This paper presents an efficient event detection and classification technique for multiple power quality (PQ) disturbances. Initially synthetic power quality disturbances are simulated and then are directly processed to proposed algorithms to generate the target feature sets which comprises of energy, entropy, root mean square (RMS), mean, standard deviation, kurtosis, variance and maximum peak respectively. After the overall data analysis, it was found that thirteen power quality events out of the overall generated PQ disturbances were distinctively classified. Eventually these target features are passed through simple decision tree based event classifier for PQ events classification. The proposed algorithms are change detection filter (CDFT) with noise, without noise and synchrosqueeze wavelet transform (SST) has been scrutinized for number of disturbances presented in the PQ events. The proposed technique SST is applied for PV based microgrid to enhance the real time performance of the proposed technique where it has been verified as a superior technique as compared with the some of the existing event classification techniques such as wavelet transform (WT), stock well transform (SR),etc. The entire process has been verified in the in the MATLAB /Editor. The proposed technique evades the need of further signal processing techniques for detection and classification PQ events, thus ensconced less computational complexity and faster execution. Hence it is an efficient algorithm for real time applications.
A new method of power quality disturbance classification based on deep belief network
