Transformer copper loss prediction method and system

The invention provides a transformer copper loss prediction method and system, and belongs to the technical field of electric power, and the method comprises the steps: collecting the historical working condition data of a transformer; preprocessing the collected historical working condition data; the power efficiency is calculated according to the collected power signals, and the copper loss degree is represented by the reciprocal of the power efficiency; generating a training sample of the transformer copper loss prediction model; an improved time sequence perception neural network rTCN-Attention prediction model is constructed, and a final prediction model is obtained through training; acquiring current working condition data, and predicting the copper loss degree at the next moment by using the improved time sequence perception neural network rTCN-Attention prediction model; and determining an early warning grade according to the predicted copper loss degree. According to the method, the three-dimensional