非线性系统模糊神经网络直接自适应控制

提出一种基于模糊神经网络的自适应控制方法。由模糊神经网络构成非线性预测器，利用使预测输出等于参考输出，生成实时控制信号。对自适应算法进行了理论分析，结合实例进行了仿真。

Direct FuzzyNeuralNetworkBased Adaptive Control for Nonlinear Systems

In this paper a nonlinear adaptive control algorithm based on fuzzy neural network is proposed. Nonlinear plant may be described by eqs.(1) and (2). For this nonlinear plant, a nonlinear predictor, eq.(3), is proposed by us. The predictor can be expressed by the fuzzy neural network. The output from the predictor is future prediction of the plant output. The current control signal is produced by letting the prediction output of the predictor equal the desired output of the plant (eq.5). A learning algorithm for the network parameters is derived (eq.6). Its convergence and properties are proved (lemma 2). The learning algorithm (eq.6), the predictor (eq.4), and the controller (eq.5) taken together is the direct adaptive control algorithm. Its convergence is also proved. The algorithm can deal with the nonlinear plant control without its mathematical model. Figs.1 and 2 show that the algorithm can follow the desired output signals well and converge quickly.