Hindawi

Mathematical Problems in Engineering

Volume 2021, Article ID 9974230, 18 pages
https://doi.org/10.1155/2021/9974230

Research Article
Multiresponse Optimization for a Novel Compliant Z-Stage by a
Hybridization of Response Surface Method and Whale
Optimization Algorithm

Minh Phung Dang,1 Hieu Giang Le,1 Ngoc N. Trung Le,1 Ngoc Le Chau,2
and Thanh-Phong Dao 3,4
1
Faculty of Mechanical Engineering, Ho Chi Minh City University of Technology and Education, Ho Chi Minh City, Vietnam
2
Faculty of Mechanical Engineering, Industrial University of Ho Chi Minh City, Ho Chi Minh City, Vietnam
3
Division of Computational Mechatronics, Institute for Computational Science, Ton Duc Thang University,
Ho Chi Minh City, Vietnam
4
Faculty of Electrical & Electronics Engineering, Ton Duc Thang University, Ho Chi Minh City, Vietnam

Correspondence should be addressed to Thanh-Phong Dao; daothanhphong@tdtu.edu.vn

Received 16 March 2021; Revised 5 April 2021; Accepted 11 April 2021; Published 23 April 2021

Academic Editor: Dr. Dilbag Singh

Copyright © 2021 Minh Phung Dang et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
A novel compliant z-stage is applied for positioning and indenting a specimen in nano/microindentation testing system. For an
excellent operation, the proposed z-stage can concurrently satisfy multicriteria comprising high safety factor, small parasitic
motion, and large output displacement. The key aims of this article are to present a novel design of the compliant z-stage as well as
an effective integration methodology of Taguchi method, response surface method, weight factor calculation based on signal to
noise, and the whale optimization algorithm to resolve a design optimal problem so as to enrich the quality performances of the
proposed stage. Primarily, the z-stage is designed based on four-lever amplifier, compliant hinge shifted arrangement mechanism,
zigzag-based flexure spring guiding mechanism, and symmetric six leaf hinges-based parallel guiding mechanism. Secondly, the
number experiment data are achieved by the Taguchi method and finite element analysis. Subsequently, the regression functions
among input variables and quality characteristics are formed by exploiting response surface method. In addition, the weight
factors for every characteristic are defined. Additionally, the sensitivity analysis is accomplished for determining influences of
input variables on quality responses. Ultimately, based on regression equations, the whale optimization algorithm is executed to
define the optimal factors. The consequences indicated that the output deformation is about 454.55 μm and the safety factor is
around 2.38. Furthermore, the errors among the optimal consequences and the confirmations for the safety factor and output
deformation are 7.12% and 4.25%, correspondingly. By using Wilcoxon and Friedman methods, the results revealed that the
proposed algorithm is better than the cuckoo search algorithm. Based on the quality convergence characteristics of hybrid
approach, the proposed method is proficient for resolving complicated multiobjective optimization.

1. Introduction mechanical components together. Consequently, there is a

complexity for obtaining precise positioning due to the
Nano/microindentation testing technique has been utilized demerits of backlash, friction as well as abrasion. As a result,
for supplying small displacement or force for measuring the an accurate positioner is crucial for locating position as well
depth in the nano/microratio quality characteristics of as checking material specimen.
material specimen. This technology is utilized to monitor Based on the essential merits of compliant mechanism
mechanical features of various materials [1, 2]. However, such as free wear, free backlash, light weight, small friction,
entire positioner utilized conventional joints to connect high precision, low cost, and compact mechanism [3–5], it
2 Mathematical Problems in Engineering

has been extensively employed to discover and gradually In addition, in previous studies, Canan Kandilli et al.
replace traditional mechanism. In addition, accurate posi- employed Ansys workbench to optimize the design and
tioning platforms are also inherited from quality charac- operation factors of a PVT system integrated with natural
teristic benefits of the compliant mechanism, and they are zeolite [20]. Maryam Safa et al. used the adaptive neuro-
widely utilized for potential applications such as constant- fuzzy inference system (ANFIS) approach to predict the
force precision positioning [6], ultra-violet nanoimprint highest significant factors on vectorial crystal growth of
lithography [7], atomic force microscopy [8], and posi- extremely oriented vertically aligned carbon nanotubes [21].
tioning system for cell micromanipulation [9]. In addition, Chunbiao Zhu et al. utilized a saliency algorithm to create a
related to positioning stage for nanoindentation tester, significance map for guiding the style transmitting proce-
Huang et al. proposed a novel in situ nanoindentation in- dure. Zeeshan Ali et al. utilized complex neutrosophic set
strument with the maximal output displacement of about (CNS) to instigate the novel dice similarity measures and
11.44 μm. The device can realize coarse modification and generalized dice similarity [22]. Furthermore, several pop-
accuracy loading and unloading procedure automatically ulation-based algorithms [23–25] were developed. For ex-
[10]. Moreover, the minimal indentation instrument with ample, Golpı̂ra et al. applied the genetic algorithm (GA) for
the operating stroke of 40 μm for indenter was developed by optimization key factors of the power system [26], Basu et al.
Huang et al. [11]. utilized the cuckoo search algorithm for solving economic
Meanwhile, the working travel of piezoelectric (PZT) dispatch [27], and Chander et al. proposed particle swarm
actuator is limited, so, in order to obtain enough significant optimization for optimizing multilevel thresholding [28].
output deformation and high precision for z-axis-indenter, Prakash et al. utilized the multiobjective particle swarm
the compliant positioning stages regularly combine various optimization (MO-PSO) method to define the electrical
displacement amplifications [12]. For specifics, Chang et al. discharge machining (EDM) factors [29]. In contrast, these
offered a Scott–Russell mechanism for expanding the stroke algorithms demand primary control factors. Other ap-
of PZT for gaining high deformation [13] and Wang et al. proaches to enhance convergence velocity with less primary
developed the displacement amplification mechanism of factors are, for instance, teaching learning-based optimi-
lever mechanism and bridge mechanism combined into zation (TLBO) [30–33] and Rao’s algorithms [34].
compliant gripper [14]. Tang et al. proposed a new hydraulic In order to conduct multicriteria, hybrid approaches
deformation magnification structure for PZT actuator for such as the TM-RSM and GA [35], the TM-fuzzy based on
achieving large stroke [15]. Xu proposed a novel flexure moth-flame optimization [36], PSO-GA [37, 38], RSM-PSO
gripper according to the Scott–Russell structure to magnify [39], gravitational search algorithm (GSA)-GA [40, 41],
output displacement [16], and Kim et al. exploited a double grey-TM, RSM, and entropy measurement [42], PSO-neural
amplifier comprising two L-shape levers in a compliant 3- network (NN) [43], TM-RSM and NSGA-II [44], and fuzzy
DOF stage [17]. Although there have been a lot of compliant logic-ANFIS and lightning attachment procedure optimi-
mechanisms with one degree of freedom, there has been a zation (LAPO) [45] have been implemented by many re-
little attention in developing a compliant z-stage for use in searchers. For more specifics, Dao proposed hybrid
nanoindentation devices. Furthermore, there is a need to approach of the grey-TM according to fuzzy logic for op-
develop a new z-stage with full consideration of multiple timizing the main parameters of a compliant stage [46].
functions, simultaneously. In order to fill in the gaps in the Dang et al. developed a combined method of the TM, RSM,
literatures, the present paper is aimed to design a novel enhanced ANFIS, and TLBO for a positioner [47].
compliant z-stage based on the four-lever amplification Meanwhile, an integration method of TM, RSM [48],
mechanism, compliant hinge shifted arrangement mecha- and whale optimization algorithm (WOA) integrated the
nism, zigzag-based flexure spring guiding mechanism, and weight factor calculation method based on the signal to noise
symmetric six leaf hinges-based parallel guiding mechanism. for optimizing a compliant z-stage applied for the indenter
In order to reach multiple good functions for the z-stage, in nanoindentation tester which has not been studied yet. As
an integrated method of the Taguchi method (TM), response a result, based on the global convergence quality charac-
surface method (RSM), and whale optimization algorithm teristics of the WOA, a proficient hybrid approach of TM,
[18] is developed to optimize main geometric parameters of RSM, and WOA [18, 49–52] is developed to resolve mul-
the proposed z-stage. Commonly, the output deformation ticriteria optimization trouble of the z-stage in this article.
and the safety factor are the greatest essential features. Additionally, the weight factors based on signal to noise [53]
However, two characteristics are struggled together. As a for each response are calculated and assigned for the total
result, in order to balance and achieve the desires of a large weighted objective function in order to enhance the optimal
output deformation as well as a high safety factor, simul- precision results.
taneously, the hybrid method is proposed for conducting the This article proposes a new optimization design for the
optimal problem. Additionally, the key geometric factors of compliant z-stage. In order to solve the optimization, a
z-stage were optimized for achieving aforementioned hybridization of the TM, RSM, weight factor calculation, and
requests. whale optimization algorithm is developed. The TM-RMS is
Generally, the TM is widespread exploited in optimal integrated with the finite element method to retrieve datasets
analysis engineering for generating initial numerical data and establish the regression equations. Based on the re-
[19]. By contrast, in order to perform multiresponse opti- gression models, the whale optimization is extended to
mization, the TM should integrate with various algorithms. optimize the geometry factors of the z-stage. Then, the
Mathematical Problems in Engineering 3

