Fault prediction method based on cooperative analysis of various fault characteristics of steam turbine

The invention discloses a fault prediction method based on cooperative analysis of various fault characteristics of a steam turbine. According to the invention, on the basis of non-stationary, non-linear, and slow-changing characteristics of the million-kilowatt ultra-supercritical set steam turbine fault degradation process in papers, characteristics capable of reflecting different fault degradation characteristics are extracted by using co-integration analysis, kernel principal component analysis and slow characteristic analysis methods; fault characteristics are examined comprehensively based on the monotonicity, robustness and correlation indicators, thereby extracting key fault characteristics and a residual service life indicator is generated by using DismatchityAnalysis; and then a fault prediction model is established by using a PLS. 本发明公开了面向汽轮机的多种故障特性协同分析的故障预测方法。本发明针对论文百万千瓦超超临界机组汽轮机故障退化过程非平稳、非线性、变化缓慢的特性，分别使用协整分析、核主成分分析和慢特征分析的方法来提取能反映不同故障退化特性的特征，然后利用单调性、鲁棒性和相关性指标综合考察故障特征，从而提取关键故障特征，并使用DissimilarityAna