Research On The Risk Warning Method Of The Ferromagnetic Resonance Over-Voltage In Distribution Network Based On The Hidden Markov Model

In undergrounded power distribution network, the ferromagnetic resonance over-voltage is a common power failure which seriously threats the safety and stable operation of power system. In recent years, many experts and scholars have done a lot of research on ferroresonance through theoretical analysis, test simulation, computer simulation methods, and proposed several measures to avoid this phenomenon. But those researches still focus on using signal processing methods to identify the type of overvoltage after the failure, which do nothing to early warning before the failure. Meanwhile, the current risk warning mechanism bases on the long-middle-term, static risk assessment, which doesn’t reveal the mechanism of ferroresonance and can’t be used to risk warning timely pre-failure.In this paper, the writer analyze the mechanism of ferroresonance, generalize the necessary conditions and characteristics, mining the signs and signal characteristics of the operation parameters before the ferroresonance, and analyzes the excitation conditions which cause the ferromagnetic resonance, studying the evolutionary trends with single phase grounding fault disappearing as stimulating condition And then, we establish a simulation model based on ATP-EMTP program to get operation data, based on the signal characteristics before and after the ferroresonance, the neutral point voltages are decomposed by wavelet package separately, and the characteristic band is chosen based on the principle of maximum energy. According to the failure process all the life can be described through the feature extraction method. Based on the above researches, a risk warning method of ferroresonance based on the Hidden Markov Model (HMM) is studied through taking wavelet packet coefficients as the prognostic feature information. Experimental results show that this method can not only effectively identify the current system in which the state, but also warning of future the risk of a period of time the system might face.