基于小波网络的传感器故障检测

Fault Detection Based On Wavelet Network

针对常规BP网络存在的问题,在小波网络的基础之上将其结构进行改进,提出复合小波网络,代替BP网络用于传感器的故障检测。为避免由于传感器输入信号突变引起的传感器输出与观测器预测值产生较大偏差而导致的误诊断,提出将检测方法加以改进。仿真实验表明,复合小波网络较BP网络和未改进的小波网络收敛速度快、预测精度高,更适用于故障检测;采用了复合小波网络的改进的检测方法既提高了检测的快速性,又提高了可靠性。

Aimming at the problem of BP network, the multiple wavelet network is raised to replace BP network in the fault detection of transducer though improving the structure of wavelet network.In order to ,avoid the fault diagnosis On account of the difference between the output of transducer and observe owing the break of transducer , detection＇s means is improved.The result of experiments indicates that the multiple wavelet network has the quicker convergence speed and the higher prediction precision and is fitter than BPnetwork and unimproved wavelet-Network in fault detection.The improved fault detection using the multiple wavelet network enhances both speedability and securlty.