The invention discloses a full-automatic power consumption prediction method, which comprises the steps of preprocessing a historical power consumption time sequence to obtain historical power consumption data; decomposing the historical power consumption data by adopting a Boosted HP filter to obtain a trend term and a cycle term; determining parameters of an Auto ARIMA model according to the data of the trend term to obtain a prediction model of the trend term; training the seq2seq model according to the data of the cycle term to obtain a prediction model of the cycle term; superposing the prediction result of the prediction model of the trend term and the prediction result of the prediction model of the cycle term to obtain an electricity consumption prediction model; wherein the powerconsumption prediction model is used for predicting power consumption. According to the method, only historical power consumption data is utilized, and dependence on exogenous data is not needed; parameters or self-adaptive learning parameters can be preset for the adopted HP filter, Auto filter and seq2seq model, and the parameters do not need to be set manually according to data. Compared with atraditional model, the method has the advantages that the prediction precision is greatly improved, and the method is convenient to use.