Research On Genetic Algorithm And Its Application In Thermal Process Identification And Control

With the increasingly rise of power demand and rapid advancement of science and technology, modern power industries are heading for large-scale, high parameters and full automation. They bring forward higher demand on control quality of control system. It is based on mastering the characteristic of controlled object that the optimal control parameters are confirmed and the suitable system is designed. The traditional methods are far insufficient for ensuring rationality and control quality of the complex system after modeling. Accordingly, it has important meaning to increase control quality of power units, optimize unit performance and improve the economy of unit, that the modeling methods and PID controller parameter optimization are researched based on advanced optimization algorithm.The thesis is made up of three parts. Genetic algorithm is researched in the first part. Aiming at the problem that the simple genetic algorithm is difficult to deal with premature and local convergence, an improved real-coded adaptive algorithm is proposed. The validity of new algorithm is tested. The experimental results show that the improved method has better ability to converge to the global optimum than the simple genetic algorithm. Second part researches a thermal process identification method based on genetic algorithm. Above all, the identification principle based on genetic algorithm is introduced. As a general method, the validity of algorithm is tested when testing signal is complex or testing signal includes noise. Because the identification method based on step response works according to overlapping degree of response curve, and one system maybe have infinite corresponding model satisfying overlapping condition in theory, the probability and reliability of this identification method is analyzed and a positive conclusion is drawn. According to the characteristic of thermal process step response, the transferring functions of thermal process are classified and synthesized. Thus the bugs can be avoided that are met while adopting general transfer functions. The simulating result shows the validity. Third part researches a tuning method of PID parameters based on genetic algorithm. In the first place, the tuning methods of the PID controller parameters are overviewed, then a new tuning method of PID parameters is put forward. The method offers a linear quadratic function including the adjusted variables and the adjusting variables and the function is optimized directly by genetic algorithm. It avoids evaluating the Riccati equation. So this method has broader general sense. The simulating result indicates that this method is valid and preferably robust.In the thesis the principle of genetic algorithm is introduced, all kinds of operators are analyzed qualitatively, the strategies of the traditional genetic algorithm are improved and the better global optimization ability is gained. The improved algorithm is applied in thermal process identification and optimization of PID parameters. The methods have preferably general sense and precision. The simulating result indicates the algorithms are valid.