【摘要】 现有的短期热负荷预测涉及参数较多，增加了热负荷预测难度，导致预测性能下降，为此，提出一种基于长短期记忆神经网络的热力站短期热负荷预测方法。使用皮尔森相关系数得出热力站热负荷主要影响因素。利用LSTM算法构建短期热负荷预测网络模型，将获取的温度、日照、用户行为三个维度特征作为LSTM算法的输入，预测未来一段时间的热力站热负荷变化需求。实验结果表明，LSTM算法各时间点的负荷误差最大仅为0.297W/m~2,预测结果与实际热负荷需求呈正相关，具有实用性。更多还原

【Abstract】 The existing short-term heat load forecasting involves many parameters, which increases the difficulty of heat load forecasting and leads to a decline in forecasting performance. Therefore, a short-term heat load forecasting method for heat stations based on long-term and short-term memory neural networks is proposed. The Pearson correlation coefficient is used to obtain the main influencing factors of the heat load of the thermal power station. The LSTM algorithm is used to build a short-term heat load forecasting network model, and the obtained temperature, sunshine, and user behavior characteristics are taken as the input of the LSTM algorithm to predict the heat load change demand of the heating station for a period of time in the future. The experimental results show that the maximum load error of the LSTM algorithm at each time point is only 0.297W/m2, and the prediction results are positively correlated with the actual heat load demand, which is practical.更多还原