Research on fault diagnosis of UPFC based on wavelet packet energy

In order to make up for the lack of UPFC protection system in fault diagnosis and early warning, a new UPFC fault diagnosis method based on wavelet packet energy and OVO SVM has been proposed. During the simulation, the signals of DC side capacitor voltage of the UPFC model in the states of normal and 4 shunt valve side internal fault are collected. The wavelet packet analysis is used to decompose the signal into three layers and the wavelet packet energy of each frequency band is extracted to form the original sample space. The sample data are filtered by principal component analysis (PCA) algorithm to filter out the effective fault information. Then the data after dimension reduction are sent into one-versus-one SVM (OVO SVM) for training and testing. Experiments prove that the 5 running states of UPFC mentioned above can be identified in a very short period of time effectively and accurately by this UPFC fault diagnosis technology and to some extent, the safe and stable operation of UPFC is able to be prospectively guaranteed.