基于多传感器振动信号的机械设备故障预测

Mechanical Equipment Fault Prediction Based on Multi-sensor Vibration Signals

目前相关研究主要集中在故障诊断和剩余使用寿命预测上,无法提前提供旋转机械设备的具体健康状况和故障类型等信息。本文将CNN、LSTM和支持矢量分类(SVC)相结合,建立基于多个传感器振动信号的旋转机械设备三阶段故障预测模型,实现故障分类和故障类型的同时识别。实验结果表明,随着迭代次数增加,故障预测准确率呈非线性递增趋势,但当迭代次数大于600时,故障预测准确率增加幅度缓慢。迭代次数600次时,三阶段故障预测模型的故障预测准确率较优。且机械设备轴承故障越严重,故障预测准确率越高。与其他三种模型的故障诊断方法相比,基于多传感器振动信号的三阶段故障预测模型能在一定程度上提高故障诊断率,并具有识别新故障的特点。

Current research mainly focuses on fault diagnosis and remaining useful life prediction,which cannot provide specific information about the health condition and fault types of rotating machinery equipment in advance.This article combines CNN,LSTM,and Support Vector Classification(SVC)to establish a three-stage fault prediction model for rotating machinery equipment based on multiple sensor vibration signals,achieving fault classification and simultaneous identification of fault types.Experimental results show that as the number of iterations increases,the fault prediction accuracy exhibits a nonlinear increasing trend.However,when the number of iterations exceeds 600,the increase in fault prediction accuracy becomes slower.When the number of iterations is 600,the fault prediction accuracy of the three-stage fault prediction model is optimal.Moreover,the more severe the bearing fault of the mechanical equipment is,the higher the fault prediction accuracy will be.Compared with the fault diagnosis methods of the other three models,the three-stage fault prediction model based on multi-sensor vibration signals can improve the fault diagnosis rate to a certain extent and has the characteristic of identifying new faults.