A review of decoupling control based on multiple models

Guohai Liu, Zhaoxia Wang, Congli Mei, Yuhan Ding

School of Electrical and Information Engineering, Jiangsu University, Zhenjiang, 212013
E-mail: ghliu@ujs.edu.cn

Abstract: In this article, we review and discuss methods for decoupling control which include conventional decoupling
and adaptive decoupling. Decoupling methods which is used to solve the multivariable coupled in nonlinear control are
always a hot issue. The major algorithms such as neural network and fuzzy control are described in details and critically
reviewed in this work. The advantages and disadvantages of these algorithms are analyzed. The review reveals the
tremendous prospect of decoupling algorithms in nonlinear control. It is important to seek effective and simple
decoupling methods.
Key words: multivariable decoupling control, adaptive decoupling control, neural network, fuzzy control

uncertainties, multivariable and strong coupling, but also

large degree of complexity and nonlinearity.
1. INTRODUCTION ŁThe purpose of decoupling is to eliminate interference
The design of decoupling for multivariable system has among circuits and loops, so the variety of an output
shaped in the early development of control field and requires only affected by the specific input. Since a
multivariable model is established for process control in complex multivariable system is transformed into several
practical industry. Most industrial processes are single variable systems, it is important to design the
characterized with large uncertainty, time-varying and decoupling controller for choosing the advisable input and
strong coupling. Variables in multivariable systems usually output.
exist coupling and its variables or parameters are uncertain. łMost decoupled systems require that the closed- loop
Part of these parameters may be time-varying or unknown system can be adjusted online according to the needs of
and also may have random unknown interference, so it is control or the change of conditions. However, the controller
essential to decouple for multivariable system to obtain cannot be adjusted online directly due to there exists so
desired control effects. much association among variables.
There are many problems in complex industrial process ŃThe variables of coupled system cannot be considered
with characteristics of multivariable, nonlinear, individually, so these variables need to be adjusted in real
time-varying, and dynamic. It cannot achieve satisfactory time to obtain reasonable results. However, it usually
results and meet the needs of industrial production using cannot meet the require.
traditional decoupling methods. Fixed-value control, PID Thus, it is necessary to decouple multivariable system
control and other conventional control methods are used to into single variable system in nonlinear time-varying
obverse the parameters of the fermentation process for most multivariable system with strong coupling.
existing fermentation control. For the process of growth
and reproduction of microorganism, a complex 3 THE CURRENT SITUATION OF
multivariable system with strong coupling [1, 2], is difficult DECOUPLING CONTROL
to model using mechanism and realize decoupling.
Experienced operators can identify by the experience of Multivariable system must be decoupled in order to
operation. The growth and reproduction of fermentation are obtain satisfactory control effect in nonlinear coupled
affected when an operating mistake occurs which even systems. According to the degree of decoupling, it can be
makes the whole fermentation tanks scrapped and causes divided into the whole and approximate decoupling.
heavy economic losses. Therefore, the design requirement According to the time characteristics of decoupling, it can
for decoupling has become higher because of the be divided into static and dynamic decoupling. In this
characteristics of complex multivariable process. Nowdays, paper, the most common traditional decoupling methods
transforming the multivariable system into single variable are analyzed. Adaptive decoupling control and its
system has become a hot issue in the research field of application are introduced.
control theory and control engineering. 3.1 THE CONVENTIONAL DECOUPLING
Modern frequency-domain method is the representative
2. THE IMPORTANCE OF DECOUPLING of the conventional decoupling control methods, applied to
CONTROL MIMO linear systems, which also include time-domain
To deal with couplings in multivariable nonlinear methods. These methods [3-5] are mainly consist of
systems, decoupling control had received extensive Diagonal Matrix Synthesis, Characteristic Locus Analysis,
research during the last few decades. However, most Dyadic Expansion, Sequence Return Difference, Inverse
multiple control systems not only have properties of Nyquist Array Method and Singular Value Decomposition.
Diagonal Matrix Synthesis is presented for dual-variable
This work is a Project Funded by the Priority Academic Program control system in order to remove the linkages between the
Development of Jiangsu Higher Education Institutions under Grant two control loops using matrix transformation. It makes the
[2011]6, and a project supported by agricultural science and technology of
Zhenjiang under Grant 2007AA04Z179.

