Multiple Correspondence Analysis-Based Evaluation of Power Asset Condition Monitoring Techniques

Abstract

Condition monitoring (CM) plays an important part in power asset management to sustain the high-level reliability of a power grid. Given various power asset CM techniques (CMTs), it is necessary to evaluate and select the most suitable technique that can effectively reflect equipment failure. This paper proposes a systematic method to evaluate CMTs that allows decision makers to successfully address this problem. Firstly, a multiple correspondence analysis (MCA) model is used to analyze and visualize the association among the power asset technical parameter, failure modes, and CMTs, ensuring that they can be analyzed in the same framework. Secondly, a performance score is proposed to quantitatively estimate the correspondence of each CMT in detecting failure. Thirdly, hierarchical clustering is performed to identify the most suitable CMTs for equipment with different technical parameters. Case studies are conducted with actual data in Singapore, and the results validate the effectiveness of the proposed method.