元胞萤火虫算法在Wiener模型辨识中的应用

Improved glowworm swarm optimization based on cellular automata and its application in Wiener model identification

基本萤火虫算法存在陷入局部最优、后期收敛慢等固有缺点,为此将元胞自动机机理融入自适应步长萤火虫算法,即将邻域规则和演化规则融合在萤火虫算法中。通过其邻域模型选择邻域集合,在其邻域结构内通过一种融合生命游戏与优胜劣汰的演化规则进行迭代寻优。对4种典型的测试函数进行实验,实验结果表明,该算法能跳出局部最优,有较强的收敛速度和精度,可应用于非线性系统中Wiener模型的参数辨识。因Wiener模型含有非线性部分,导致不易辨识,采用改进的元胞萤火虫算法将该参数辨识问题转变为优化函数问题,利用元胞萤火虫算法进行函数寻优。数值仿真验证了改进算法能够有效地进行非线性系统参数辨识。

Basic glowworm swarm optimization possesses slow convergence speed,poor local search ability and easiness to fall in local peak.To overcome these problems,an adaptive step algorithm integrating the mechanism of cellular automata was proposed,namely applying the evolutionary rule and domain rule to glowworm swarm optimization.Neighborhood was selected via domain model and ierative refinement was proceeded by means of evolutionary rule coalesced with game of life and survival of the fittest within the domain structure.Typical test functions were simulated and tested,the results of which reveal the proposed algorithm has better global searching ability,convergence speed and precision.Because Wiener model of the nonlinear system possesses nonlinear drag casing to identify difficultly,the improved cellular glowworm swarm optimization was put forward for model parameter identification.Parameter identification problem was converted to function optimization problem and solved using cellular glowworm swarm optimization.It is verified to be effective and feasible in identification problem with numerical simulation.