numerical validations are performed for confirming the thin thickness, the rotation characteristics of different
optimal consequences. Furthermore, the sensitivity analysis hinges, the travel loss as well as distance among continued
and analysis of variance are studied for all parameters. Fi- levers, so, equation (2) is not precise for amplification ratio
nally, a few statistical analysis [54] is executed for evaluating calculation. It should be checked by finite element analysis
the manners of the proposed algorithm with another (FEA) to obtain the accurate consequences. Likewise, so as
approach. to increase high stiffness as well as decline decoupling
error, the zigzag-based flexure springs and parallel guiding
2. Conceptual Design of Compliant Z-Stage mechanism were integrated into the magnification
mechanism of the z-stage.
In this study, an intended prospective application of the
compliant z-stage is suggested, as illustrated in Figure 1. This
system comprises a coarse z-stage, a coarse xy-stage, a fine 2.2. Compliant Z-Stage. In this research, based on the great
xy-stage, and a fine z-stage. For specifics, the coarse xy-stage characteristics of the material Al 7075 (good yield strength
is applied for the initial coarse positioning and a fine xy-stage of 503 MPa, Young’s modulus of 71700 MPa, light density
is used for the fine positioning the material specimen as well of 2770 kg/m3, and Poisson’s ratio of 0.33), this material
as the coarse z-stage is employed for moving the sample to was chosen for designing the proposed z-stage. In addi-
near the indenter. Finally, the fine z-stage is named as the tion, in order to achieve more than the working travel of
compliant z-stage which is utilized for the fine positioning 40 μm, the Physik-device PZT actuator (P-225.10) with a
the indenter to reach to the location of material sample. maximal stroke variety [0, 15] μm or translational crew for
creating more higher displacement will be utilized for
2.1. Integrated Amplification Mechanism of Displacement. generating input displacement and connected with pro-
A lever structure is beneficial structure for amplifying the posed hybrid amplification mechanism. For specifics,
values of force and displacement. First and foremost, Figures 4(a) and 4(b) illustrate the details and main
Figure 2 depicts an operating principle of one lever geometric dimensions of the proposed compliant z-stage,
mechanism for magnifying the displacement along the z- respectively. The z-stage comprises following parts: (i)
axis. For specifics, point O represents a fixed joint that is fifteen fixed holes are used for locating the z-positioner
the rotary position of the lever. In addition, M and N are and (ii) a PZT actuator or translational screw can be
the input and output points, respectively. The working utilized for generating the input displacement for the
principle of lever mechanism is demonstrated as follows: z-stage. Additionally, force transferring guiding mecha-
when a vertical displacement Δd1 impacts into the input nism based on zigzag flexure springs and parallel guiding
point M, the lever will rotate a relative angle c around the mechanism are integrated into the z-stage for reducing
z-axis. Consequently, the point N moves to N′ and the decoupling error.
output displacement Δd2 will obtain in vertical y-axis The sum dimension of z-stage is around
direction. In contrast, this mechanism generates easily 301 mm × 162 mm × 16 mm. Especially, compared with
high decoupling error. Therefore, displacement amplifi- common flexure hinges, elliptical hinge is proposed for
cation mechanism with symmetric multilever is integrated utilizing so as to gain accurate rotary-center and high output
to amplify the output displacement as well as reduce the displacement, simultaneously. Besides, in order to reduce
parasitic motion. Later on, in order to decline more the decoupling error as well as gain mobility, parallel guiding
decoupling error, a guiding mechanism with zigzag-based mechanism is also used to integrate into z-stage. In addition,
flexure spring and parallel guiding mechanism are inte- by checking initial quality responses, in order to avoid plastic
grated into the magnification mechanism, as exhibited in failures as well as obtain a high safety factor, an input
Figure 3. In order to decline more the error of decoupling displacement of 65 μm is proposed.
error, parallel guiding mechanism-based symmetric six The proposed z-stage would be advanced for creating
leaf hinges was integrated into a hybrid amplification linear deformation employed for in situ nanoindentation
mechanism. Especially, this mechanism affects signifi- tester. Table 1 provides geometric dimensions of the z-stage.
cantly in reducing parasitic motion for precise z-stage. Thus, some key factors of the stage need to be considered in
According to the lever mechanism physical features, the optimizing procedure in order to boost the quality features
magnification proportion can be around gained: of the z-stage.
Δd2 d2
rlever � � . (1)
Δd1 d1 3. Methodology
The function of multiple levers magnifier is supposed in 3.1. Formulation of Optimal Problem. The compliant z-stage
form as follows: ought to satisfy the subsequent requisites: (1) the safety
d2 factor (F1) is expected to obtain high value in order to assure
rlever � k × . (2) the strength of the z-stage and (2) high z-axis displacement
d1
(F2) is also demanded for gaining high value for expanding
For a four-lever magnification, k is supposed for locating competence to position as well as indent for
selecting 4. Due to the influences of flexure hinges such as checking a material specimen.
4 Mathematical Problems in Engineering

Fine Z
Nanoindentation positioning
stage Multiple
tester microscope

Z
Column
Y
Material sample
O X
Fine XY positioning stage

Coarse XY positioning stage

Coarse Z
positioning stage

Base

Figure 1: A proposed nano/microindentation system with the compliant z-stage.

Y Y
X d2 d1
O

Δd1
Δd2

N O M′ X
M
N′
Output displacement Input displacement
Fixed
Input
(a) (b)

Figure 2: Scheme: (a) lever-structure operating; (b) magnification proportion.

Especially, based on the global convergence quality F1 (M, N, P, K). (3)

characteristics of the WOA, an integration approach of the
TM, the RSM, weight factor calculation method based on Maximize F2 (X) as follows:
signal to noise, and the whale optimization algorithm
(WOA) is proposed for balancing among its characteristics F2 (M, N, P, K), (4)
as well as enhancing the quality responses of the z-stage. A
WOA algorithm would be exploited via MATLAB 2017. This and subject to the following constraints:
integration method diagram is suggested, as shown in
σy
Figure 5. σ max ≤ , (5)
The optimal trouble for the z-stage is shortened as n
follows.
Seek the input variables: X � [M, N, P, K]. 0.55mm ≤ M ≤ 0.65 mm; 0.6 mm ≤ N ≤ 0.9 mm,
􏼨 (6)
Maximize F1 (X) as follows: 0.9 mm ≤ P ≤ 1.1 mm; 49 mm ≤ K ≤ 53 mm,
Mathematical Problems in Engineering 5

Input
displacement Lever 1
(PZT)
Zigzag-based
flexure hinge

Elliptical hinges

Lever 2
Parallel guiding mechanism Output displacement
based on symmetric six
leaf hinges
Figure 3: Proposed hybrid displacement amplifier.

N k
j
h
K l n
P o
f g q
a i m

e M r
t

s u c

Figure 4: Main geometric factors of the z-stage.

Table 1: Geometric factors of the compliant z-stage.

