Thermal Load Forecasting And Planning Of Central Heating System

Short-term heat load forecasting is the basis and prerequisite for achieving"on-demand heating" of central heating systems,solving the problem of uneven heating,and ensuring the comfort of hot users.Heat load is affected by many factors,and the relationship between each factor and heat load is both linear and non-linear,making the heat load prediction more difficult.Therefore,it is very important to select the appropriate forecasting method and establish the appropriate thermal load forecasting model to provide the operator with pertinent guidance.This article takes the self-discipline of heat load and its influencing factors as the research object.The main research contents include the following points:Firstly,the historical data are preprocessed,then the dynamic influence of meteorological factors on heat load is explored.Through the analysis of the correlation between each influencing factor and the thermal load and the multivariate stepwise regression method to determine the input variables,a multiple stepwise regression equation was obtained.Since the heat load is also affected by its own laws,the impact of meteorological factors on the thermal load is not entirely linear,The multiple stepwise regression model can better reflect the change trend of the heat load within a specific time period.Exceeding this time period,the generalization ability of the model will deteriorate and the accuracy will become lower.Therefore,it is not appropriate to linearize the relationship between heat load and meteorological factors over a full period of time.Secondly,in order to solve the nonlinear problem between thermal load and meteorological factors and to improve the search ability of standard particle swarm optimization algorithm,a thermal load forecasting model based on particle swarm optimization support vector machine(DPSO-SVM)with dynamic adjustment of similarity weights is proposed.The results of DPSO-SVM model show that the DPSO-SVM model has better performance than the PSO algorithm.The DPSO-SVM model can well solve the problem that the multiple stepwise regression model has strong generalization ability in a specific time period and is weak in the ability to generalize beyond the time period.compared to the multiple stepwise regression model,the accuracy of the model increased by 4.9%.Finally,the DPSO-SVM thermal load forecasting model based on major holidays is proposed.the accuracy of the revised model is improved by 12.7%.Therefore,the DPSO-SVM model is more suitable for predicting the thermal load on a time-of-day basis.The DPSO-SVM prediction model was applied to a heating company in Dalian to predict the heat load value in the future at a time and perform heat supply according to the heat planning value.According to the feedback from thermal users,the DPSO-SVM thermal load forecasting model is reliable and can provide targeted guidance for operators.Therefore,the DPSO-SVM thermal load forecasting model has played a positive guiding role for heating enterprises.