Dissolved gas analysis in oil can reflect the operation status of transformer. The accurate prediction of volume fraction of the dissolved gas can provide theoretical support for discrimination and warning of early fault of the transformer. Therefore

a prediction method of dissolved gas volume fraction in transformer oil based on time varying filter empirical mode decomposition and multi-model fusion is proposed. Firstly

the gas volume traction sequence is decomposed into multiple sub-sequences by time varying filter empirical mode decomposition to reduce its non-stationarity. Then

the multi-model fusion strategy is used to fuse and reconstruct the prediction results of four different single models. Since the weight coefficient of single model has a significant impact on the prediction results

the improved slime mold algorithm is used to optimize the weight coefficient to improve the prediction accuracy. Finally

it is shown through calculation variation that the proposed method

compared to traditional prediction model

has higher prediction accuracy

and can better predict the variation trend of gas volume fraction in the oil.