Factor Worth Factor Worth Factor Worth Unit
A 114.5 J 6.5 i 15 mm
B 301 K 3.5 t 15 mm
C 60 M 32.5 u 73 mm
D 20 N 16 M 0.55 ≤ M ≤ 0.65 mm
E 17 O 1 N 0.6 ≤ N ≤ 0.9 mm
F 21 Q 8 P 0.9 ≤ P ≤ 1.2 mm
G 5 R 8 K 49 ≤ K ≤ 53 mm
H 22 S 35

where F1 and F2 denote the safety factor and the displace- (lever-amplifier 2), and distance between second elliptical
ment, respectively. There are many factors affecting to the hinge center and guiding elliptical hinge center,
output responses. Meanwhile, M, N, P, and K are four main correspondingly.
factors affecting significantly the output characteristics. Additionally, σ y symbolizes the yield strength of the
Therefore, M, N, P, and K were chosen as input design material (Al 7075), σ max symbolizes the maximal stress of the
variables. For more specifics, M, N, P, and K are thickness of compliant stage as well as n is the safety factor.
zigzag-based compliant spring, thickness of first elliptical A safety factor of more than 1.8 was recommended to
joint (lever-amplifier 1), thickness of second elliptical hinge design z-stage for avoiding plastic breakdowns.
6 Mathematical Problems in Engineering

CAED
Computer-aided engineering design Whale optimization algorithm
Start optimize parameters of z-satge

Define optimization problem Determine the parameters of the algorithm

Determine design variables and

Create initial candidate solution
objective functions

Create 3D model by FEM Generate initial optimal search agent

No Update the position of the search agent

Comparision and selection according to the values of A and p
of initialization designs

Yes Update the position and score

of the optimal search agent
Build design of experiments and collect
Refine range of design variables

data by integrating of TM and FEM

and objective functions

Establish regression equations Interaction max?

using RSM

Statistic based
on weight Yes
factor calculation
Calculate S/N ratio for
objective functions
No Are solution satisfied
by validating?
Normalize the S/N ratio
Yes
Calculate weight factors End
by established equations

Figure 5: Flowchart of the proposed optimization approach.

Equation (6) illustrated limitation range of the input vari- where βi (i � 0, 1, 2, . . ., n) denote unknown regression
ables. The smaller and the higher bounds for the input variables coefficients, βij (i < j) denotes collaboration coefficients, x1,
were chosen for ensuring a compact compliant structure. x2, . . ., xn are series of n predictors considered to be asso-
ciated to a response variable Fj, and ε denotes an arbitrary
error.
3.2. Hybrid Approach. In this research, the proposed method Additionally, analysis of variance (ANOVA) would be
was developed to optimize the key geometric factors for implemented for defining meaningful influence proportion
enhancing the output responses of the stage. of every factors on the quality characteristics.
Primarily, the TM is exploited for building initial nu- Thirdly, weight factors were determined for each re-
meric experiments. Secondly, the RSM is employed to form sponse according to the signal to noise for enhancing op-
regression functions to map connections among input timal calculation results. Based on the weight factor method
variables and the both quality characteristics. A complete with details in Reference [53], the WFs for each response
quadratic formulation is appropriate type for the proposed were computed accurately. Every objective is normalized in
z-stage as following equation: the variety from 0 to 1. Later on, WFs were assigned for both
n n n−1 n output responses.
Fj � β0 + 􏽘 βi xi + 􏽘 βij x2i + 􏽘 􏽘 βij xi xj + εi , (7) Based on the calculated weight factors, the total weighted
i�1 i�1 i�1 j�i+1 objective function was conveyed as follows:
Mathematical Problems in Engineering 7

f � w1 f1 + w2 f2 , (8) and m � 1, 2, . . . , l, where l denotes the digit of

experiment grade of every input variable.
where f1 , f2 , and f are safety factor, output displacement, (5) Identify the Weight Factors (WFs). The WFs are in
and integrated function, correspondingly and w1 and w2 form as follows:
denote the weight factor worths of f1 and f2 , corre- q
spondingly. In addition, the weight factor is determined as 􏽐j�1 rij
wi � m q , (13)
follows: 􏽐i�1OF 􏽐j�1 rij
(1) Identify Signal to Noise Scale (S/N). The S/N scale is
formed by the Taguchi method, and it demonstrates where wi (wi ≥ 0) symbolizes the WF of every
a better ability corresponding to a higher signal to performance function ith and mOF symbolizes the
noise scale worth. digit of performance functions.
Based on the technical requirements of responses, The equation for checking the calculation the sum
the higher the better is exploited, that is, in form as of weight factors is given by
follows: mOF

n
􏽘 wi � 1. (14)
⎝1 􏽘 1 ⎞
η � −10 log⎛ ⎠, (9)
i�1
k i�1 y2i
According to the regression functions assigned
calculated weight factors, the whale optimization
where k and y symbolize the number of reiterations
algorithm was employed for seeking the optimi-
of ith testing and the property.
zation factors. A WOA was proposed for multiple
(2) Normalize the S/N Scale. The normalization-based responses optimization because of the quality
signal/noise scale of every rank for every perfor- convergence characteristics and high convergence
mance is expressed as follows: speed.
ηj − min ηj
zi � , (10)
max ηj − min ηj 3.3. Whale Optimization Algorithm. Finally, the WOA im-
itated from hunting manners of the humpback whales in
where ηi symbolizes the S/N worth, max ηi and seeking and tackling preys considering bubble-net feeding
min ηi denote the greatest and lowest worths of ηi , manner was suggested by Mirjalili and Lewis [18]. The WOA
respectively, zi symbolizes the normalization-based comprises two key phases as follows: (1) exploitation phase:
S/N worth of characteristic function (j � 1, 2, . . ., k), surrounding quarry and helix apprising location can be
and k is the digit of characteristic functions. executed and (2) Exploration phase: investigating for a prey
(3) Define the Average Worth of Normalized S/N Scales. is conducted erratically. The more details can be read in
The average worth of normalized S/N ratios at every Reference [18]. The mathematical equation of every phase is
grade for design variables was determined for expressed in the subsequent parts.
characteristic functions. The S/N scale worths are
normalized, and the average worth of normalized S/ 3.3.1. Exploitation Phase (Bubble-Net Attacking Method).
N scales is in form as follows: Both progresses can be established so as to establish
1 m mathematical function for the bubble-net manner of
aLi � 􏽘z , (11) humpback whales. The both progresses are expressed as
NLi i ij
follows.
where NLji symbolizes digit of reiteration of grade
ith, zij symbolizes the worth of S/N scale of grade (1) Surrounding Quarry. After identifying the location of the
ith of performance function jth, and Ali is the mean quarry, humpback whales surround the quarry. Because the
value of S/N ratio of grade ith of every input var- situation of the optimization design in the investigation area
iable of every performance function. can be not defined, the WOA algorithm supposes that the
recent finest applicant resolution is the objective quarry or is
(4) Determine the Variety of Every Design Variable. The proximate the optimum. Later on, the greatest investigation
variety of the every grade of every design variable is agent is determined. Therefore, the other investigation
expressed as follows: agents can endeavor for adapting their situations nearby the
rij � max􏽮zi,j,1 , zi,j,2 , . . . , zi,j,m 􏽯 finest investigation agent. Mathematical equations of the
(12) manner are in form as follows:
− min􏽮zi,j,1 , zi,j,2 , . . . , zi,j,m 􏽯, → 􏼌􏼌􏼌􏼌→ �→∗ → 􏼌􏼌􏼌
D � 􏼌􏼌 C .X (t) − X (t)􏼌􏼌􏼌, (15)
where rij denotes the variety (max-min) of the S/N
value for every grade of every parameter; → �→ →→
X (t + 1) � X∗ (t) − A .D , (16)
j � 1, 2, . . . , q, whereq is the digit of input variables;
8 Mathematical Problems in Engineering