978-1-4577-2074-1/12/$26.00 2012
c IEEE 1077

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 07:17:13 UTC from IEEE Xplore. Restrictions apply.
 system matrix transformed into a diagonal matrix and then decoupling control combined adaptive control with
can be achieved by feed-forward compensation. decoupling technology has its unique advantages and
Characteristic Locus Analysis is a precise method application on solving the adjustment of parameters
analyzing the state of MIMO system, which has important automatically in the complex industrial process. The
meaning for decoupling theory of MIMO system. But the purpose of adaptive decoupling control [6,7] is to transform
accuracy of process which need for approximation is not closed-loop system transfer function into a diagonal form.
guaranteed during the realization of compensation. Optimal control is used to established the objective function
Dyadic Expansion which solves the difficult and optimize parameters, which is fit for time-varying
achievement of the root locus gain in Characteristic Locus systems.
Analysis is an application of characteristic curve. However, For uncertain linear multivariable systems, adaptive
its function cannot follow the law which achieved by decoupling control had received extensive research.
attempting and it must consider precision of measurement. Different decoupling designs for adaptive controllers had,
Sequence Return Difference which strings the controller respectively, been proposed in this paper.
of compensation one by one and the feedback of closed 3.2.1 Adaptive decoupling control with pole placement
loop, solves the multivariable control problem in the design
This method firstly uses decoupled pre-compensation for
of single-variable. Considering all aspects of performance,
decoupling design, and then applies the traditional pole
it is easy to implement computer programming. But it
assignment to control the decoupled SISO system[8]. The
designs the system only in order without better way to
advantages in the design of pre-decoupled controller
choose the root locus gain.
consist of avoiding unstable pole-zero cancellation,
Based on the advantage of the diagonal matrix, Inverse
ensuring system stability and tracking a bounded output
Nyquist Array Method which has a certain robustness
which can reduce the online computation largely. However,
realizes the pre-compensation easily and simply by using
this mean requires that the pole-zero dots are cancelled
classic design. In addition, it can transform the closed loop
precisely and the estimated parameters converge to the true
function matrix into the diagonal advantaged matrix.
value, so this algorithm is complex to achieve.
However, it cannot guarantee that the closed loop function
3.2.2 Adaptive decoupling control based on generalized
matrix has the diagonal advantage. The system may exist
minimum variance
association which cannot be ignored, so it needs to get the
parameters of compensate controller repeatedly to meet the This method designs decoupled controller for random
idea of diagonal advantage. multivariable systems with general interact matrix using the
Singular Value Decomposition is used widely in the integration idea of open-loop decoupled compensator and
study of system control and has characteristic of low generalized minimum variance with self-adapting. Its
sensitivity and high degree of numerical calculation in the merits lie in identifying the parameters of controller online,
design of multivariable system. then avoiding the need for unstable zero problems of
The traditional thought of decoupling control transforms open-loop decoupling control algorithm. Practices show
the function matrix of generalized system which consisted that this method can give a good result adapting to the
of compensation controller and the controlled objects into changing conditions[9]. But it is difficult to realize in a
diagonal matrix, then a coupled multivariable system distributed computer control system.
change into several single-variable system without 3.2.3 Adaptive decoupling control based on feedback
coupling. These decoupling methods rely heavily on control
precise mathematical model of controlled object. However,
most industry control processes are uncertain nonlinear and The method of feedback decoupling considers the
coupled. When the conditions change, the traditional influence of this channel to others is to weaken or eliminate
coupling. Due to its complex mechanism, the industry
decoupling methods are difficult to meet the needs of actual
process characteristics of nonlinear are difficult to be
control. This requires the novel decoupling methods based
on traditional idea and added other methods in order to described precisely using mathematical model. Adaptive
meet the actual needs. decoupling control for random multivariable system with
transmission delay can be realized by introduced this
3.2 ADAPTIVE DECOUPLING METHODS method. In addition, neural network method can be added
INTEGRATED WITH OTHER METHODS and combined with feedback control [2]. The effect of
higher order nonlinear eliminated by feedback
Adaptive control makes target output track the desired compensation could be considered as interference, and then
output by designing parameter adaptive laws. It can adaptive decoupling algorithm is used to achieve neural
compensate for the unpredictable change of model, network adaptive decoupling control [10, 11].
parameters and input singles automatically. When Alternatively, neural network can be used to realize
characteristics and surrounding environment is partially feedback compensation [12], which makes compensated
known or normal operation change, it can adjust the control system decoupled and the decoupled SISO systems keep
law automatically by the practical operation of objects in the characteristics of main channels. Combined with
order to achieve the desired objects. The state, performance multi-model adaptive control, multivariable feedback
and parameters of the objects must be updated timely to adaptive decoupling method [13] design the adaptive
define the current running index of system and to adjust the decoupling control law. According to the performance
structure and parameters. This can guarantee that the indicators of system, the method achieves the control
system runs in the state of optimal or sub-optimal. Adaptive

