Genetic Algorithm (GA) Based Optimal Feedback Control Weighting Matrices Computation -ENG

Abstract

Linear Quadratic Regulator (LQR) is one of the most interesting control techniques adopted as a control strategy in state feedback. These types of techniques achieve good results but suffer from the problem of trial and error involved in the computation of weight matrices. The trial and error technique leads to hard tuning of the LQR controller parameters. This of course will lead to difficulty in reaching the optimal system performance. The paper attempts to solve the above difficulty via the selection of the LQR weight matrices using Genetic Algorithm GA. This proposed solution will avoid the trail and error involved in the state feedback technique. The proposed solution has been adopted in the design of position controller of a robot arm and the results of computer simulation have shown that the proposed solution fulfill specifications, for minimum overshoot , settling and rising times.

Aircraft pitch control design using LQG controller based on genetic algorithm

“…The genetic tuning method starts with the process of selecting real numbers assortment chromosomes, which serve as an initial population. During an optimization process, populations of the chromosomes are iteratively manipulated by GA using reproduction, crossover and mutation operations, then each solution can be validated using a fitness function [29], [30]. Flowchart for optimizing LQG controller based on genetic algorithm is shown in Figure 5.…”

Optimal tuning linear quadratic regulator for gas turbine by genetic algorithm using integral time absolute error

“…The matrix K is giving by (6) The symmetric definite matrix P is the solution of the algebraic Riccati equation given by (7) The closed-loop system which has the optimal Eigen values is given by (8) The block diagram of LQR controller of the gas turbineis shown in the…”

“…Therefore, such optimization technique can be used to design LQR for MIMO systems in a more systematic way. The GA was used to determine the weight matrices for single-input single-output systems [7]. Other reserchers used Ant Colony Optimization technique to determine the tuning parameters of the LQR [8,9].…”