Effective power quality disturbance (PQD) events recognition is important for electricity network operation. This paper proposes an improved methodology for detecting and classifying multiple PQD events, which may simultaneously occur in the electricity network. It employs an adaptive dual resolution Stockwell-transform for signal processing and a short-term Renyi entropy estimation for feature extraction. Also, a multi-class support vector machine is employed to classify the types of PQD events. A variety of power quality disturbance signals are synthesized to evaluate the effectiveness and efficiency of the proposed method. These signals include both single fault/abnormal event and mixed events (e.g. a mixture of several PQD events occurring simultaneously). Real-life disturbance recordings from a transmission network service provider are also collected and used to verify the proposed method. The results show that the proposed method can achieve high classification accuracy at noisy environment. Especially, it is capable of identifying the disturbance signals comprising both harmonics with low total harmonic distortion levels and transients. Moreover, the algorithms developed in this paper can achieve a high computational efficiency.

No abstract