Condition Information Extraction Model Based on Dynamic Principal Component Analysis for On-Condition Maintenance

To the vast condition information variables of high dimension obtained from condition monitoring, the condition information extraction model is studied. Only cross correlation is considered in current condition information extraction (such as principal component analysis), which is excessively broaden the hypothesis condition of the model. It doesn´t agree with the fact. So the dynamic principal component analysis is applied to extract condition information for on-condition maintenance. The detailed application process of dynamic principal component analysis is studied, and auto-regression model is adopted to determine delay time. And condition information extraction model is established considering time series which is not smooth.