Power Quality Disturbances Classification Based On Wavelet Transform And HMM

With the improvement of the new industrial techniques, the amount of non-linear load in the grid is increasing and the grid is suffering from more and more serious power quality issue. In order to ensure the safety, reliability and economical operation of the power system, it is necessary to study the power quality problem in depth. The analysis of the power disturbances are the foundation of power quality management. The power quality can be enhanced at the premise of locating the fault. Thus there is no denying the importance of researches on the classification of power quality event.Based on the PSCAD/EMTDC software platform, the voltage sag, voltage swell, voltage interrupt, voltage oscillation and harmonic five kinds of power quality signal are stored. This paper proposed a new signal character extracting method, using the wavelet and multi-solution theory. This method can reflect how the signal decomposition’s energy changes with time.This paper introduced the basic theory of HMM and its algorithm in the process of learning and recognition. Based on the MATLAB software platform, 5 HMMs about the power quality event were trained and then unknown signals coming from the five events were classified. At the end of the paper, a system that combines a signal producer, a sampler, a controller and a computer is introduced. The system can sample and recognize an unknown power disturbance signal.The experience results showed that the method proposed in this paper could recognize the five power quality event effectively. The method has advantages of less calculation complexity and high reliability. The method can be used to classify other power quality events as well.