where t denotes the present repetition, X∗ symbolizes the chosen investigation agent in preference to the greatest
→
location vector of the greatest gained resolution, X sym- investigation agent realized. This structure and | A | > 1
bolizes the location vector, and | | is the absolute value. X∗ highlight exploration as well as permit the WOA algo-
ought to be apprised in every iteration if it has a greater rithm to implement a universal investigation. The alge-
resolution. A and C are coefficient vectors computed and are braic equation can be computed by
in form as follows: → 􏼌􏼌􏼌􏼌→ ⇀ →
→ D � 􏼌􏼌 C · Xrand − X |, (21)
→ → → (17)
A � 2a · r − a,
→ ⇀ → →
→ → X (t + 1) � Xrand − A · D , (22)
C � 2. r , (18)
⇀
→ where Xrand symbolizes an arbitrary location vector (an
where a is declined in range (2, 0) throughout repetitions arbitrary whale) preferred according to present population.
→
and r symbolizes an arbitrary vector in range (0, 1).
→
To employ shrinking, the worth of a is decreased → in
equation (17). In addition,
→ the oscillation variety of A can be 3.6. Process of Whale Optimization Algorithm. The WOA
→
reduced via a . Besides, A symbolizes an arbitrary worth in algorithm starts via setting arbitrary resolutions. Until now,
the interval from–a to a where a is declined in (2, 0) over the
→ investigation agents apprise their places with respect to
course of repetitions. Establishing arbitrary worths for A either an arbitrarily selected investigation agent or the
from −1 to 1, the updated location of an investigation agent greatest resolution gained for every repetition. The factor is
is specified anywhere from the initial point of the agent to fallen in range (2, 0) so as to offer exploration and ex-
the location of the recent greatest agent. ploitation, correspondingly. An arbitrary investigation agent
→
is preferred when→ | A | > 1, while the greatest resolution is
3.4. Helix Updating Location. After computing the dimen- chosen when | A | < 1 for apprising the location of the in-
sion among the whale and quarry corresponding with po- vestigation agents. Therefore, if its value is more than 0.5,
sitions at (X, Y) and (X∗ , Y∗ ), Later on, a spiral model for the then the investigation agents vary their points using equa-
place of whale and prey to imitate the spiral shape-based tion (19), else utilizing equation (16). WOA is able to ex-
motion of humpback whales is computed as the following change among either a helix or rounded travel. Ultimately,
equation: the WOA algorithm is ended by the fulfilment of an ending
→ �→ �→ condition.
X (t + 1) � D′ · ebm · cos(2πl) + X∗ (t), (19) Based on theoretical view, the WOA can be concerned
�→ �→∗ → about a universal optimization algorithm according to its
where D′ � |X (t) − X (t)| and shows the distance of the ith merits of exploration/exploitation ability. Moreover, the
whale to the quarry, b symbolizes a stable worth for de- proposed hypercube structure determines an investigation
termining the form of the logarithmic helix, and m sym- region in the area of the best resolution and permits other
bolizes an arbitrary digit form −1 to 1. This manner in WOA investigation agents to utilize the current greatest evidence
is illustrated for changing locations of whales in optimal in the domain. Apprised alteration of the search vector A
process. To formulate simultaneously this manner, it has a permits the WOA algorithm to smooth journey among
50% possibility for preferring amongst the shrinking exploration as well as exploitation: by declining A, several
encircling procedure and the helix pattern. The mathe- repetitions can be dedicated to exploration (|A| ≥ 1), and the
matical model for updated location of whales is calculated as rest is allocated to exploitation (|A| < 1). Extraordinarily, the
following equation: proposed algorithm comprises purely both key internal
�→ →→ factors modified (A and C). The more details can be read in
→ ⎪
⎧
⎨ X∗ (t) − A .D ⎪
⎫
if p < 0.5 ⎬
X (t + 1) � ⎪ �→ , Reference [18].
⎩ bl
�→∗
⎪
⎭
′
D · e · cos(2πl) + X (t) if p < 0.5
(20) 4. Results and Discussion
where p denotes the contingent value from 0 to 1.
4.1. Assessment of Primary Characteristics and Decoupling
Error. Firstly, a compliant z-stage was constructed via In-
3.5. Exploration Phase (Investigation for Quarry). →
The sim- ventor software. Subsequently, the z-axis output displace-
ilar method according to the difference of the A vector ment as well as safety factor of the proposed z-stage was
can be exploited for investigating for quarry (exploration). retrieved based on FEA in Ansys software. For specifics, a
Actually, humpback whales are investigated arbitrarily → coarse meshing was employed in the automatic mode.
according to the place of every another. Consequently, A Thereafter, compliant joints were purified for achieving
can be utilized as the arbitrary worths better than 1 or less accurate analysis results, as demonstrated in Figure 6.
than −1 to control investigation agent to journey remotely In order to check the safety factor, the decoupling error,
with a reference whale. Until now, compared with the and magnification proportion, as illustrated in Table 2, the
exploitation phase, the place of an investigation agent in output deformation of z-stage was examined via altering the
the exploration phase is updated based on an arbitrarily input displacement from 5 μm to 60 μm, correspondingly.
Mathematical Problems in Engineering 9

Refine mesh

Coarse mesh

Figure 6: Mesh establishment of the proposed z-stage.

Table 2: The output deformation, magnification, safety factor, and stress.

Equivalent stress
Input displacement (μm) Output deformation (μm) Magnification scale Safety factor
(MPa)
65 387 5.95 1.8092 278.02
70 416.77 5.95 1.68 299.41
75 446.54 5.95 1.568 320.8
80 476.31 5.95 1.47 342.18
85 506.07 5.95 1.3835 363.57
90 535.84 5.95 1.3066 384.96
95 565.61 5.95 1.2379 406.34
100 595.38 5.95 1.176 427.73
105 625.15 5.95 1.12 449.12
110 654.92 5.95 1.0691 470.5
115 684.69 5.95 1.0226 491.89
120 714.46 5.95 0.97998 513.28

In order to check strength conditions of the proposed When the z-stage travelled along the z-direction, simi-
z-stage, the input movement was changed in range (65, 120) larly, the motion of x-direction also occurred. The unex-
with the changing step of 5 μm. In addition, the corresponding pected travel created the decoupling error. Based on the
output deformation could be retrieved by FEA. Based on the results of Table 3, the decoupling error is really minor with
consequences of Table 2, corresponding to input displacement 0.0389% (<1%), so, this error did not influence meaningfully
of 120 μm, the output deformation was 714.46 μm, the safety to the output deformation accuracy. The decoupling error
factor was lower than 1 as well as the stage could be damaged. could be in form as follows:
Consequently, the maximal input displacement ought to be x
lower than 120 μm. By contrast, when the input displacement ed � o 100%, (23)
zo
was 115 μm, the equivalent stress was around 491.89 MPa
(lower than the yield strength of material) corresponding to the where ed symbolizes the decoupling error and xo and zo are
safety factor of 1.0226. Especially, in order to ensure a good the x-direction output deformation and the z-direction
working life for the stage, a safety factor over 1.8 was expected. output deformation.
In addition, the input displacement was proposed range in (5,
65) for checking output deformation as well as the decoupling
error which was computed via exploiting equation (21). The 4.2. Orthogonal Array Experiment and Mathematical Model.
consequences showed that the decoupling error was moder- Every parameter could be partitioned to triple grades
ately minor with 0.0389% (<1%), as shown in Table 3. according to specialized experience as well as machining
Therefore, it guaranteed the accuracy for the proposed z-stage. capacity, as revealed in Table 4. The L9 (34) orthogonal array
Moreover, the consequences exhibited the safety factor of of the Taguchi method was implemented for forming the
1.8092 and magnification scale (MS) of 5.95. Conversely, all initial quantity of experimentations. The safety factor (F1)
consequences could not meet magnification demands (MS is and z probed deformation (F2) were gathered by the finite
more than 6). Consequently, an optimal procedure should be element method in Ansys software and consequences of the
implemented for enhancing the quality characteristics. numeric testing, as exhibited in Table 5.
10 Mathematical Problems in Engineering

Table 3: Consequences of output deformation as well as decoupling error.

Input displacement (μm) Z-axis output deformation (μm) X-axis output deformation (μm) Decoupling error (%) Magnification scale
5 29.77 0.0116 0.0389 5.95
10 59.54 0.0232 0.0389 5.95
15 89.31 0.0348 0.0389 5.95
20 119.08 0.0464 0.0389 5.95
25 148.85 0.0579 0.0389 5.95
30 178.61 0.0695 0.0389 5.95
35 208.38 0.0811 0.0389 5.95
40 238.15 0.0927 0.0389 5.95
45 267.92 0.1043 0.0389 5.95
50 297.69 0.1159 0.0389 5.95
55 327.46 0.1275 0.0389 5.95
60 357.23 0.1391 0.0389 5.95

Table 4: Input dimensions and their grades (unit: mm).

Parameters Variety Grade 1 Grade 2 Grade 3
M 0.55–0.65 0.55 0.6 0.65
N 0.6–0.9 0.6 0.75 0.9
P 0.9–1.2 0.9 1.05 1.2
K 49–53 49 51 53

Table 5: Numerical results.

No. M N P K F1 F2 (μm)
1 0.55 0.6 0.9 49 2.1846 423.89
2 0.55 0.75 1.05 51 1.8968 392.88
3 0.55 0.9 1.2 53 1.8956 253.77
4 0.6 0.6 1.05 53 1.615 221.14
5 0.6 0.75 1.2 49 2.1052 401.83
6 0.6 0.9 0.9 51 1.944 370.74
7 0.65 0.6 1.2 51 1.6641 309.1
8 0.65 0.75 0.9 53 1.7837 211.66
9 0.65 0.9 1.05 49 2.2807 381.09

The regression equations were gained as follows:

F1 � 109.2 − 31.00∗ M + 0.5896∗ N − 0.1536∗ P − 3.740∗ K + 25.14∗ M∗ M
+ 0.09333∗ N∗ N − 0.05778∗ P∗ P + 0.03562∗ K∗ K,
(24)
F2 � −25134 + 640.1∗ M + 646.2∗ N + 252.9∗ P + 1028∗ K − 1002∗ M∗ M
− 392.7∗ N∗ N − 142.4∗ P∗ P − 10.50∗ K∗ K.

