混合编码免疫算法在非线性系统辨识中的应用

Application of Combinated-encoding Immune Algorithms for Nonlinear System Identification

基于免疫算法优越的全局搜索性能与GP算法简洁的结构树编码方法,提出了一种混合编码免疫辨识算法,通过对模型结构与参数分别编码及免疫操作,同时实现了非线性模型的结构与参数辨识,实现了全局寻优,辨识的模型结构简单、易于理解。仿真验证了本算法的有效性及较强的非线性逼近能力。

Based on global search performance of traditional immune algorithm and simple hierarchical classification tree of GP algorithm, the Combinated-Encoding Immune algorithm was proposed. The nonlinear model＇s global optimized structure and parameters were both achieved through encoding and immunizing operations. The simulation result shows that achieved model structure is simple and easy to comprehend. The validity and the nonlinear approach ability of this algorithm are also proved.