基于气象因素的集中供热系统热负荷预测研究

Research of District Heating System Heat Load Prediction Based on Weather Factors

集中供热系统各子系统的给定值都是由预报热负荷决定的,提供准确的热负荷预测是提高供热质量的基础;传统热负荷预测仅考虑室外温度的影响,并且热负荷预测模型参数的辨识仍依靠传统数学工具,精度不够高;为了更准确预测,研究了多个气象参数对集中供热系统热负荷的影响,采用了多元回归法,利用1stOpt软件中的LM-UGO算法建立了集中供热系统热负荷预测模型;实验结果显示,室外温度对热负荷有直接影响,风速或日照对室外温度有直接影响,然后间接影响热负荷,同时,多元回归拟合的平方相关系数均在0.900 0以上,模型训练、测试的平均绝对百分比误差均在4.00%以下;应用实例表明,热负荷预测模型的训练与测试均比较合理,这种多元回归法适用于在热负荷预测邻域推广与使用。

The set value of district heating system＇ s subsystem depends on forecasting heat load, providing accurate heat load predic- tion is the basis to improve the heating quality. For the traditional heat load prediction, the outdoor temperature was considered as the only one influence factor, the heat load prediction model＇ s parameters identification depend on traditional mathematical tools, and the model＇ s accuracy isn＇ t high. In order to predict the heat load more accurately, the multiple weather factors＇ influences on the district heating sys- tem heat load were researched in this paper, moreover, the multiple regression method was used in this paper, and then, the district heating system heat load prediction models were established by Levenberg Marquarat--Universal Global Optimization （LM--UGO） algorithm in lstOpt software, The experimental results shown that the outdoor temperature has a direct influence on the heat load, the wind speed or so lar radiation has a direct influence on the outdoor temperature so as to impact the heat load indirectly. The results shown that the multiple regression fitting＇ s square correlation coefficient were all greater than 0. 900 0, model training and testing＇ s mean absolute percentage er ror were all less than 4.00%. This application indicated that the training and testing process of the heat load prediction models were reason- able, it proved that this multiple regression method can he promoted into the heat Load prediction field.