并行模糊系统的预测和辨识收敛性

自1965年L.A.Zadeh提出模糊集理论以来,模糊理论已经在非线性系统辨识、函数逼近、模式识别、机器学习等领域得到了极广泛的应用。由于人的推理在本质上是模糊的,因此使用模糊理论处理实际问题时更符合人的处理过程,这就极大地提高了人类解决问题的能力。在模糊理论的应用当中,非线性系统辨识是最重要的方向之一。一般说来,模糊系统可以以两种方式用于非线性系统辨识:串并行方式和并行方式(见图1,图2)。其中TDL表示时间延迟逻辑,RS表示待辨识的实际系统,FS表示模糊系统。图1称为串并行方式,图2称为并行方式。在大部分实际应用当中使用的是串并行方式,

Prediction and Identification Algorithm Convergence of Parallel Fuzzy Systems

Fuzzy systems can be used to identify nonlinear dynamic systems in two modes. One is series-parallel modeand the other is parallel mode. The prediction and identification algorithm convergence of nonlinear dynamic system i-dentification using parallel fuzzy systems is discussed in this paper. It is proved that as long as the parameters of par-allel fuzzy systems meet some prerequisites, the parallel prediction procedure converges and the parallel identificationalgorithm locally converges. This conclusion has instructive significance for parallel fuzzy systems' application.