Tables 6 and 7 show the ANOVA consequences of the relation K and K was substantial higher, at 8.73% and
safety factor as well as output deformation. The assessment 6.50%, respectively. In contrast, the rate of N, P, relation N
could be implemented with 5% substance grade and 95% and N, relation P and P as well as relation M and M was
assurance grade. slightly lower, at 0.81%, 0.53%, 0.29%, 0.04%, and 0.02%,
As depicted in Table 6, the impact rate on the safety respectively. Hence, so as to climb the worth of F1, pa-
factor F1 of K was highest, at 84.34%, while the effect figure rameters K and M ought to be considerably monitored.
for N and interaction K and K was substantial higher, at Furthermore, the influence rate of error for F1 as well as F2
17.43% and 9.85%, respectively. In contrast, the rate of M, P was 0%.
as well as interaction M and M was slightly lower, at 2.50%,
2.47%, and 1.92%, respectively. Consequently, so as to climb
the worth of F1, factors K as well as N ought to be con- 4.3. Sensitivity Analysis. Statistic engineering was imple-
siderably monitored. mented to define the influence level of input variables on the
As illustrated in Table 7, the effect rate of K on F2 was output performances. In the variety (0.55, 0.65), factor M
greatest, at 83.07%, while the effect figure for M and influenced to F1 and F2 in reducing slightly, as depicted in
Mathematical Problems in Engineering 11

Table 6: ANOVA analysis for safety factor (F1).

Source DF Seq SS Contribution (%) Adj SS Adj MS P value
Model 8 0.412320 100.00 0.412320 0.051540 Substantial
Linear 4 0.363795 88.23 0.363795 0.090949 Substantial
M 1 0.010292 2.50 0.010292 0.010292 Substantial
N 1 0.071854 17.43 0.071854 0.071854 Substantial
P 1 0.010201 2.47 0.010201 0.010201 Substantial
K 1 0.271448 65.83 0.271448 0.271448 Substantial
Square 4 0.048525 11.77 0.048525 0.012131 Substantial
M∗M 1 0.007900 1.92 0.007900 0.007900 Substantial
N∗N 1 0.000009 0.00 0.000009 0.000009 Substantial
P∗P 1 0.000003 0.00 0.000003 0.000003 Substantial
K∗K 1 0.040613 9.85 0.040613 0.040613 Substantial
Error 0 — — — — —
Total 8 0.412320 100.00 — — —

Table 7: ANOVA analysis for output displacement (F2).

Source DF Seq SS Contribution (%) Adj SS Adj MS P value
Model 8 54299.7 100.00 54299.7 6787.5 Substantial
Linear 4 50580.8 93.15 50580.8 12645.2 Substantial
M 1 4742.7 8.73 4742.7 4742.7 Substantial
N 1 441.5 0.81 441.5 441.5 Substantial
P 1 288.3 0.53 288.3 288.3 Substantial
K 1 45108.3 83.07 45108.3 45108.3 Substantial
Square 4 3718.9 6.85 3718.9 929.7 Substantial
M∗ M 1 12.6 0.02 12.6 12.6 Substantial
N∗ N 1 156.1 0.29 156.1 156.1 Substantial
P∗ P 1 20.5 0.04 20.5 20.5 Substantial
K∗ K 1 3529.7 6.50 3529.7 3529.7 Substantial
Error 0 — — — — —
Total 8 54299.7 100.00 — — —

Surface plot of F1 vs. M, N Surface plot of F2 vs. M, N

2.2 400
F2 (µm)

2.0
300
F1

1.8 0.9 0.9

0.8 0.8
1.6 200
)
)

m
m

0.7 0.7
(m
(m

0.55 0.55
0.60 0.6 0.60 0.6
N
N

0.65 0.65
M (mm) M (mm)

(a) (b)

Figure 7: Influence scheme of M and N on (a) safety factor and (b) the output deformation.

Figure 7. By contrast, element N revealed that in the variety To sum up, total influences of input variables on output
in (0.65, 0.75), it influenced a gradual increase to F1 as well as responses were demonstrated, as in Figure 9. These depict a
affected a slow fluctuation to F2. rise and decline variety in every parameter. Based on these
In the variety (0.9, 1.2), factor P influenced a slow decline results, the main factors could be adjusted for obtaining a
to F1 and F2, as depicted in Figure 8. In addition, in the greatest construction for the z-stage.
variety (50, 51.5), factor K influenced a sharp decline to F1
and affected a gradual decline to F2. Meanwhile, from
51.5 mm to 53 mm, factor K influenced a slight decline to F1 4.4. Calculation of Weight Factor. In general, total of both
as well as caused a dramatic decline to F2. weight factors were equivalent to one. In addition, this
12 Mathematical Problems in Engineering

Surface plot of F1 vs. P, K Surface plot of F2 vs. P, K

2.2 400

F2 (µm)
2.0
F1

300
1.8 53.0 53.0
1.6 51.5 51.5
200

)
m

m
50.0 50.0

(m
0.9 0.9

(m
1.0 1.0

K
1.1 1.1

K
1.2 1.2
P (mm) P (mm)

(a) (b)

Figure 8: Influence scheme of P and K on (a) safety factor and (b) the output deformation.

0.60 0.75 0.90 50.0 51.5 53.0

2.2

2.0
F1

1.8

1.6

400

350
F2 (µm)

300

250

200
0.55 0.60 0.65 0.90 1.05 1.20
M N P K
Figure 9: Sensibility scheme of input variables on two quality characteristics.

weight factor for every response was allocated 0.5 but this 4.5. Optimal Results and Verifications. Initial numeric ex-
worth could be not exact as well as found optimal solutions perimentations were generated via exploiting the Taguchi
could be imprecise. Consequently, this paper proposed to method. Subsequently, the z-stage design was constructed and
exploit an efficient method to define the weight factors. the numeric calculation results were retrieved. According to
Afterward, the numerical data were transferred into the collected information, the response surface method was
signal to noise proportions applying equation (9), as employed for forming the regression models for the safety
revealed in Table 8. Subsequently, the proportions were factor as well as output deformation. Afterward, the WFs were
normalized via equation (10), as demonstrated in Table 9. determined using equations (9)–(14). According to the weight
For specifics, the normalized signal to noise proportions for factors assigned in equation (8), the multiobjective optimiza-
safety factor (η1) and the displacement (η2) symbolized z1 tion trouble was resolved via the WOA method. The optimal
and z2, correspondingly. Employing equations (12) and (13), procedure was executed by utilizing MATLAB 2017 software.
the WF for the safety factor and the displacement was The optimization consequences were detected at M � 0.55 mm,
computed, as indicated in Tables 10 and 11, correspondingly. N � 0.82377 mm, P � 0.9 mm, K � 49 mm, F1 � 2.382456817 as
The weight factor worths for F1 and F2 were 0.5150 and well as F2 � 454.5511266 μm. The consequences exposed that
0.485, correspondingly. Later on, the calculated WFs for the optimal values of both response were appropriate for the
each response were allocated for the optimal procedure in mirco/nanoindentation device and precise positioning system
the WOA. as well as assured strength conditions of proposed material.
Mathematical Problems in Engineering 13

Table 8: The investigational consequences and S/N proportions.

No. F1 F2 (mm) η1 of F1 (dB) η2 of F2 (dB)
1 2.1846 423.89 6.787438586 52.54506342
2 1.8968 392.88 5.560430819 51.88519842
3 1.8956 253.77 5.554934001 48.08880559
4 1.615 221.14 4.163450533 46.89334611
5 2.1052 401.83 6.465867227 52.08084715
6 1.944 370.74 5.773925212 51.38138891
7 1.6641 309.1 4.423588412 49.8019801
8 1.7837 211.66 5.026436246 46.51277584
9 2.2807 381.09 7.161363251 51.62055106

Table 9: The worths of normalized S/N proportions (zi).

