工作站广泛应用于工业生产.现有的PCA故障诊断模型在检测工作站故障时存在模型过度拟合、指标冲突、参数敏感度低等问题,误检漏检较多.针对上述问题,提出一种基于改进PCA模型的工作站故障诊断方法,即MA-PCA(Multi-index Principal Component Analysis).在传统PCA主成分分析故障诊断模型基础上,在主元空间和残差空间计算T2和SPE统计量,引入控制限融合深度系数的统计量指标,调整控制限,根据主成分中的参数贡献量反馈调整诊断模型,构建自适应更新的故障诊断模型.最后通过实例分析,验证了所提出的改进PCA模型故障诊断方法的可行性,并与传统PCA模型和三项故障诊断方法的诊断结果进行比对,该方法在故障诊断准确率上可提高2.7％～8.2％.
Research on Workstation Fault Diagnosis Method Based on Improved PCA
Workstations are widely used in industrial production.The existing PCA fault diagnosis models have problems such as overfitting,conflicting indicators,resulting in a high number of false positives.To address the above issues,a workstation fault diagnosis method based on an improved PCA model,namely MA-PCA(Multi index Principal Component Analysis),is proposed in this paper.This method is based on the traditional PCA fault diagnosis model and calculates T2 in the principal component space and SPE in residual space.The control limit is introduced to fuse the depth coefficient with the statistical index,and the control limit is adjusted.Based on the parameter contribution in the principal components,the diagnostic model is adjusted through feedback,and an adaptive updated fault diagnosis model is constructed.Finally,the feasibility of the proposed improved PCA model fault diagnosis method was verified through case analysis,and com-pared with the diagnostic results of traditional PCA model and three fault diagnosis methods.The proposed method can improve the fault diagnosis accuracy by 2.7％to 8.2％.
万方数据
维普
