基于概率神经网络的高压加热器故障诊断及仿真

Fault Diagnosis and Simulation for High-pressure Heaters Based on Probabilistic Neural Network

针对汽轮发电机组高压加热器发生的故障问题,提出了一种基于概率神经网络(PNN)的故障诊断方法。通过对该设备故障的定性分析,对比研究了PNN与误差反向传播神经网络(BPNN)分类模型对故障的分类效果。仿真表明,对高压加热器的故障分类中,在分类速度、精确度等方面,PNN模型均优于BPNN分类模型。

In order to solve the problem encountered by a turbine-generator＇s high-pressure heater,a fault diagnosis method based on PNN（probabilistic neural network） was proposed.By qualitatively analyzing the equipment fault,a comparison was made between classification results obtained respectively with PNN and BPNN（back-propagation neural network） models.Simulation results show that for fault diagnosis of high-pressure heaters,PNN model is better than BPNN in the aspect of classification speed and precision.