基于小波网络的非稳态故障诊断方法
Unsteady Fault Diagnosis Based on Wavelet-Networks
-
摘要: 针对现有BP网络在故障诊断中存在的缺陷,提出了利用小波函数构造故障诊断神经元网络,利用变尺度算法进行网络训练,加快了网络的训练速度。并交结果运用到一个CSTR非稳态故障诊断的过程中,仿真实例表明,小波网络较BP网络具有更好的故障诊断能力。Abstract: In order to improve the performance of fault diagnosis systems based on BP networks, a new method based on wavelet networks is presented which incorporated an algorithm of varying scal for training networks so as to accelerate the training process. Simulation research on its application to unsteady fault diagnosis of a CSTR process has proved that it performs more effectively than that based on BP networks.