Heat and Power Load Prediction for Combined Heat and Power Station with Different Methods

The accurate short-term load prediction is an essential prerequisite for ensuring the reliable and economic operation of combined heat and power stations. In this study, the heat load and power load prediction models of combined heat and power station were developed with long short-term memory (LSTM) and multiple linear regression (MLR), respectively. The preprocessing and correlation analysis of heat and power loads were conducted. The prediction effects of different prediction models were also compared and analyzed. Taking the data of February as an example, the mean absolute percentage error (MAPE) of heat load prediction with LSTM and MLR models are 3.20% and 3.75%, respectively, and the MAPE of power load prediction are 5.59% and 8.76%, respectively. The prediction results show that the performance of the LSTM model is better than that of the MLR model, and the prediction accuracy of the heat load is higher than that of the power load.