一种模式关联主元分析的海流机故障检测方法

针对海流机复杂工况下发电过程数据的多模式和模式频繁变动的问题,提出一种模式关联主元分析方法。从理论上分析模式变化对传统主元分析(principal component analysis, PCA)的影响,描述了过程数据多模式下的故障检测问题。提出一种模式标准化算法,动态拟合多模式数据特征。通过构建多模式关联关系,将变化模式引起的统计量差值剔除。通过搭建海流机试验平台,对比所提方法与传统检测方法验证了所提方法的有效性。理论分析和试验结果表明:在海流机变转速同时变载荷工况下,所提方法能够快速准确的检测出故障。

A mode-correlation principal component analysis for the fault detection of marine current turbine

To solve the problem of multi-mode characteristic and frequent mode changes, a detection method for marine current turbine which called mode-correlation principal component analysis was proposed. The influence of modal change on the traditional principal component analysis(PCA)was analyzed in theory. The detection problem caused by multi-mode characteristic was described. A mode normalized algorithm was proposed in the proposed method to dynamic fitting the mode. The statistical difference value of different modes was removed due to relationships between modes. Compared with other methods, the experimental platform was built to verify the effectiveness of the proposed method. Theoretical analysis and experimental results showed that the proposed method could detect the fault quickly and accurately under the condition of variable speed and variable load.