Ivan Sekaj

Robust output feedback controller design: genetic algorithm approach

This paper proposes guaranteed cost design of robust output feedback controller for continuous linear parametric uncertain systems. New necessary and sufficient conditions for static output feedback stabilizability of linear continuous time systems underly the design procedure. Proposed algorithms are computationally simple and tightly connected with the Lyapunov stability theory and the LQR optimal state feedback design. The proposed approach allows for prescribing the structure of the output feedback gain matrix (including the decentralized one) by the designer. New design method proposed in this paper, exploit genetic algorithm to design robust controller with guaranteed cost for polytopic linear continuous time systems. Numerical example is given to illustrate the performance of the proposed robust controller.

Optimization of Robotic Arm Trajectory Using Genetic Algorithm

Abstract The paper presents a genetic algorithm - based design approach of the robotic arm trajectory control with the optimization of various criterions. The described methodology is based on the inverse kinematics problem and it additionally considers the minimization of the operating-time, and/or the minimization of energy consumption as well as the minimization of the sum of all rotation changes during the operation cycle. Each criterion evaluation includes the computationally demanding simulation of the arm movement. The proposed approach was verified and all the proposed criterions have been compared on the trajectory optimization of the industrial robot ABB IRB 6400FHD, which has six degrees of freedom.

Genetic Algorithm Based Controller Design

Abstract Controller design approach is described, which is based on genetic algorithms. The approach uses an optimisation procedure, where the cost function to be minimised consists of the closed-loop system simulation and a performance index evaluation. The designed system may be complex and practically without any limitations on its type, structure and size. The proposed approach is demonstrated on a robust controller design example

Selected population characteristics of fine-grained parallel genetic algorithms with re-initialization

A class of fine-grained parallel genetic algorithms (F-PGA) are analyzed and experimentally compared. Each node of the F-PGA represents a single individual. Selected topologies are proposed, which are using various parent selection and offspring selection methods. Also the influence of population re-initialization on the parallel genetic algorithm performance is analyzed and selected characteristics of evolutionary algorithm population are proposed. These characteristics represent such properties as relative number of modified genes and number of duplicate individuals in population. The results are demonstrated on examples with minimization of selected test functions.

ROBUST OUTPUT FEEDBACK CONTROLLER DESIGN VIA GENETIC ALGORITHM

Abstract This paper proposes guaranteed cost design of robust output feedback controller for continuous linear parametric uncertain systems. Proposed algorithms are computationally simple and tightly connected with the Lyapunov stability theory and the LQR optimal state feedback design. The proposed approach allows for prescribing the structure of the output feedback gain matrix (including the decentralized one) by the designer. New design method proposed in this paper, exploit genetic algorithm to design robust controller with guaranteed cost for polytopic linear continuous time systems. Numerical example is given to illustrate the performance of the proposed robust controller.

Design of Continuous-time Controllers using Cartesian Genetic Programming

Abstract An evolutionary computation - based design/optimisation approach using the Cartesian Genetic Programming is proposed for non-linear continuous-time process control. It is a simplification of a more general Genetic Programming – based design, which is powerful, but more computationally demanding. The approach is able to produce effective and non-intuitive controllers in the form of a network of interconnected elementary building blocks, which minimize the defined performance index. Each building block performs mathematic operations between its inputs, next it contains gain and an elementary dynamic part as integrator, derivative or unity gain. The proposed design method is demonstrated on water turbine control design, and the results are compared with the genetic algorithm-based PID controller design.

Optimization of power system stabilizer by genetic algorithm

Abstract Power system stabilizers (PSS) play an important role in damping of power system oscillations. An intensive research activity has been devoted to design of their structure and optimal setting of their parameters. In this paper the simultaneous optimization of multiloop PSS and automatic voltage regulator parameters (AVR) by means of genetic algorithm is proposed. Using an example of the 259 MVA turbogenerator excitation system in the nuclear power plant Mochovce (Slovak Republic) it is shown that the genetic algorithms are able to find the optimal parameters of excitation system so that the requirements on terminal voltage performances as well as on damping of active power oscillations are satisfied.

Fuzzy Logic Controllers and Their Replacement by a Non-Fuzzy Approximation Algorithm

Abstract As an alternative to the defuzzification a new computation method of fuzzy logic controller output is presented. In connection with this method an other procedure is proposed, which allowes to substitute also for fuzzification. It uses the same knowledge base in the form of a rule-table like the fuzzy logic controller, but without using the fuzzy logic. The proposed algorithm is simple and the results which where verified in simulation and in the real-time control as well are the same like with fuzzy logic controllers.

Parameter optimisation of Artificial Pancreas adaptive controller using genetic algorithm

We propose an optimisation method for parameter setting of a MRAC-based adaptive controller of blood glucose (glycemia) in the human body. The glycemia dynamics is a complex process, which can be controlled by external bio-cybernetic devices as Artificial Pancreas. The artificial pancreas consists of glycemia sensor, controller and insulin pump. Such systems are using complex algorithms with many parameters which have to be set by the designer and which have significant influence on the quality of glycemia control. Since conventional parameter design methods do not give satisfactory results a simulation-based optimisation method which is based on genetic algorithms was proposed. The proposed method gives good results and it can be used also for design of other complex bio-cybernetic systems.

Modeling of the Blood Pressure Regulation System in Rats Using Genetic Algorithms

Abstract Blood pressure (BP) is one of the principal vital signs. The regulation of normal BP is critical for maintaining the normal functioning of an organism. The structure of the BP regulation in vertebrates, including humans, is very complex and it is not yet fully understood. There are several BP regulation subsystems which interact together. In this study, we analysed and modelled the role of the most important BP regulation systems (sympathetic nervous system, renin-angiotensin-aldosterone system, L-Arginine/nitric oxide) as well as the role of reactive oxygen species, the imbalance of which is implied in the development of hypertension disease. Our aim is to design a dynamic model of the BP regulation in normotensive rats, which can help us better understand the complex process of BP regulation. The BP responses were measured in conscious rats after acute stress with or without the application of selected drugs, which can inhibit particular regulation subsystems. Linear dynamic models of particular regulation subsystems were identified using genetic algorithms. Each subsystem is a feedback loop, which contains both the system dynamics and the controller. We describe the simulation-based identification procedure of these particular models as well as of the entire BP regulation system.