基于数据增强的滚动轴承智能故障诊断方法

目的 针对包装机械设备中滚动轴承应用场景多且有效故障数据难采集而导致的智能诊断方法诊断准确率较低的问题,提出一种基于数据增强的滚动轴承智能诊断方法.方法 首先根据轴承振动信号的故障特征,提出一种数据增强方法,有效扩充训练数据样本多样性.然后采用卷积神经网络对原始样本和增强样本进行故障诊断训练,从而大幅度提高诊断模型的诊断性能.为了验证所提方法的有效性,建立滚动轴承故障试验台并采集轴承故障数据.结果 实验结果表明,在标签训练样本不充足的情况下,提出的方法与不使用数据增强方法相比,模型在诊断准确率方面取得了较大的提高,能够准确地识别各类轴承故障.结论 该方法实现了准确地对稀缺标记样本下滚动轴承故障的诊断,为保证包装机械滚动轴承故障诊断的诊断精度提供了可靠的方法.

Intelligent Fault Diagnosis Method of Rolling Bearing Based on Data Enhancement

In order to solve the deficiency of accurate analytical dynamic response model for dynamic parameter identification of packaging machinery, this paper presents a new method of parameter identification for packaging pushing mechanism using Lie symmetry analytic model. Firstly, the nonlinear dynamic equation of packaging pushing crank slider mechanism is established; secondly, the analytic mathematical model of crank and slider motion displacements are obtained using Lie symmetric integration method, and then the iterative formulas of parameter identification are obtained by combining Lie symmetry analytical model with recursive least square algorithm. Finally, an example shows that the analytical model values are consistent with the experimental values, and the accuracy of parameter identification is higher, thus the parameters identification using Lie symmetry analytical method can provide important reference for model control and path planning of packaging pushing mechanism.