基于改进飞蛾优化算法的Hammerstein系统辨识

Hammerstein system identification based on improved moth⁃flame optimization algorithm

研究了在重尾噪声影响下多输入多输出(MIMO)Hammerstein系统的辨识问题。考虑到传统辨识方法在重尾噪声干扰下可能会失效,结合RBF神经网络和飞蛾优化(MFO)算法的优势,提出一种新型的辨识方案。利用RBF神经网络拟合静态非线性模块,通过将辨识问题转化为优化问题对线性部分和非线性部分的参数同时进行更新。为了提升飞蛾优化算法的辨识性能,将高斯混合分布思想引入到飞蛾位置更新中,提出一种新型的高斯混合飞蛾优化(GMFO)算法。实验结果证明了所提方法的有效性。

The identification of MIMO(multiple⁃input multiple⁃output)Hammerstein system under the background of heavy⁃tailed noise is studied.In view of the fact that the traditional identification method may fail under the interference from heavy⁃tailed noise,a novel identification scheme is proposed in combination with the advantages of RBF(radial basis function)neural network and MFO(moth⁃flame optimization)algorithm.The RBF neural network is used to fit the static nonlinear module and update the parameters of the linear part and nonlinear part simultaneously by converting the difficulties of identification to the optimization.In order to improve the identification performance of MFO algorithm,the idea of Gaussian mixture distribution is introduced into the moth position update,and a novel version named Gaussian mixture moth⁃flame optimization(GMFO)algorithm is proposed.The simulation results have verified the effectiveness of the proposed method.