【摘要】 提取滚动轴承有效的故障特征参数是轴承故障诊断重要的组成部分,为改善核极限学习机(Kernel Extreme Learning Machine,KELM)高维数据特征选取的问题,提出一种结合稀疏自动编码器(Sparse Auto-Encoder,SAE)与KELM的方法。首先,提取振动信号的时域、频域和时频域特征构成高维特征向量;其次,采用多层SAE融合高维特征来消除特征的冗余性;最后,采用融合后的特征训练KELM,得到故障诊断模型。针对KELM对参数敏感的缺陷,采用萤火虫算法(IF)进行参数优化。为评估方法有效性,采用实验数据进行测试,并与传统KELM方法进行比较,结果显示该方法具有更好准确性和稳定性。更多还原

【Abstract】 Extracting effective rolling bearing fault feature parameters is an important part of the bearing fault diagnosis. In order to improve the high-dimensional data feature selection of kernel extreme learning aachine(KELM), a novel method of combining sparse auto-encoder(SAE) with KELM was proposed. Firstly, the vibration signal of time domain, frequency domain and time-frequency domain features were extracted to constitute a high-dimensional feature vector. Then, multi-layer SAE fusion was used to eliminate the redundancy of the features. Finally, the fused characteristics were used to train the KELM and the fault diagnosis model was obtained. According to the sensitivity of KELM to parameters, the firefly algorithm was used to optimize the parameters. To assess the validity of this method, the laboratory test data was adopt to compare the proposed method with the traditional KELM. The results show that this method has better accuracy and stability.更多还原