Multiple Data Fusion for fault diagnosis of power grid

Abstract

When transmission line faults occur in a power grid, the control center would receive a lot of data from different sources. These multi-source data should be taken together to diagnose faults accurately. In this paper, we proposed a new fault diagnosis method named Multi-source Data Fusion (MDF), which is based on Dempster-Shafer Theory of evidence using both the digital and analogue data. The digital data is the status of protective relays and circuit breakers and the analogue data is the transient waveforms of voltage and current recorded by fault recorders. By employing a fuzzy Petri network and the wavelet analysis method, we extract the fault feathers from the multi-source data and obtain four basic sets of evidence. Then, Dempster-Shafer Theory (DST) is used to combine the evidence and get the information fusion result. Finally, a C-means algorithm is adopted to identify the fault equipment. The simulation studies have been undertaken in PSCAD and Matlab. The experimental tests show that the proposed method can significantly improve the performance of fault diagnosis, in terms of accuracy and real-time response, in identifying the transmission line faults occurring in a complex power grid.