Accurate prediction of concentration of the dissolved gases with fault characteristic in power transformer oil is of great significance to discovery of early latent power transformer faults and to achievement of safe and economical operation of power system.A grey-Markov combination prediction model of concentration of the dissolved fault-characteristic gases in power transformer oil is proposed based on the multivariable grey model and Markov prediction theory.The multivariable grey model established with the optimized formula of background value is used to dynamically predict the macro development trend of the fault-characteristic gases for power transformer.Moreover

a state transition probability matrix of time series of the fault-characteristic gases is obtained

thus high-precision prediction of the concentration of the fault-characteristic gases is implemented by taking the normalized self-correlative coefficients as weights to correct the predictive value of the multivariable grey model via the weighted Markov chain model.The practicality and effectiveness of the model is demonstrated by results of an actual numerical example.