Data-Driven and Interactive Fault Diagnosis of Distribution Switches

In recent years, with the increase in the scale of distribution networks and the type of devices, traditional fault diagnosis approach relying on human experiences is faced with serious challenges. Meanwhile, through testing and online monitoring of various devices, a large amount of operating data has been acquired and stored, providing the necessary foundation for the proposed data analysis. To enhance the efficacy of various distribution devices maintenance, a data-driven and interactive diagnosis platform is designed and implemented. In this platform, physical models of devices, historical diagnosis records and offline experiment data are comprehensively utilized. After extracting association rules between fault characterization and potential defects, a Bayesian network is constructed based on the models of devices and trained by the historical diagnosis records. Using this Bayesian network, the fault diagnosis is made to facilitate function experiments and health examinations of distribution device. An interactive diagnosis method is also proposed, which enables interactive and incremental reasoning based on feedback from test personnel.