Research On Operation Control Strategy Of District Heating System Based On Load Forecasting

Under the background of building a smart city,the urban district heating system in China is developing towards intelligent system,and the professional and technical personnel committed to the research of heating engineering have achieved fruitful results.In view of the development of district heating technology in China,most of the current systems are at a low level of intelligence,especially the lack of advanced operation control means,low automatic control rate,and the problem of imbalance between heat supply of heating station and heat demand of users is quite prominent.In this paper,Kaifeng JS district heating system is taken as the research object,and mainly completed the following work:Based on the Laplace transform of the heat balance equation in the typical heating station of Kaifeng JS district heating system,the secondary network backwater temperature control model is established.Based on the intelligent management platform of the target system,2664 sets of data are collected after 121 days of monitoring,and abnormal data are screened,eliminated and supplemented by horizontal and vertical processing and wavelet threshold denoising.Based on the preprocessing results of measured data,the CV-PSO-LSSVM heating load prediction model is established by using the combinatorial methods of cross validation,particle swarm optimization and least squares support vector machine.Based on the load forecasting results of CV-PSO-LSSVM model,the set values of the control system are designed,and the parameters of the generalized predictive control algorithm are determined by particle swarm optimization,and the generalized predictive controller of the backwater temperature of the secondary network is designed.The data preprocessing method used in this paper can effectively remove abnormal data and smooth the high frequency noise,and improve the accuracy of modeling.The heat load forecasting model based on cv-pso-lssvm has high forecasting accuracy and stability.Its average relative error is 1.16%,which is 62.09% lower than BP model.The particle swarm optimization algorithm can solve the difficulty of parameter setting of the generalized predictive control algorithm.Compared with PID control,the generalized predictive controller based on load forecasting designed in this paper can better track the secondary network backwater temperature.The controller has a small amount of overshoot and the advantages of fast response speed,which can effectively solve the problem of imbalance between heat supply of heating station and heat demand of users,and finally achieve on-demand heating.