Multivariable Control Theory:

Classical Robust Control

Methods
 General Robust Control Problem Formulation
The concepts of gain margin and phase margin are relevant only when there is either pure gain uncertainty or phase uncertainty in a system. However, in practical
systems, it is common for both uncertainties to coexist simultaneously. In such cases, simply having good gain and phase margins does not guarantee the robustness
of the system.
Having small gain or phase margins indicates that the system is not robust, meaning it is sensitive to variations in gain or phase. On the other hand, having good gain
and phase margins does not necessarily imply that the system is robust. Other factors beyond gain and phase margins must be considered to ensure the overall
robustness of the system.

In numerous control scenarios, the primary objective is to minimize the Hinf or H2 norm of a transfer function (t.f.) by utilizing a controller, denoted as K. To tackle this
problem effectively, it is common to formulate it with the assistance of a generalized plant, represented as P. The generalized plant encompasses all the known
transfer functions, while the generalized controller, K, is the controller that needs to be designed. It's important to note that P not only represents the physical system
but also encapsulates the problem concept. The ultimate goal of the control process is to minimize a specific norm of the transfer function between the input signal
(w) and the output signal (z).

𝑤 𝑧

P
𝑣
𝑢

K
 Example: Classical Disturbance Rejection Problem
The aim of the control is to find a controller K to minimize/reduce/attenuate the effect of the disturbance to the output.
The problem is to minimize the H2 and Hinf norm of the closed loop system transfer function between z and w.

+ 𝑑 𝑦
+ u
K G
r e +
-

u 𝑦 𝑣
G -I

K
Example:
𝑑
𝑒𝑚 𝑢 𝑦
𝑟
K G
-
𝑦𝑚
𝑛

It is required to minimize the Hinf norm of the transfer function between

 − 𝑧
𝑟
𝑛
𝑑
𝑢 𝑦 v
G
−

𝑃
K
For the control problem to be meaningful, we may need to use weight
transfer functions on w, d, or z.
𝑤 𝑧ǁ 𝑧
𝑤
ෝ Wz
Ww
P’
𝑣
𝑢

Usually it is assumed that Ww and Wz are stable and minimum phase

transfer function matrices.
 Robust Control Toolbox (Matlab)
1. The robust control toolbox is a comprehensive set of functions and tools that assist in the examination and
creation of Multiple-Input Multiple-Output (MIMO) control systems, especially those with uncertain elements.
These tools enable analysis and design in the presence of uncertain parameters and dynamics.
2. The toolbox allows you to construct models of Linear Time-Invariant (LTI) systems that encompass uncertain
parameters and dynamics. This capability is essential for handling system uncertainties that may arise due to
varying operating conditions or imprecise knowledge about the system.
3. By utilizing the toolbox, you gain access to tools that enable the evaluation of stability margins and worst-case
performance of MIMO systems. This analysis is crucial for ensuring that the control system remains stable and
performs satisfactorily even under unfavorable scenarios.
4. The toolbox includes model reduction tools that simplify and reduce the complexity of intricate models. These
tools minimize both additive and multiplicative error bounds, making it easier to work with large-scale models
without sacrificing accuracy.
5. The toolbox provides advanced robust control methods such as H-infinity (H∞), H2, linear matrix inequalities
(LMI), and µ-synthesis robust control. These methods are based on mathematical optimization techniques and
allow for the design of controllers that are more robust against uncertainties and disturbances.
6. Additionally, the toolbox offers the capability to shape the frequency responses of MIMO systems and design
controllers that can tolerate uncertainties. This means you can tailor the system's behavior to desired frequency
ranges and ensure the controllers remain effective even in the presence of uncertainties.
• In control engineering, modeling uncertainty refers to situations where the
precise values of system gains or other parameters are not known, or they can
vary within a certain range. Real parameter uncertainties can include
uncertainties in pole and zero locations as well as uncertain gains.
• The Robust Control Toolbox provides the capability to create uncertain Linear
Time-Invariant (LTI) models. These models are specifically designed as objects that
can be used in robust control applications. By incorporating uncertainty into the
models, control engineers can effectively account for the variations and lack of
precise knowledge regarding system parameters, ensuring robust performance in
the face of such uncertainties.
Sources of Uncertainty
Uncertainty in an system can occur in various forms and from various sources.

Multiplicative Uncertainty
Additive Uncertainty

Feedback Uncertainty
In the Robust Control Toolbox, several functions are available to define and create various types of uncertain parameters and systems:
1.UComplex(): This function is used to define complex uncertain parameters. It allows you to specify uncertainties in the real and
imaginary parts of complex-valued parameters. These uncertainties can be expressed in terms of uncertain gain, uncertain pole/zero
locations, or other forms of uncertainty.