1078 2012 24th Chinese Control and Decision Conference (CCDC)

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 07:17:13 UTC from IEEE Xplore. Restrictions apply.
purpose by switching two models. It cancels the state time-domain response curve. Then the controller for each
require for system zero dynamics. decoupled loop could be designed.
3.2.4 Multivariable adaptive decoupling control based A class of nonlinear multivariable systems have been
on generalized predictive control divided into two parts [21], the linear term of the system
identified by a trained neural network offline and the
Generalized predictive control adopts the ways of
nonlinear term of the system viewed as the measured
multiple forecasting and rolling optimizations to obtain
disturbance and eliminated using feed-forward strategy.
better dynamic quality and better control robustness [14].
Neural network is used to approximate passages coupling
The design of generalized predictive linked with feedback
[22], nonlinear and unmodeled dynamics which could make
decoupling control brings out multivariable adaptive
identification much better adding improved RLS
decoupling control, which is appropriate for generalized
identification algorithm. Then control precision and
predictive control in DCS. The advantages of this mean in
robustness for adaptive control will be improved. The
control systems, such as turbine, large-scale cookery, lie in
singular problem of the controller is solved by practical
modeling simply and having better robustness, but these
neural network control and an appropriate
systems can only realize approximated decoupling.
Lyapunov-Krasovskii [23] is used to compensate for the
3.2.5 Multivariable PID adaptive decoupling control
unknown time delays. In this paper, the continue function is
Decoupling control law is divided into open loop approximated by RBF neural network. The presented
decoupling control law and close loop decoupling control design can be carried out in the decoupled back stepping
law. Open-loop decoupling control law is consisted of the design and the control can be carried out in a nested matter
object’s decoupled compensator and controller. to guarantee semi-global ultimate boundedness of all
Multivariable system is transformed into several SISO signals in the closed-loop system.
systems by compensating, while the desired performance 3.2.7 Adaptive decoupling control based on fuzzy
will be achieved by the controllers of SISO systems. At control
present, PID control law belongs to more typical and riper
The selection of control rules for conventional fuzzy
method than others in the design for controller. The
controller has great impact on the result of control. Fuzzy
parameters of conventional PID controller are obtained by
control does not rely on the model of object and has
artificial attempting. When the operating conditions
advantages of fast control and good robust. In order to
change, the quality of controller is likely to deteriorate and
avoid chattering phenomenon, fuzzy control [24] is
that may lead to run unusually. The conventional PID
introduced based on sliding model control so that a class of
control combined with multivariable adaptive decoupling
fourth-order nonlinear system can be decoupled into two
control [15], united design, will have a significant effect on
second-order subsystems which have a separate control
the application in industrial process and have advantages of
purpose expressed in terms of a sliding surface. The
simple structure, easy to implement, better robustness and
adaptive fuzzy sliding model control is applied to regulate
better stability. Zhai L F proposed PID decoupling
the parameters of the fuzzy controller which is used to
controller [16] for multivariable system, which required the
approximate the equivalent control of decoupled sliding
parameters of balance known and did not give way to
model control. Considering the perspective of the control
eliminate the error of stable state. On the basis of this, this
object, a cascade-connected self-adaptive fuzzy-neural
paper put forward multivariable adaptive decoupling
network decoupling controller [25] to improve the
controller[17] for nonlinear discrete-time systems with
dynamical property and decoupling capability for
strong coupling combined PID controller with multiple
multivariable nonlinear systems with strong coupling
models, which included a fully adaptive PID decoupling
factors is presented. Self-tuning fuzzy controller is
controller, a function with parameters re-assignment, a
designed by using the intelligent weight function rulers and
number of fixed PID decoupling controller and a switching
self-adaptive neural network decoupling controller is based
mechanism. This cancels that the parameters of balance
on the learning algorithm of dynamical coupling
must be known and gives way to eliminate stable state
characteristic. A class of under actuated systems with
error. The performance of the system can be guaranteed at
strong coupling [26] has been decoupled into several
this condition.
subsystems by using the concept of hierarchical decoupling
3.2.6 Adaptive decoupling control based on neural
method. To design the adaptive control law, adaptive fuzzy
network
inference system is used to approximate the unknown
Neural networks have abilities of self-study, coupling factors and optimal coupling factors can be
generalization and nonlinear mapping and its function can estimated automatically.
be used to compensate [18-20] for the uncertain part of the 3.2.8 Adaptive decoupling control using reference
unknown systems or approximate it. Adaptive decoupling model
method integrated the means, such as neural network,
The design of the adaptive law [23] for control system is
self-adaptive control, adopts neural network to achieve the
the primary target of adaptive decoupling control. A
purpose of identification or approximation for a class of
nonlinear-state-feedback speed control scheme[27] for the
nonlinear dynamical discrete time systems. Adding a neural
induction-motor drive with unknown load torque using the
network decoupling controller before multivariable
adaptive rotor-flux estimator insensitive to both rotor and
nonlinear system, generalized objective can be transformed
stator-resistance variations has been developed, which
into a system without coupling or with little coupling after
decouples the control of flux and speed. In this paper, a
the effects are eliminated by the training of generalized