S/N proportions Normalized S/N proportions (zi)
η1 (dB) η2 (dB) z1 of η1 z2 of η2
6.787438586 52.54506342 0.8753 1.0000
5.560430819 51.88519842 0.4660 0.8906
5.554934001 48.08880559 0.4642 0.2613
4.163450533 46.89334611 0.0000 0.0631
6.465867227 52.08084715 0.7680 0.9230
5.773925212 51.38138891 0.5372 0.8071
4.423588412 49.8019801 0.0868 0.5453
5.026436246 46.51277584 0.2879 0.0000
7.161363251 51.62055106 1.0000 0.8467

Table 10: The WF for F1. 4.5.2. Statistical Analysis. The proficiency of the hybrid
The average worth of normalized S/N scales at
approach was assessed with the cuckoo search algorithm
every grade [55]. In this article, for a behavior assessment between
Grade
hybrid approach and CSA, a statistic analysis was exploited
M N P K
by Wilcoxon’s rank signed assessment at 5% substantial
Grade 1 0.6018 0.3207 0.5668 0.8811
grade as well as 95% guarantee intervals.
Grade 2 0.4351 0.5073 0.4887 0.3633
Grade 3 0.4582 0.6671 0.4396 0.2507
Every algorithm was retrieved to achieve optimal results
Variety rij 0.1667 0.3464 0.1271 0.6304 for 35 times with the same constraint conditions. Tables 12
WF for F1: w1 = 0.5150 and 13 present the assessment consequences of the Wil-
coxon method. It is clear that no significant difference be-
tween average worths of the two algorithms is presumed via
a null hypothesis. Based on the p-worth consequences (less
Table 11: The WF for F2. than 0.05), there is a convincing confirmation in opposition
The average worth of normalized S/N to the null hypothesis, as provided in Tables 12 and 13.
Grade proportions at every grade Therefore, the behaviors of the proposed algorithm are
M N P K different to its CSA.
Grade 1 0.7173 0.5361 0.6024 0.9233
In addition, the Friedman nonparametrics test method
Grade 2 0.5977 0.6046 0.6001 0.7477 was utilized for determining the contrast among the pro-
Grade 3 0.4640 0.6384 0.5765 0.1081 posed hybrid approach and the CSA at meaningful grade of
Variety rij 0.2533 0.1022 0.0258 0.8151 α � 0.05. Similarly, every algorithm was retrieved to achieve
WF for F2: w2 = 0.485 optimal results for 35 times with the same constraint con-
ditions. The consequences exhibited that the p-worth is
smaller than 0.05. Thus, the null hypothesis would be re-
4.5.1. Dynamic Analysis. These optimal parameters of pudiated. The noteworthy observation was that there is a
z-stage built 3D model again in order to check dynamic convincing confirmation in opposition to the null hy-
characteristics for six modes of natural resonant frequency. pothesis, as provided in Tables 14 and 15.
The modal analysis is conducted to define the six modes of
resonant frequency for the optimal z-stage, as illustrated in
Figure 10. The resonant frequency values of six modes from 4.5.3. Verification. The optimization values (M � 0.55 mm,
mode 1 to mode 6 were 336.11 Hz, 589.53 Hz, 594.75 Hz, N � 0.82377 mm, P � 0.9 mm, and K � 49 mm) would be
596.36 Hz, 601.63 Hz, and 605.49 Hz, respectively. In order employed for designing again model of z-stage for simu-
to avoid the destroy of the optimal z-stage, we should evade lation confirmation to evaluate the error among the FEA and
aforementioned resonant frequency values. the combination method. The procedure would be
14 Mathematical Problems in Engineering

B: model B: model
Total deformation Total deformation 2
Type: total deformation Type: total deformation
Frequency: 336.11Hz Frequency: 589.53Hz
Unit: mm Unit: mm
2/16/2021 4:15 PM 2/16/2021 4:15 PM
124.4 max 1031.5 max
110.58 916.9
96.757 802.28
82.935 687.67
69.112 573.06
55.29 458.45
41.467 343.84
27.645 229.22
13.822 114.61
0 min 0 min

(a) (b)
B: model B: model
Total deformation 3 Total deformation 4
Type: total deformation Type: total deformation
Frequency: 594.75Hz Frequency: 596.36Hz
Unit: mm Unit: mm
2/16/2021 4:16 PM 2/16/2021 4:16 PM
1014.8 max 918.78 max
902.02 816.7
789.26 714.61
676.51 612.52
563.76 510.43
451.01 408.35
338.26 306.26
225.5 204.17
112.75 102.09
0 min 0 min

(c) (d)
B: model B: model
Total deformation 5 Total deformation 6
Type: total deformation Type: total deformation
Frequency: 601.63 Hz Frequency: 605.49Hz
Unit: mm Unit: mm
2/16/2021 4:17 PM 2/16/2021 4:17 PM
932.84 max 757.41 max
829.19 673.26
725.54 589.1
621.89 504.94
518.24 420.79
414.59 336.63
310.95 252.47
207.3 168.31
103.65 84.157
0 min 0 min

(e) (f )

Figure 10: Six modes of resonant natural frequency of the optimal z-stage: (a) Mode 1: resonant frequency of 336.11 Hz; (b) Mode 2:
resonant frequency of 589.53 Hz; (c) Mode 3: resonant frequency of 594.75 Hz; (d) Mode 4: resonant frequency of 596.36 Hz; (e) Mode 5:
resonant frequency of 601.63 Hz; (f ) Mode 6: resonant frequency of 605.49 Hz.

Table 12: Wilcoxon’s benchmark of the proposed algorithm vs. CSA for the safety factor.
Quantity for assessments Anticipated medium P-worth Wilcoxon statistic
35 0.237596 ≤0.001 630

Table 13: Wilcoxon’s benchmark of the proposed algorithm vs. CSA for the displacement
Quantity for assessments Anticipated medium P-worth Wilcoxon statistic
35 17.5383 ≤0.001 630
Mathematical Problems in Engineering 15

Table 14: Friedman tests for the safety factor.

Characteristic Quantity for assessments Medium Total of ranks
F1 via proposed algorithm 35 2.17429 70
F1 via CSA 35 2.38246 35
Total 70 2.27837
DF Chi-square P-worth
1 35 ≤0.001
Null hypothesis H0: all treatment effects are zero
Substitute hypothesis H1: not all treatment effects are zero

Table 15: Friedman test for the displacement.

Characteristic Quantity for assessments Medium Total of ranks
F1 via proposed algorithm 35 437.583 70
F1 via CSA 35 454.551 35
Total 70 446.067
DF Chi-square P-worth
1 35 ≤0.001
Null hypothesis H0: all treatment effects are zero
Substitute hypothesis H1: not all treatment effects are zero

Table 16: Error between anticipated results and verifications.

Characteristics Anticipation Confirmation Error (%)
F1 2.382457 2.224 7.12
F2 (mm) 454.5511 436.04 4.25

Table 17: Advancement between primary design and optimal design.

Characteristics Primary design Optimal design Advancement (%)
F1 2.224 2.292 3.708
F2 (mm) 436.04 468.076 18.498

Table 18: Comparison of the proposed z-stage with previous studies.

Studies Dimensions Output displacement (μm)
H. Huang et al. [10] 103 mm × 74 mm × 60 mm 11.44
H. Huang et al. [11] 200 mm × 135 mm × 200 mm 40
The proposed z-stage 301 mm × 162 mm × 16 mm 454

accomplished with the same constraints as well as input In addition, the proposed z-stage was compared with
displacement. Based on the predicted consequences of the previous studies, as illustrated in Table 18. Compared with
hybrid approach, a 3D model was established again for the the several commercialized nanoindentation devices, the
validation. The output z-displacement was about 436.04 μm, proposed z-stage has more the output displacement. In
and the safety factor was around 2.224. It can be clearly seen addition, this z-stage has small parasitic motion error for
that that the error among anticipated consequences and ensuring the indenting accuracy.
verifications for the displacement and safety factor stand at
4.245282% and 7.124857%, respectively, as depicted in Ta- 5. Conclusions
ble 16. Therefore, the anticipated consequences are a
trustworthy confirmation with verified consequences. This article proposed a novel design of the compliant z-stage
Additionally, the optimization consequences were better integrating the four-lever amplification mechanism, compliant
than the results of initial design, as illustrated in Table 17. hinge shifted arrangement mechanism, zigzag-based flexure
Moving to a more detailed analysis, an advancement for the spring guiding mechanism, and symmetric six leaf hinges-based
safety factor was about 3.708% as well as its ratio for the parallel guiding mechanism applied for nano/microindentation
displacement was around 18.498%. The consequences testing system and an efficient integration method of the TM,
demonstrate that the proposed combination approach is a FEM, response surface method, weight factor calculation
proficient methodology for solving multicriteria optimal method based on signal to noise and WOA in order to optimize
problem of the z-stage. the key geometrical parameters for improving the essential
16 Mathematical Problems in Engineering

