【摘要】 为了进一步提高齿轮裂纹故障诊断精度，设计了一种基于KPCA和PSO-SVM方法的齿轮裂纹故障信号诊断方法。对时频域进行特征降维处理时通过主成分分析(PCA)的方式实现高维信号噪声。为了抑制支持向量机(SVM)算法分类精度低问题，引入粒子群算法(PSO)进行加强，并使SVM算法获得更优的核函数。研究结果表明：以SVM-PSO算法进行处理时相对其它算法表现出了更高的精度，能够满足优异的分析性能。以SVM-PSO算法进行处理时的精度最高，能够满足高稳定性要求，所需计算时间也较合适，PSO方法有助于算法分类精度及效率都获得显著提升。样本数对处理时间影响很大，综合判定训练样本数设置1600是相对比较合理的。采用时域特征时会导致分类精度下降，外频域特征相对时域特征可以达到更高的精度。更多还原

【Abstract】 In order to further improve the precision of gear crack fault diagnosis, a gear crack fault signal diagnosis method based on KPCA and PSO-SVM is designed. Principal component analysis(PCA) is used to realize high-dimensional signal noise when feature dimension is reduced in time-frequency domain. In order to suppress the low classification accuracy of support vector machine(SVM) algorithm, particle swarm optimization algorithm(PSO) is introduced to strengthen the SVM algorithm and obtain better kernel function. The results show that the SVM-PSO algorithm has higher precision than other algorithms and can meet the excellent analysis performance. The SVM-PSO algorithm has the highest accuracy in processing, which can meet the requirements of high stability and the required calculation time is appropriate. The PSO method helps to significantly improve the classification accuracy and efficiency of the algorithm. The number of samples has a significant influence on the processing time, so it is relatively reasonable to set 1600 samples for training comprehensively. The use of time domain features will lead to a decline in classification accuracy, and the outer frequency domain features can achieve higher accuracy than the time domain features.更多还原