论文部分内容阅读
提出应用小波神经网络实现非线性系统模型的在线建立及自校正方法。首先提出了小波网络节点库的构成方法和一种小波网络模型结构确定和权值估计方法。在此基础上运用限定记忆最小二乘法,设计小波网络自学习建模和在线校正的算法。该算法能根据系统输入输出数据自动地建立小波网络模型,并使得在线校正得到的小波神经网络在某种准则下是最优的。
A method of online establishment and self-tuning of nonlinear system model based on wavelet neural network is proposed. First of all, the method of constructing wavelet network node library and the method of structure determination and weight estimation of wavelet network model are proposed. On the basis of this, a finite memory least squares method is used to design the algorithm of wavelet network self-learning modeling and online calibration. The algorithm can automatically build a wavelet network model based on the input and output data of the system, and make the wavelet neural network corrected online is optimal under certain criteria.