The Application Of GAs In Human-simulated Intelligent Control

Control on under-actuated systems has extensive application in the engineering technology fields, for example, space station, nuclear power station, space robot etc. Under-actuated systems are difficult to be controlled, and the main reason is that the dimensions of controlling inputs are less than one of generalizing coordinate. To resolve the controlling problems of under-actuated systems, researchers have developed all kinds of control theories and control methods for different tasks. But these methods are of strong limitation, that is that these methods are only fit to resolve some special controlling problems of under-actuated problems, and have not general meaning. Aiming at control of complex systems, Professor Li Zhushu and his research group applied their originally-developed Human-Simulated Intelligent Control theory based on Sensor-motor Intelligent Schema to the control of under-actuated systems and provided a feasible way resolving the control problems of under-actuated systems. One of main problems when applying human-simulated intelligent control is parameters optimization of controllers. With the increase of complexity of systems, the structure of human-simulated intelligent controller becomes also more and more complex, therefore there is much difficulty for some experts in the parameters optimization. Hereby, the paper developed a method resolving the parameters optimization of human-simulated intelligent controller based on evolution thought.In the research process, the following works have been completed:(1) Researching the relations of human-simulated intelligent control and GA;(2) Developing an improved genetic algorithm;(3) Providing a modeling method of complex systems;(4) Taking the parameters optimization of human-simulated intelligent controller for the swing-up and handstand of three-link acrobot as an example, and discuss a optimizing method of human-simulated intelligent controller's parameters.Practices demonstrate the effectiveness of proposed methods resolving the parameters optimization problems of human-simulated intelligent controller based on evolution thought.