responses of the z-stage such as the output displacement and the Acknowledgments
safety factor. For more specifics, the weight factors for both
responses were computed via forming the equations chain so as This work belongs to the project grant no. T2020-61TÐ
to enhance the calculation precision of optimal results. funded by Ho Chi Minh City University of Technology and
First of all, a preliminary design was tested and enhanced Education, Vietnam.
by FEA to gain good features according to the merits of a new
integrated compliant mechanism of the four-lever amplifier, References
zigzag guiding mechanism, and parallel guiding mechanism.
Then, the numerical experiment data were established by TM. [1] J. Nohava, N. X. Randall, and N. Conté, “Novel ultra nano-
Later on, the output displacement and the safety factor are indentation method with extremely low thermal drift: prin-
defined by FEA. Afterward, the regression functions among ciple and experimental results,” Journal of Materials Research,
four design variables and quality responses are formed via vol. 24, no. 3, pp. 873–882, 2009.
exploiting the response surface method. In addition, signal to [2] Z. Hu, K. J. Lynne, S. P. Markondapatnaikuni, and
noise ratios as well as the weight factors for every characteristic F. Delfanian, “Material elastic-plastic property characteriza-
tion by nanoindentation testing coupled with computer
are defined. Eventually, according to regression equations, the
modeling,” Materials Science and Engineering: A, vol. 587,
WOA is executed for determining the optimal parameters. pp. 268–282, 2013.
The results determined that the weight factors are 0.515 [3] Q. Xu, “Design and testing of a novel multi-stroke micro-
for F1 and 0.485 for F2 where both these values were allo- positioning system with variable resolutions,” Review of
cated in the WOA algorithm in order to resolve the mul- Scientific Instruments, vol. 85, no. 2, Article ID 025002, 2014.
ticriteria optimization trouble. The sensitivity analysis and [4] K. Lu, J. Zhang, W. Chen, J. Jiang, and W. Chen, “A
ANOVA were implemented for defining the influences and monolithic microgripper with high efficiency and high ac-
substantial contributions of input variables on the output curacy for optical fiber assembly,” in Proceedings of the 2014
characteristics. The consequences revealed that the optimal 9th IEEE Conference on Industrial Electronics and Applica-
factors were detected at M � 0.55 mm, N � 0.82377 mm, tions, pp. 1942–1947, Hangzhou, China, June 2014.
P � 0.9 mm, and K � 49 mm. Besides, the results revealed that [5] A. S. Putra, S. Huang, K. K. Tan, S. K. Panda, and T. H. Lee,
the optimization safety factor is 2.382456817 as well as the “Design, modeling, and control of piezoelectric actuators for
intracytoplasmic sperm injection,” IEEE Transactions on
optimization displacement is around 454.5511266 mm.
Control Systems Technology, vol. 15, no. 5, pp. 879–890, 2007.
Based on the results of the Wilcoxon and Friedman
[6] P. Wang and Q. Xu, “Design of a flexure-based constant-force
nonparametrics testing methods, the proposed algorithm was XY precision positioning stage,” Mechanism and Machine
better than the cuckoo search algorithm. Additionally, the Theory, vol. 108, no. 2017, pp. 1–13, 2017.
consequences demonstrated that the errors among the opti- [7] S. Fan, H. Liu, and D. Fan, “Design and development of a
mization consequences and the FEA validations for the safety novel monolithic compliant XY stage with centimeter travel
factor and z-axis deformation are 7.12% and 4.25%, corre- range and high payload capacity,” Mechanical Sciences, vol. 9,
spondingly. The FEA affirmations were nearby with the an- no. 1, pp. 161–176, 2018.
ticipated consequences from the integration method. [8] S. P. Wadikhaye, Y. K. Yong, and S. O. R. Moheimani, “Design
Moreover, the optimization consequences were better than its of a compact serial-kinematic scanner for high-speed atomic
initial design. For specifics, after optimal process, an ad- force microscopy: an analytical approach,” Micro & Nano
vancement for the safety factor and the deformation was Letters, vol. 7, no. 4, pp. 309–313, 2012.
around 3.708% and 18.498%, correspondingly. [9] M. Gauthier and E. Piat, “Control of a particular micro-macro
positioning system Applied to cell micromanipulation,” IEEE
In addition, this research meets several challenges such
Transactions on Automation Science and Engineering, vol. 3,
as taking lots of time for creating the great initial design with no. 3, pp. 264–271, 2006.
good performances as well as proposing an effective ap- [10] H. Huang, H. Zhao, J. Mi et al., “A novel and compact
proach for multiobjective optimization problem. Therefore, nanoindentation device for in situ nanoindentation tests
topology optimization should be considered to make an inside the scanning electron microscope,” AIP Advances,
initial design of structure, reduce the time consuming, and vol. 2, no. 1, Article ID 012104, 2012.
enhance the characteristics of the design. [11] H. Huang, H. Zhao, J. Mi et al., “Experimental research on a
In the future study, a prototype will be manufactured modular miniaturization nanoindentation device,” Review of
and its manners will be checked for assessing with the Scientific Instruments, vol. 82, no. 9, Article ID 095101, 2011.
numerical analysis consequences. [12] W.-L. Zhu, Z. Zhu, P. Guo, and B.-F. Ju, “A novel hybrid
actuation mechanism based XY nanopositioning stage with
totally decoupled kinematics,” Mechanical Systems and Signal
Data Availability Processing, vol. 99, pp. 747–759, 2018.
[13] S. H. Chang and B. C. Du, “A precision piezodriven
The data used to support the findings of this study are in-
micropositioner mechanism with large travel range,” Review
cluded within the article. of Scientific Instruments, vol. 69, no. 4, pp. 1785–1791, 1998.
[14] F. Wang, C. Liang, Y. Tian, X. Zhao, and D. Zhang, “Design
Conflicts of Interest and control of a compliant microgripper with a large am-
plification ratio for high-speed micro manipulation,” IEEE/
The authors declare that they have no conflicts of interest ASME Transactions on Mechatronics, vol. 21, no. 3,
regarding the publication of this article. pp. 1262–1271, 2016.
Mathematical Problems in Engineering 17