2.UComplexM(): This function is specifically designed for the creation of complex-valued uncertain matrices. It enables you to define
uncertainties in the elements of matrices, which can be useful in representing uncertain system dynamics or uncertain controller
parameters.

3.UDyn(): By using this function, you can create an unstructured uncertain dynamic system class. The input/output dimension of the
system is specified by the iosize parameter. This function allows you to represent systems with uncertain dynamics, where the exact
structure of the uncertainty is not explicitly known or modeled.

4.ULTIDyn(): This function is used to create an uncertain linear time-invariant (LTI) object, where only bounds on the frequency
response are known. It is particularly helpful when you have limited information about the exact frequency response characteristics of
the uncertain system. The bounds on the frequency response can be defined to represent the uncertainty.

5.UReal(): This function is employed to define real uncertain parameters. It is used in various analysis and design functions within the
Robust Control Toolbox. The uncertainties in the parameters can be expressed in terms of ranges, intervals, or other forms of
uncertainty, allowing for robust analysis and design of control systems.

These functions play a crucial role in representing and handling uncertainties in control systems, enabling robust control analysis and
design by incorporating uncertainties into the models and algorithms.
 Mu-Synthesis Toolbox
The Mu-Synthesis toolbox, which was created by renowned experts in Robust Control named Balas, Doyle, Glover,
Packard, and Smith, offers various features that enable robust controller design in the frequency domain.
These individuals are highly respected in the field of control engineering worldwide.
We will focus on highlighting a few noteworthy aspects of the toolbox, providing a glimpse into its capabilities.

Basic Definitions:
Given A, B, C, and D matrices in state-space representation a system matrix can be defined using:

Sys=pck(A,B,C,D);

It is also possible to unpack a system into state space matrices

[A,B,C,D]=unpck(sys);
 System Interconnections
To construct a generalized system by connecting multiple subsystems, the sysic command can be employed.
However, before utilizing the sysic command, certain prerequisites need to be fulfilled. Firstly, the names of the subsystems must be defined, indicating how each
subsystem is identified.
Additionally, the names of the inputs and outputs of the overall system need to be specified, denoting the signals that enter and exit the system.
Finally, the inputs to each subsystem should also be defined, outlining the signals that are provided as inputs to individual subsystems.
By fulfilling these requirements, the sysic command can be effectively utilized to interconnect the subsystems and form a comprehensive system described in a
generalized form.
Optimal Controller Design
 Generalized plant has been formed:
system: 4 states 3 outputs 2 inputs

• nd2sys(), which enables the construction of a Single Input Single Output (SISO) system using a numerator and denominator representation. By providing the
appropriate numerator and denominator coefficients, the nd2sys() function allows for the creation of a mathematical representation of the system.
• Another useful function is conv(), which facilitates the multiplication of polynomials. This function is commonly used when manipulating transfer functions or
performing operations involving polynomial expressions in control systems.
• It is worth noting that the Mu-Synthesis toolbox considers a constant input Wu=1 as a system. While this may seem unconventional, it is a valid approach
within the context of the toolbox and can be utilized accordingly for system analysis and design.
 # of measurement
𝑇𝑜𝑙𝑒𝑟𝑎𝑛𝑐𝑒
Gamma range
# of control

[K,CL,gamma] = hinfsyn(P,nmeas,ncont,gamTry) calculates a controller for the target performance level gamTry. Specifying gamTry can be useful when the optimal controller performance is better
than you need for your application. In that case, a less-than-optimal controller can have smaller gains and be more numerically well-conditioned. If gamTry is not
achievable, hinfsyn returns [] for K and CL, and Inf for gamma.
example
[K,CL,gamma] = hinfsyn(P,nmeas,ncont,gamRange) searches the range gamRange for the best achievable performance. Specify the range with a vector of the form [gmin,gmax]. Limiting the
search range can speed up computation by reducing the number of iterations performed by hinfsyn to test different performance levels.
Robust Performance
 Test the system response for
Frequency response of the closed loop system several allowed perturbations
 References
• https://www.mathworks.com/help/robust/ref/dynamicsystem.hinfsyn.ht
ml
• https://www.mathworks.com/help/robust/controller-synthesis.html
• https://www.mathworks.com/videos/robust-control-part-5-h-infinity-and-
mu-synthesis-1586760454181.html
• Mu-Analysis and Synthesis Toolbox_For Use with MATLAB, Gary
J.Balas, John C.Doyle, Keith Glover, Andy Packard, Roy Smith
• Robust Control Toolbox_Getting Started Guide, User’s Guide
• Use of mu-Synthesis Toolbox. M. Turan Söylemez