Genetic Algorithms for Control System Design Applications

The paper describes genetic algorithms application for control system design and for dynamic system model identification. The approach is based on the search for the multivariable function global optimum with cost functions which consist of dynamic system simulation and integral performance criterion evaluation. The proposed methods are demonstrated on design examples of control structures under PID controllers for single input – single output (SISO) and multi input – multi output (MIMO) systems as well as on an example of a selftuning control structure for a 2x2 system. A real-time identification example is presented, too.