Prediction of Water Supply Temperature in Secondary Pipe Network Based on Machine Learning
Taking a heating station in a residen-tial area in Weifang City as the research object,the da-ta such as wind direction,wind speed,weather condi-tions(referring to physical phenomena such as cloudy,sunny,rainy and snowy skies),average outdoor temper-ature,and average indoor temperature were used as in-put variables,prediction models of water supply tem-perature of secondary pipe network based on machine learning were constructed,and the prediction effects of the prediction models were compared.The prediction models include back propagation neural network(BPNN)model,support vector regression(SVR)model,and random forest model.All three prediction models can predict the water supply temperature of the secondary pipe network.Among the three prediction models,the predicted values obtained by the random forest model have a higher degree of agreement with the measured values,and the error fluctuation range be-tween the predicted value and the measured value is smaller.Regardless of weather conditions,the evalua-tion indicators of the random forest model are superior to the other two prediction models.The random forest model has the best prediction performance.Compared with not considering weather conditions,the prediction performance of the random forest model considering weather conditions is improved.
water supply temperature of second-ary pipe networkmachine learningprediction model
国家科技期刊平台
NSTL
万方数据
