Maximum likelihood estimation of energies in partial discharge pulse signals based on 3F-C method

To suppress the influence of white noise and narrowband interference on the energy measurement of partial discharge pulse signals, we proposed an energy estimation method based on noise parameters. Furthermore,we investigated the field signal model and the probability distribution of the energy spectral coefficients, and obtained the maximum likelihood estimation of the pulse energy with known noise parameters. Meanwhile, to acquire the noise parameters, the sampled data were divided into signal frames and noise frames, and the 3F-C method was adopted to estimate the parameters of the white noise and narrowband interference. Through simulation experiments and field data processing, the results of the proposed method were compared with those of the conventional wavelet packet de-noising method. The data analysis results show that, compared with the wavelet packet de-noising method, the proposed method has a better accuracy in energy estimation for either white-noise environment or mixed-noise environment with narrowband interference. With the decrease of the signal-noise ratio and the increase of the signal time window length, the advantage of the proposed method becomes more significant. For measured partial discharge pulse signals with a signal-noise ratio of -8 dB, the relative error of energy estimation using the proposed method is only 4.07%, while that of the wavelet packet de-noising method is 44.12%. © 2017, High Voltage Engineering Editorial Department of CEPRI. All right reserved.