基于自适应遗传随机共振的滚动轴承微弱故障诊断

Rolling bearings weak fault diagnosis based on adaptive genetic stochastic resonance

随机共振(SR)能够利用噪声能量增强微弱信号,有效降低了噪声信号对特征提取的影响,针对SR方法参数选择时缺少交互以及提取特征诊断效果缺乏验证的不足,提出自适应遗传随机共振(AGSR)的滚动轴承微弱故障诊断方法。AGSR方法利用遗传算法(GA)寻找随机共振的最优系统参数,在考虑参数间交互作用的同时对其进一步优化,有效提高了轴承微弱故障特征的提取效果,随后将AGSR方法提取的特征信号输入堆叠自动编码器(SAE),通过反向传播算法多次迭代优化整个SAE网络,最终实现故障诊断。滚动轴承实测数据的检验结果表明,该方法可有效实现滚动轴承早期微弱故障检测。

Stochastic resonance(SR)can be used to enhance weak signals by means of noise energy,which effectively re?duces the influence of noise signals on feature extraction.In allusion to the lack of interaction in parameter selection of SR meth?od and the lack of validation in diagnosis effect of feature extraction,a rolling bearing weak fault diagnosis method based on adaptive genetic stochastic resonance(AGSR)is proposed to improve the recognition rate of weak faults.The AGSR method is used to find the optimal system parameters of SR by means of genetic algorithm(GA),the parameters are further optimized while considering the interaction between them,and the extraction effect of bearing weak fault features is effectively improved.Then,the feature signal extracted by AGSR method is input into stacked autoencoder(SAE),the total SAE network is opti?mized by multiple iterations by means of back?propagation algorithm,and the fault diagnosis is achieved ultimately.The test re?sults of the measured data of rolling bearing show that this method can effectively realize the early weak fault detection of the rolling bearings.