2012 24th Chinese Control and Decision Conference (CCDC) 1079

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 07:17:13 UTC from IEEE Xplore. Restrictions apply.
 stable model reference adaptive system rotor-resistance 4. CONCLUSION
estimator is investigated and an error signal between two
calculated air-gap reactive powers is used to correct For uncertain nonlinear multivariable systems, adaptive
adaptively the rotor-resistance error. A decoupled adaptive decoupling control had been researched and different
controller [28] for controlling the dissolved oxygen and adaptive controllers with decoupling designs had,
glucose feeding simultaneously in fed-batch fermentation respectively, been proposed. However, many decoupling
production has been designed. The adaptive controller for theories are difficult to generalize for its design way and
decoupled the system is presented using a linear reference complex algorithm. Adaptive decoupling control is based
model for controlling these two variables. The adaption on the traditional decoupling methods which is inclined to
gain coefficient is an important parameter and has a investigate the design of controller and has been
significant effect on the performance of the adaptive successfully applied in some areas. At present, the designs
controller. Within a certain range, a larger gain can achieve of adaptive decoupling controllers are usually achieved by
better test result. attempting, training or optimizing. But its theory is not
3.2.9 Adaptive decoupling control based on feedback perfect and is mainly proved by Lyapunov theorem, which
has a lot of work to do. If we want to use widely, we must
The almost disturbance problem is formulated and investigate into the combination with multivariable
solved by the design of feedback controls which attenuates decoupling technology and actual industrial process much
up to an arbitrary degree the influence of unknown bounded deeper.
disturbances on the output. For a class of SISO uncertain
nonlinear systems[29] influenced by both uncertain REFERENCES˖
constant parameters and unknown time-varying bounded
[1]G. H. Liu, Y. Kun, L. Quan, X. X. Liu, X. Q. Liu, Decoupling control of
disturbances, the output feedback disturbance decoupling multivariable biology fermentation process, Journal of Northeast
control is improved by adding an adaptive part so that University(Natural Science),34(sup)˖155-159, 2004.
asymptotic tracking is achieved in the presence of constant [2]Y. Q. Chang, Y. C. Li, Z. Lv, F. L. Wang, Multivariable feed forward
parameters. The adaptive part consists of reduced order decoupling control based on double-Level neural network in a
observer dynamics, parameter estimation dynamics with fermentation process, Journal of Northeast University(Natural
Science),Vol.28,No.7,925-928,2007.
projection and auxiliary signals generated by output filers. [3]Y. X. Sun, Multivariable decoupling control system, Journal of
Adaptive neural network control [30] is presented for a Chekiang University, 1980.
class of block triangular MIMO nonlinear discrete-time [4]P. Cheng, Multivariable linear control system, Beijing, Press of Beijing
systems with every subsystem in pure-feedback form with University Aeronautics and Astronautics, 1990.
[5]X. H. Gu, L. F. Zhou, Linear Multivariable system frequency –domain
unknown control directions. The MIMO system is method, Shanghai Press of Shanghai Jiaotong University,1990.
transformed into a predictor form by rigorous derivation in [6]Yu Wen, Adaptive decoupling controller, [Graduate Paper], Shenyang,
