In view of the traditional power load forecasting algorithm model of slow training speed and the prediction problem of poor effect

a parallel load forecasting method is proposed based on the deep belief network. Based on parallel computing framework and deep belief network

the method is parallel train the history pow-er load and weather information data

and load values is forcasted through the training model. The experimental results show that the average error between the predicted power load value and the actual value is low and the prediction accuracy is higher than the traditional method. It effectively reduces the elapsed time of consuming training and prediction

and can adapt to the prediction demand in large-scale power data scenarios.