基于小波神经网络的船舶电气故障诊断

Ship electrical fault diagnosis based on wavelet neural network

为保证船舶安全航行,需实时掌握电气系统运行状态,设计基于小波神经网络的船舶电气故障诊断模型。将小波分析方法引入神经网络模型中,采用小波函数替换网络模型隐含层的Sigmoid函数,设计小波神经网络模型;通过小波自适应软阈值降噪处理信号中的噪声,获取包含船舶电气系统运行特征信息的降噪后信号分量;改进BP神经网络依据该分量实现船舶电气故障分类诊断。测试结果显示:该方法的降噪效果良好,能量比在0.15以下;标准差结果在0.922以上;能够精准完成操作机构脱扣卡滞、电路过热以及绝缘体受潮3种故障诊断。

In order to ensure the safe navigation of ships,it is necessary to grasp the operating status of the electrical system in real time,in order to study the ship electrical fault diagnosis model based on wavelet neural network.This model introduces wavelet analysis method into the neural network model,replaces the sigmoid function of the hidden layer of the network model with wavelet function,and designs a wavelet neural network model;This model uses wavelet adaptive soft threshold denoising to process the noise in the signal and obtain the denoised signal components containing the operational characteristics of the ship electrical system;Improve the BP neural network to achieve classification and diagnosis of ship electrical faults based on this component.The test results show that the noise reduction effect of this method is good,with an energy ratio below 0.15.The standard deviation result is above 0.922.Capable of accurately diagnosing three types of faults:tripping and jamming of the operating mechanism,overheating of the circuit,and dampness of the insulation.