Forecasting of dissolved gas in transformer oil by optimized nonlinear Grey Bernoulli Markov model

Accurate forecasting of concentration of the dissolved gas with fault characteristic in power transformer oil is of great significance to detect the incipient fault in transformer and to achieve safe and economic operation of power system. This paper proposes a novel comprehensively optimized nonlinear Grey Bernoulli Markov model which suitable for forecasting fault characteristic gas in oil. In order to improve the precision, this paper use Metabolic model to modify the original data and we also smoothing the gas data by Exponential smoothing operation. In addition, the optimized model is improved by the modification of the background value. The ONGBM(1,1) can be used to forcecast Weishan Dali and Jinghong Xishuangbanna 110kV Substation faulty with real measured dissolved gas data. Predictive analytics use the GM(1,1) and ONGBM(1,1), respectively. The results show that the optimized prediction accuracy is greatly improved.