基于遗传优化的RBF-BP网络的实时故障检测

Real-time fault detection based on genetic optimization of RBF-BP network

针对单一神经网络对复杂模型难以实时做出准确预测和BP神经网络自身的缺陷,结合RBF神经网络可以逼近任意函数的特性,提出了基于遗传优化的混合神经网络模型(RBF-BP)。由RBF网络和BP网络并联作为一个神经网络(简称为RBF-BP)的隐层,利用该网络对被控对象进行逼近训练、实时故障检测,该算法同时具有RBF网络和BP网络的优点,适用于复杂非线性系统的故障检测。

Complex for a single neural network model is difficult to make accurate and timely forecasts and BP neural network to their own shortcomings,combine with the characteristics of RBF neural network can approach to any functions.This paper proposes genetic optimization of the hybrid neural network model,that the RBF network and BP neural network as a parallel network（referred to as RBF-BP） of the hidden layer,diagnose real-time fault for the controlled object using the neural network.The algorithm has advantages of RBF network and BP network,it can apply to fault diagnosis for complex nonlinear system.