基于PCA 和KNN 的电主轴故障诊断方法研究

电主轴是数控机床中重要的部件之一, 其性能的优劣直接影响机床工况和加工零件质量。对电主轴进行故 障诊断能很大程度上提高数控机床的加工精度, 并且能够有效地增加其可靠性和安全性。在一般诊断过程中, 原 始数据的高维特征量处理较为困难。为顺应实际应用中对电主轴故障诊断的精度要求, 提出一种基于主成分分析 (PCA) 与K 最近邻(KNN) 的电主轴故障诊断方法。此方法利用PCA 对原始非线性时间序列数据的特征向量进行 降维, 并选取其中主成分特征向量。将得到的主成分特征向量作为KNN 的输入进行故障分类。最后将该方法的预 测结果与决策树和随机森林的分类结果进行对比, 结果表明, PCA-KNN 算法在故障分类精度上相较于其他两种算 法有显著提高, 是一种有效的电主轴故障分类方法。

Research on Fault Diagnosis Method of Motorized Spindle Based on PCA and KNN

Motorized spindle has become the most important part of modern numerical control. The performance of motorized spindle can directly affect the efficiency of computerized numerical control (CNC). Diagnosing when the potential failure occurs can highly extend CNC life, and increase its reliability and safety. But during the practical diagnosis process, it meets so many difficulties to handle high dimensional eigenvalues of tremendous amount of original collected data. To fulfill the demand of accurate classification of spindle failures, this paper reaches out a solution which based on principal component analysis (PCA) and K nearest neighbor (KNN) for bearings failure classification. This solution is first using PCA to reduce dimensionality of eigenvector of original data to select the principal component eigenvector and output the non-linear time series data. Set the outcome of PCA dimensionality reduction-the non-linear time series data-as the input of KNN, thus to get the bearings failure classification. And comparing to the results that come from decision tree and SVM, the conclusion shows that the accuracy of KNN combined with PCA is highly above decision tree and SVM, which indicates that the method in this paper is an efficient way of spindle failure diagnosis method.