基于BP神经网络的变速箱故障诊断方法研究

A Research of Gearbox Fault Diagnosis Based on BP Neural Network

摘要:

针对变速箱的工作时间不能真实反映实际健康状况的问题,通过提取变速箱的振动信号作为状态参数,建立了基于BP神经网络的变速箱故障诊断模型;该模型首先提取振动信号中对故障反映灵敏的成分作为特征值,获得BP神经网络的训练数据,并通过对比确定最优的隐含层节点数,确定BP神经网络的结构参数;模型训练结束后,以验证数据为例进行故障诊断研究,并对诊断结果进行评估;评估结果表明,该模型准确度高,具有较好的应用和推广价值。

Abstract:

The useful life of gearbox can not truly reflect its healthy condition. Aiming at this problem and choosing appropriate condition parameter which is the vibration data of the gearbox, a model of gearbox fault diagnosis based on BP neural network is established. This model needs to extract the Eigen value of vibration signal which is sensitive to fault firstly, and the training data for BP neural network can be got. The optimal number of hidden nodes is determined by comparing each other, then the structural parameters of BP neural network is determined. When the model training is over, the research on fault diagnosis is carried out using the test data and to evaluate the result of diagnosis. The application shows that the model runs accurately, and it is practical and worthy of using abroad.