基于快速算法的模糊神经网络自适应控制

针对过程控制中被控对象常具有非线性、不确定性及参数时变等复杂因素，而难以建立精确的数学模型的情况，提出了一种基于快速学习算法的模糊神经网络自适应预测控制方案。该方案用神经网络作辨识器，模糊神经网络作控制器来实现非线性系统的自适应预测控制。为了克服传统的梯度下降法收敛速度慢、容易陷入局部极小值的缺点，该方案采用递推最小二乘法训练模糊神经网络。仿真结果表明，该方案可以实现模糊控制和神经网络的优势互补，对不确定非线性系统具有很好的控制效果。

Fuzzy Neural Network Adaptive Control Based on Fast Parameters Learning Algorithm

To the problem that the controlled object is often accompanied by some complex factors such as non-linearity,uncertainty,time-varying parameters,and the difficulty of building an accurate mathematical model,a fuzzy neural network adaptive predictive control scheme based on fast parameter learning algorithm is proposed.A neural network identifier and a fuzzy neural network controller are applied to realize the adaptive control of nonlinear systems.Instead of gradient descent method,recursive least-square algorithm is used to train the fuzzy neural network to overcome the defect of the low convergence rate and local minimal.Simulation results show that the fuzzy control and neural network can take advantage of each other to possess a good performance in the uncertain nonlinear system.