基于小波包—灰色神经网络的机电作动系统故障预测

Fault Prediction of Electromechanical Actuating System Based on Wavelet Packet-gray Neural Network

摘要:

针对健康状态难以预测的机电作动系统,提出一种基于小波包—灰色神经网络的机电作动系统故障预测方法;该方法利用小波包分解对机电作动系统的故障特征进行提取,利用灰色GM(1,1)模型对提取出的12种不同故障模式下的频带值进行预测,并作为神经网络的输入,采用BP三层结构,通过验证取隐含层数为10进行训练,预测结果相对误差只有-0.014 6,取得了较好的预测效果;仿真结果表明:小波包—灰色神经网络方法在机电作动系统故障预测中具有一定的有效性。

Abstract:

Aiming at the evaluate problem of electromechanical actuating system health status, a method based on wavelet packet-gray neural network is proposed for the fault prediction of electromechanical actuating system. This method made a fault feature extraction in electromechanical actuating system applying wavelet packet decomposition, used the grey model of GM(1,1)to predict the twelve frequency band values under the different fault modes, then as the input of the neural network, adopted BP three layer structure, took hidden layer of 10 for training, relative error of predicted results is just -0.014 6, achieved good prediction effect. Finally, according to analysis of the fault of electromechanical actuating system, the correctness and effectiveness of this method are validated by the result of practical fault diagnosis examples.