[15] H. Tang, Y. Li, and X. Xiao, “Development and assessment of a [31] R. V. Rao, V. J. Savsani, and D. P. Vakharia, “Teaching-
novel hydraulic displacement amplifier for piezo-actuated learning-based optimization: a novel method for constrained
large stroke precision positioning,” in Proceedings of the 2013 mechanical design optimization problems,” Computer-Aided
IEEE International Conference on Robotics and Automation, Design, vol. 43, no. 3, pp. 303–315, 2011.
pp. 1409–1414, Karlsruhe, Germany, May 2013. [32] R. V. Rao, V. J. Savsani, and D. P. Vakharia, “Teaching-
[16] Q. Xu, “Structure design of a new compliant gripper based on learning-based optimization: an optimization method for
Scott-Russell mechanism,” in Proceedings of the 2013 IEEE continuous non-linear large scale problems,” Information
International Conference on Robotics and Biomimetics Sciences, vol. 183, no. 1, pp. 1–15, 2012.
(ROBIO), Shenzhen, China, December 2013. [33] N. Le Chau, N. L. Ho, N. T. Tran, and T.-P. Dao, “Analytical
[17] H.-Y. Kim, D.-H. Ahn, and D.-G. Gweon, “Development of a model and computing optimization of a compliant gripper for
novel 3-degrees of freedom flexure based positioning system,” the assembly system of mini direct-current motor,” Inter-
Review of Scientific Instruments, vol. 83, no. 5, Article ID national Journal of Ambient Computing and Intelligence,
055114, 2012. vol. 12, no. 1, pp. 1–28, 2021.
[18] S. Mirjalili and A. Lewis, “The whale optimization algorithm,” [34] R. V. Rao and R. B. Pawar, “Constrained design optimization
Advances in Engineering Software, vol. 95, pp. 51–67, 2016. of selected mechanical system components using Rao algo-
[19] J. A. Ghani, I. A. Choudhury, and H. H. Hassan, “Application rithms,” Applied Soft Computing, vol. 89, Article ID 106141,
of Taguchi method in the optimization of end milling pa- 2020.
rameters,” Journal of Materials Processing Technology, [35] T.-H. Hou, C.-H. Su, and W.-L. Liu, “Parameters optimiza-
vol. 145, no. 1, pp. 84–92, 2004. tion of a nano-particle wet milling process using the Taguchi
[20] C. Kandilli and B. Mertoglu, “Optimisation design and op- method, response surface method and genetic algorithm,”
eration parameters of a photovoltaic thermal system inte- Powder Technology, vol. 173, no. 3, pp. 153–162, 2007.
grated with natural zeolite,” International Journal of [36] M. P. Dang, H. G. Le, N. Le Chau, and T. Dao, “Optimization
Hydromechatronics, vol. 3, no. 2, pp. 128–139, 2020. for a flexure hinge using an effective hybrid approach of fuzzy
[21] M. Safa, M. Ahmadi, J. Mehrmashadi et al., “Selection of the logic and moth-flame,” Optimization Algorithm, vol. 2021,
most influential parameters on vectorial crystal growth of Article ID 6622655, 2021.
highly oriented vertically aligned carbon nanotubes by [37] H. Garg, “A hybrid PSO-GA algorithm for constrained op-
adaptive neuro-fuzzy technique selection of the most influ- timization problems,” Applied Mathematics and Computa-
ential parameters on vectorial crystal growth of highly ori- tion, vol. 274, pp. 292–305, 2016.
ented vertically aligned carbon nanotubes by adaptive neuro- [38] H. Bhattacharya, S. Chattopadhyay, M. Chattopadhyay, and
fuzzy technique,” vol. 3, no. 3, pp. 238–251, 2020. A. Banerjee, “Storage and bandwidth optimized reliable
[22] Z. Ali and T. Mahmood, “Complex neutrosophic generalized distributed data allocation algorithm,” International Journal
dice similarity measures and their application to decision of Ambient Computing and Intelligence, vol. 10, no. 1,
making,” CAAI Transactions on Intelligence Technology, vol. 5, pp. 78–95, 2019.
no. 1, 2020. [39] N. Le Chau, H. G. Le, T. Dao, M. P. Dang, and V. A. Dang,
[23] B. R. Murlidhar, R. K. Sinha, E. T. Mohamad, R. Sonkar, and “Efficient hybrid method of FEA-based RSM and PSO al-
M. Khorami, “The effects of particle swarm optimisation and gorithm for multi-objective optimization design for a com-
genetic algorithm on ANN results in predicting pile bearing pliant rotary joint for upper limb assistive device,”
capacity,” International Journal of Hydromechatronics, vol. 3, Mathematical Problems in Engineering, vol. 2019, Article ID
no. 1, pp. 69–87, 2020. 2587373, , 2019.
[24] M. Kaur and D. Singh, “Multiobjective evolutionary opti- [40] H. Garg, “A hybrid GSA-GA algorithm for constrained op-
mization techniques based hyperchaotic map and their ap- timization problems,” Information Sciences, vol. 478,
plications in image encryption,” Multidimensional Systems pp. 499–523, 2019.
and Signal Processing, vol. 32, no. 1, pp. 281–301, 2020. [41] H. Garg, “A hybrid GA-GSA algorithm for optimizing the
[25] M. Kaur, D. Singh, and V. Kumar, “Color image encryption performance of an industrial system by utilizing uncertain
using minimax differential evolution-based 7D hyper-chaotic data,” Handbook of Research on Artificial Intelligence Tech-
map,” Applied Physics B, vol. 126, no. 9, 2020. niques and Algorithms, pp. 620–654, 2015.
[26] H. Golpı̂ra, H. Bevrani, and H. Golpı̂ra, “Application of GA [42] S.-C. Huang and T.-P. Dao, “Multi-objective optimal design
optimization for automatic generation control design in an of a 2-DOF flexure-based mechanism using hybrid approach
interconnected power system,” Energy Conversion and of grey-taguchi coupled response surface methodology and
Management, vol. 52, no. 5, pp. 2247–2255, 2011. entropy measurement,” Arabian Journal for Science and
[27] M. Basu and A. Chowdhury, “Cuckoo search algorithm for Engineering, vol. 41, no. 12, pp. 5215–5231, 2016.
economic dispatch,” Energy, vol. 60, pp. 99–108, 2013. [43] S. Chatterjee, S. Sarkar, S. Hore, N. Dey, A. S. Ashour, and
[28] A. Chander, A. Chatterjee, and P. Siarry, “A new social and V. E. Balas, “Particle swarm optimization trained neural
momentum component adaptive PSO algorithm for image network for structural failure prediction of multistoried RC
segmentation,” Expert Systems with Applications, vol. 38, buildings,” Neural Computing and Applications, vol. 28, no. 8,
no. 5, pp. 4998–5004, 2011. pp. 2005–2016, 2016.
[29] C. Prakash, S. Singh, M. Singh, K. Verma, B. Chaudhary, and [44] C. Prakash, H. K. Kansal, B. S. Pabla, and S. Puri, “Multi-
S. Singh, “Multi-objective particle swarm optimization of objective optimization of powder mixed electric discharge
EDM parameters to deposit HA-coating on biodegradable machining parameters for fabrication of biocompatible layer
Mg-alloy,” Vacuum, vol. 158, pp. 180–190, 2018. on β-Ti alloy using NSGA-II coupled with Taguchi based
[30] R. V. Rao and V. Patel, “Multi-objective optimization of heat response surface methodology,” Journal of Mechanical Science
exchangers using a modified teaching-learning-based opti- and Technology, vol. 30, no. 9, pp. 4195–4204, 2016.
mization algorithm,” Applied Mathematical Modelling, [45] N. Le Chau, N. T. Tran, and T.-P. Dao, “A multi-response
vol. 37, no. 3, pp. 1147–1162, 2013. optimal design of bistable compliant mechanism using
18 Mathematical Problems in Engineering

efficient approach of desirability, fuzzy logic, ANFIS and

LAPO algorithm,” Applied Soft Computing, vol. 94, Article ID
106486, 2020.
[46] T.-P. Dao, “Multiresponse optimization of a compliant
guiding mechanism using hybrid taguchi-grey based fuzzy
logic approach,” Mathematical Problems in Engineering,
vol. 2016, Article ID 5386893, 17 pages, 2016.
[47] M. P. Dang, H. G. Le, N. Le Chau, and T. P. Dao, “A multi-
objective optimization design for a new linear compliant
mechanism,” Optimization and Engineering, vol. 21, no. 3,
2020.
[48] D.-N. Nguyen, T.-P. Dao, C. Prakash et al., “Machining
parameter optimization in shear thickening polishing of gear
surfaces,” Journal of Materials Research and Technology, vol. 9,
no. 3, pp. 5112–5126, 2020.
[49] H. M. Mohammed, S. U. Umar, and T. A. Rashid, “A sys-
tematic and meta-analysis survey of Whale optimization al-
gorithm,” Computational Intelligence and Neuroscience,
vol. 2019, Article ID 8718571, 2019.
[50] Q. H. Zhai, T. Ye, M. X. Huang, S. L. Feng, and H. Li, “Whale
optimization algorithm for multiconstraint second-order
stochastic dominance portfolio optimization,” Computational
Intelligence and Neuroscience, vol. 2020, Article ID 8834162,
2020.
[51] I. Aljarah, H. Faris, and S. Mirjalili, “Optimizing connection
weights in neural networks using the whale optimization
algorithm,” Soft Computing, vol. 22, no. 1, pp. 1–15, 2016.
[52] G. Kaur and S. Arora, “Chaotic whale optimization algo-
rithm,” Journal of Computational Design and Engineering,
vol. 5, no. 3, pp. 275–284, 2018.
[53] T.-P. Dao, S.-C. Huang, and N. Le Chau, “Robust parameter
design for a compliant microgripper based on hybrid Taguchi-
differential evolution algorithm,” Microsystem Technologies,
vol. 24, no. 3, pp. 1461–1477, 2017.
[54] A. C. Analysis, “Statistical study of machine learning algo-
rithms using parametric and non-parametric tests,” Inter-
national Journal of Ambient Computing and Intelligence,
vol. 11, no. 3, pp. 80–105, 2020.
[55] A. H. Gandomi, X.-S. Yang, and A. H. Alavi, “Cuckoo search
algorithm: a metaheuristic approach to solve structural op-
timization problems,” Engineering with Computers, vol. 29,
no. 1, pp. 17–35, 2013.