Abstract: In order to meet the demand of high-precision fault early warning for 35 kV power transformers, an automatic early warning method based on dissolved gas analysis in oil was developed. By establishing the complex relationship function between the internal fault of transformer and the dissolved gas in oil, the accurate identification of the internal fault type of transformer is realized. Further, key features, including gas content ratio and temperature trend, are extracted from transformer operation data to provide input to the prediction model. The LSTM network is used to predict the transformer gas content, and the alarm threshold is set to realize the automatic fault warning. The practical application shows that the method can accurately predict the content of dissolved gas in transformer oil, send early warning signal in time, effectively avoid power interruption and safety accident caused by transformer fault, and improve the stability and reliability of power system. This method provides powerful technical support for the operation and maintenance of 35 kV power transformers.