order to avoid no causal problem in the control design Northeastern University, 1994.
process. The implicit controls are developed for every [7]J.Min,Z.C.Huang,Method of multivariable decoupling control,Control
Engineering of China,125-127,2005
subsystem so that the couplings of inputs and states among [8]B.Wittenmark,R.Middleton,G.C.Goodwin,Adaptive decoupling of
subsystems can be decoupled completely by exploring the multivariable systems,Int J Control,1993-2009,1987
properties of the block triangular form. [9]T.Y.Chai,G.Wang,Globally convergent multivariable adaptive
In this paper, most relevant design and methods for decoupling controller and its application to a binary distillation
column,Int J Control,415-429,1992
decoupling in multivariable system are listed in detail. [10] P. Ma, J. F. Yang, C. C. Cui, S. K. Hu, Current situation and
There also have other methods which are not mentioned development of decoupling control, controll engineering of china,
that amalgamate adaptive decoupling control with other Vol. 12, No. 2, 97-100, 2005.
methods such as using single-neuron PID [31], employing [11] B. H. Sang, X. Z. Xue, A Summary of multivariable decoupling
control methods, Fire Control and Command Control, Vol.32,
dimension-by-dimension approach [32], based on state No.11, 1297-1300, 2007.
observer [33]. [12] Q. B. Jin, D. N. Zeng, Y. H. Wang, S. S. Gu, Multiple models neural
Traditional decoupling control methods are difficult to network decoupling new method, Journal of Northeast University
guarantee the quality of decoupling without adaptive law. (Natural Science), Vol.20, No.3, 250-253, 1999.
[13] Y. Fu, T. Y. Chai, H. Yue, Multiple models adaptive decoupling
Even it can lead to unstable state of controlled system when control for a class of nonlinear multivariable systems, Control and
the process changes. Most multivariable systems are Decision,Vol.21, No.2, 139-142, 2006.
uncertain, nonlinear and coupled. When the operation [14] Z. J. Wang, J. H. Li, T. Q. Zhang, Application of generalized
conditions change, decoupled controller without adaptive prediction adaptive control algorithm in the design of missile control
system, Journal of Beijing Institute of Technology, Vol.10, No.4,
law cannot modify parameters to meet the control effect 356-363, 2001.
and guarantee the stabilization of MIMO systems. [15] W. Liu, X. J. Feng, Design of Adaptive PID Decoupling controller
Multivariable adaptive decoupling control method is used based on multiple models, Micro-Computer Information
in industry process. It avoids the disadvantage of traditional (Monitoring and Control Automation), Vol.20,No.4,27-28, 2004.
[16] L. F. Zhai, T. Y. Chai. Nonlinear decoupling PID control using
decoupling methods which is not fit for uncertain nonlinear neural networks and multiple models, Journal of Control Theory and
industrial process and is applied in many fields such as Applications, Vol.2, No.1, 62-69, 2006.
industrial electric furnace, electric machine[34-36], [17] L. Li, Z. Z. Mao, F. Liu, Nonlinear Adaptive PID decoupling
distillation tower, and large-scale thermal units. Thus, in controller based on multiple models, Journal of Northeast
University(Natural Science),Vol.31,No.11,1546-1549,2010.
solving automation of complex industrial control process [18] Beibei Ren, Shuzhi Sam Ge, Chun-Yi Su, Tong Heng Lee, Adaptive
multivariable adaptive decoupling control has its unique neural control for a class of uncertain nonlinear systems in
advantages and wide application prospect. pure-feedback form with hysteresis input, IEEE Trans. on Systems,

1080 2012 24th Chinese Control and Decision Conference (CCDC)

