Identification Of Nonmonotonic Nonlinear System Based On Genetic Programming

There are many nonlinear systems with their characters slow time varying in control domain. Those with intrinsical nonlinear portions are even much more hard to control successfully. System identification is the base of control, however, nowadays there are few satisfying traditional methods that are able to be applied to all kinds of nonlinear systems universally. In view to that, with the aid of a newly developed evolutionary computation technology, namely, genetic programming, it is significative to study the identification of certain nonmonotonic nonlinear systems based on genetic programming.The following aspects are principally studied in the thesis:The information of several respects on genetic programming is learned in depth relatively and broadly. With this understanding, generalization is made. First, the basic theory of genetic programming is illuminated here and its attributes are summarized. Second, certain phenomena in the run of GP system and the correlative theories are described. Then, the tendencies in the field are presented.The trials of some nonmonotonic nonlinear system identification using genetic programming are made under the assumption that neither the structure nor the parameters are known. First, it is carried out under the condition that the parameters are constant. Second, the situation that certain parameters are changed during the run of the system is tried. The observation results show that the approximate forms of the nonlinear part could be acquired under both the two conditions.Finally, some of the ideas of the further work are depicted.