摘要
系统辨识是控制系统设计的基础，对于非线性系统辨识，传统的辨识方法往往得不到全局最优解，为此，提出了基于遗传算法的非线性系统辨识方法．遗传算法在解空间中多点寻化、工作对象是参数编码集、不需要导数信息和其它辅助信息、用概率性规则指导搜索，因而具有很强的鲁律性和广泛的适应性，为非线性系统辨识的研究与应用开辟了新途径采用的遗传算法是在简单遗传算法的基础上加以改进的改过遗传算法采用适应值比例法与最优保留策略相结合的方法进行繁殖操作，同时又自适应地改变了交叉和变异概率．该方法成功地辨识了非线性静态、动态模型，仿真结果表明了该方法的有效性.
Identification of system is the foundation for design of control systems, global optimal results can not be obtained for identilication of nonlinear system by using traditional identilication methods. An identification method based on genetic algorithm (GA) has been proposed for non-linear systems. GA searches from a population in the result space, manipulates a coding of parameter sets, is blind to derivative and other auxiliary information, and uses randomized operations. All these charecteristics contribute to GA's good robustness and extensive application. Ge-netic algorithm has opened a new path for system identilication. Based on simple genetic algorithms(SGA), an im-proved genetic algorithm (IGA) is studied. During repeduction, roulette selection and optimal reservation method are adopted. Moreover, the probability of crossover and mutation are adjusted adaptively. IGA can be successfully uted to identify nonlinear static and drnamic molels. Simulations show the effectiveness of IGA.
出处
《哈尔滨工业大学学报》
EI
CAS
CSCD
北大核心
1999年第2期39-42,共4页
Journal of Harbin Institute of Technology
