论文部分内容阅读
本文首先将自组织神经网络算法向一般化情形引伸,接着把自组织过程应用到一般非线性系统的动态过程分类,使得整个非线性系统能够按照输入输出样本空间的概率密度自组织,成为许多具有不同分类核心和感受野的线性子空间逼近。在此基础上,我们采用通用最小二乘算法,以子空间的非线性问题线性化误差作为依据,并进一步运用自组织神经网络的合作与竞争思想,最终得到一般情形的非线性系统的最小二乘辨识。仿真结果表明了本方法的可行性与优越性。
In this paper, we first extend the self-organizing neural network algorithm to the generalized case and then apply the self-organizing process to the dynamic process classification of the general nonlinear system, which makes the whole non-linear system self-organized according to the probability density of input and output sample space, Approximation of Linear Subspaces with Different Categorical Cores and Feelings. On this basis, we use general least squares algorithm, based on the linearization error of subspace nonlinearity, and further use the idea of cooperation and competition of self-organizing neural network to finally obtain the minimum Take identification. The simulation results show the feasibility and superiority of this method.