基于Spiking的RBF神经网络故障诊断算法

神经网络是一种不依赖模型的控制方法,其自身并不需要给定预先需要的有关先验知识和判断函数,因此能对变化的环境（包括扰动和噪声信号等等）具有良好的自适应性.RBF神经网络是具有单隐层的三层前馈网络,由输入到输出的映射是非线性的,而隐含层空间到输出空间的映射是线性的.其优点在于收敛速度快,具有唯一最佳逼近的特性,且不会陷入局部最小的问题.Spiking神经网络采用时间编码的方式来进行数据处理,更接近于实际生物神经系统.基于Spiking的RBF神经网络在预测精度和误差控制上有着显著的效果.

Fault Diagnosis of RBF Neural Network Based on Spiking

Neural network is one of the independent control methods,which need not the given priori knowledge and diagnosis function.Neural network has a good adaptability to the changing environment(including disturbance and noise signal,etc).RBF neural network is a three layers feedforward network with one single hidden layer.The mapping from the input to the hidden layer is nonlinear and the mapping from the hidden layer to the output is linear.RBF neural network has a quick convergence rate,the uniqueness optimal approximation and will not fall into local minimum.Spiking neural network adopts time encoding to process data,which is more closed to the real biology nervous system.RBF neural network based on Spiking coding has a remarkable effect in forecast accuracy and error control.