遗传算法在系统识中的应用.pdfVIP

摘 要 遗传算法是一种模拟自然界生物进化的搜索算法，由于它的简单易行、鲁 棒性强、尤其是其不需要专门的领域知识而仅用适应度函数作评价来指导搜索 过程，从而使它的应用极为广泛。 现代控制工程中的系统多表现为非线性、 时变和不确定性，要实现有效控制，必须采用模型辨识的方法来获取对象的近 似模型，并加以控制。 系统辨识问题可以归纳为系统结构辨识和系统参数辨识问题。本文通过遗 传算法成功的辨识了一类线性系统和线性较强的非线性系统问题。对于那些非 线性结构比较复杂的系统，我们采用了遗传规划。遗传规划是遗传算法的扩展， 它使用变长层次结构，且所使用的个体是长度和大小可变的程序树，它把系统 中的复杂分层表示出来。这样遗传算法和遗传规划在系统辨识问题中得到了成 功的应用，即，解决了系统结构和参数的辨识问题。 关键词 ：遗传算法，系统辨识，优化, 个体树, 遗传规划 I Abstract Genetic algorithm is a searching algorithm that simulates biology evolvement in nature.Because of its briefness ，strong robustness ，especiallyit need not expert domain knowledge but fitness function as the value guidesearching process ，its applications are very abroad.Based on variousMany systems in modern control project are shown as nonlinear, time-vari antand uncertain. To achieve effective control, we should use model identification to obtain the approximate model to the object. Genetic algorithm is a kind of search algorithms that can be used in model identification and optimality calculation in complex system. Syetem identification problem can conclude system configuration identification and parameter identification.This thesis use genetic algorithm successful identification linearily syetem and nonlinearily syst em of relatively linearity.But for nonlinearily configuration relatively complexity system, We use genetic programming. Genetic programming is genetic algorithm expand, It use changing hiberarchy, but that use individual is length size alterable programe tre