Load Forecasting Method Based on CS-DBN-LSTM

Abstract

Accurate load forecasting can improve the take-up rate of electricity, and ensure the electricity demand of residents for production and living can be met in time. A load prediction method integrated DBN and LSTM was adopted in the paper. Model used historical load data and weather data as data to improve prediction accuracy, since electricity is affected by factors mentioned above. The cuckoo search optimization was also introduced to find the best parameters for improving the prediction accuracy. The experiment result showed that the load forecasting algorithm proposed is with higher accuracy for the load forecasting compared with DBN, LSTM and DBN-LSTM.