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 07:17:13 UTC from IEEE Xplore. Restrictions apply.
 Man, and, Cybernetics—Part b: Cybernetics,Vol.39,No.2,431-443, using linear reference model. American Control conference,
2009. 5862-5867, 2010.
[19] Y. Fu, T. Y. Chai, Neural-network-based nonlinear adaptive [29] R. Marino, P. Tomei, Adaptive output feedback tracking with almost
dynamical decoupling control, IEEE Trans. on Neural disturbance decoupling for a class of nonlinear systems. Automatica,
Network,Vol.18,No.3,921-925,2007. 1871-1877, 2000.
[20] Y. Fu, T. Y. Chai, Intelligent Decoupling control of nonlinear [30] Yanan Li, Chenguang Yang, Shuzhi Sam Ge, Tong Heng Lee,
multivariable systems and its application to a wind tunnel system, Adaptive output feedback NN control of a class of discrete-Time
IEEE Trans. on Control Systems Technology, Vol.17, No.6, MIMO nonlinear systems with unknown control directions. IEEE
1376-1384,2009 Trans. on Systems, Man, and Cybernetics—Part B: Cybernetics,
[21] X. Wang, S. Y. Li, H. Yue, Multiple models neural network Vol.41, No.2, 507-517, 2011.
decoupling controller for a nonlinear system, Control and [31] M. Q. Yan, G. L. Yang, Z. X. Li, Adaptive neuron PID controller for
Decision,Vol.19,No.4,424-427,2004. the load control system of a thermal power unit, Mechanical &
[22] X. H. Shi, A Design Method for adaptive generalized predictive Electrical Engineering Magazine, Vol.21, No.12,23-26, 2004.
decoupling controller based on a neural network, Journal of System [32] X. Wang, Z. J. Wang, H. Yang, S. Y. Li, Multiple-model adaptive
Simulation, Vol.17, No.1, 178-180, 2005. decoupling controller employing dimension- by- dimension
[23] Fan Hong, Shuzhi Sam Ge, Tong Heng Lee, Practical adaptive approach, Control Theory&Applications, Vol.23, No.5, 711- 723,
neural control of nonlinear systems with unknown time delays, IEEE 2006.
Trans. on Systems, Man, and Cybernetics—Part B,Vol.35 [33] S. X. Su, Design of fuzzy-neural network decoupling controller for
No.4,849-854,2005. multiple models, CIESC Journal, Vol.61, No.8, 1949-1954, 2010.
[24] D.-W. Kim, H.-W. Yang, Ji-sup Yoon, Hong-pii Kim. Decoupled [34] David Díaz Reigosa, Pablo García,,Fernando Briz, Dejan Raca,
adaptive fuzzy sliding-mode control. IEEE International Symposium Member, Robert D. Lorenz, Modeling and Adaptive Decoupling of
on Computational Intelligence in Robotics and Automation, 96-101, High-Frequency Resistance and Temperature Effects in
2001. Carrier-Based Sensorless Control of PM Synchronous Machines,
[25] H. Li, Design of Multivariate Fuzzy-neural Network Decoupling IEEE Trans. on Industry Application,Vol.46,No.1,139-149,
Controller, Control and Decision, Vol.21, No5, 593-596, 2006. 2010
[26] Seung-Yong Shin, Joon-Yong Lee, Masanori Sugisaka, Ju-Jang [35] P. P. Acarnley and J. F. Watson, Review of position-sensorless
Lee, Decoupled fuzzy adaptive sliding mode control for operation of brushless permanent-magnet machines,IEEE Trans.
under-actuated systems with mismatched uncertainties. IEEE Ind. Electron.,vol. 53, No. 2, 352–362, 2006
International Conference on Information and Automation, 599-604, [36] Z. Gao, T. G. Habetler, R. G. Harley, and R. S. Colby, A sensorless
2010. rotor temperature estimator for induction machines based on a
[27] Mohamed Rashed, Peter F. A. MacConnell, A. Fraser Stronach, current harmonic spectral estimation scheme, IEEE Trans. Ind.
Nonlinear adaptive state-feedback speed control of a voltage-fed Electron., vol. 55, no. 1, 407–416, 2008
induction motor with varying parameters. IEEE Trans. on Industry
Applications, Vol.42, No.3, 723-732, 2006.
[28] Amalendu P. Ranjan, Ames Gomes, Decoupled adaptive control of
glucose and dissolved oxygen for fed-batch methionine production

2012 24th Chinese Control and Decision Conference (CCDC) 1081

Authorized licensed use limited to: VIT University. Downloaded on February 25,2022 at 07:17:13 UTC from IEEE Xplore. Restrictions apply.