二型模糊小波脑情感学习网络在非线性系统辨识中的研究  

Type-ⅡFuzzy Wavelet Brain Emotional Learning Neural Network for Nonlinear System Identification

作　　者：桑景福 陈向坚[1] 王平心 SANG Jing-fu;CHEN Xiang-jian;WANG Ping-xin(School of Computer,Jiangsu University of Science and Technology,Zhenjiang 212003,China;School of Science,Jiangsu University of Science and Technology,Zhenjiang 212003,China)

机构地区：[1]江苏科技大学计算机学院,江苏镇江212003 [2]江苏科技大学理学院,江苏镇江212003

出　　处：《模糊系统与数学》2022年第1期18-30,共13页Fuzzy Systems and Mathematics

摘　　要：针对现有模糊神经网络在辨识具有时变的非线性系统存在辨识精度不高,收敛速度较慢等缺点,提出了一种二型小波模糊脑情感学习网络(T2FWBELN)模型,它结合了模糊逻辑和脑情感学习网络的优点,并在网络结构中使用了小波函数。与其他算法相比,该算法在非线性系统辨识中有着更高的逼近能力。同时,采用模糊C均值算法生成模糊规则,并使用梯度下降法对T2FWBELN的各种参数进行在线调整,降低了参数调整时间。为了进一步验证该模型的有效性和优越性,仿真了两个不确定非线性系统辨识的例子,一个是Mackey-Glass时间序列预测,一个是带有噪声的动态系统辨识。测试结果表明,所提出的模型在处理非线性系统辨识中拥有更高的精度。To address the shortcomings of existing fuzzy neural networks in identification complex time-varying nonlinear systems with low recognition accuracy and slow convergence,a type-Ⅱwavelet fuzzy brain emotional learning network(T2 FWBELN)is proposed,which combines the advantages of fuzzy logic and brain emotional neural learning networks,uses wavelet functions in the network structure,and is applied to nonlinear system identification.Compared with other algorithms,this algorithm has higher approximation ability.Meanwhile,in order to reduce the parameter adjustment time,the fuzzy C-mean algorithm is used to generate fuzzy rules,and the gradient descent method is used to adjust the various parameters of T2 FWBELN.To further validate the effectiveness and superiority of the model,two examples of nonlinear system identification are simulated,one for Mackey-Glass time series prediction and one for dynamic system identification with noise.The test results show that the proposed model possesses higher accuracy in dealing with nonlinear system identification.

关 键 词：非线性系统辨识 脑情感学习网络 二型模糊集 

分 类 号：O159[理学—数学]