Fault Classification for Smart Switchgear Based on Multivariate Multiscale Cloud Sample Entropy and Fuzzy Support Vector Machine

1Multi-source monitoring data of smart switchgear can be used for abnormal state recognition and fault classification to realize efficient operation and management of distribution equipment. In this paper,sensors were used to monitor features such as voltage, current, temperature, flash signal, and so on.Multi-variable multi-time-scale cloud sample entropy fault features of switchgearwere obtained with lower half-trapezoid cloud model to quantify similarity of composite delay vectors of featuring time series and to soften similar tolerance criterion of multi-variable multi-scale sample entropy. This paper used piecewise half-trapezoid cloud model to quantify relationship uncertainty between fault samples. Regional difference and dispersion of sample space were synthesized to calculate sample membership, and classification method based on fuzzy support vector machine was formed to identify different switchgear fault types. Through case analysis on monitored data, correctness of the proposed scheme was validated.