Identification Of Building(Group) Thermal Model In Heating System

Central heating is the main form of heating in northern China,but most central heating systems are difficult to adjust in time according to climate changes,resulting in uneven heating and cooling in heating areas.Therefore,during the operation of the heating system,it is of great significance to predict the indoor air temperature of the heating user and use the result to guide the heating,to ensure that the heating user maintains a constant and comfortable indoor air temperature.In this paper,the building(group)thermal model in the heating system is identified and studied.For the buildings(group)in the heating system,the gray-box model is established according to the first-order dynamic model of the heating room temperature,the least square method is used to identify the thermal parameters in the model,and the blackbox model is established according to the autoregressive AR model in the time series.The least squares method and AIC estimation criterion are used to determine the parameters and order of the model,and the identification procedures of the two types of models are written.Based on the simulation experiment,the thermal parameters of the two types of models were identified under five different simulated working conditions,and the fitted value of the indoor air temperature was calculated according to the identified model and compared with the indoor air temperature value obtained based on the simulation experiment simulation to verify the rationality of the identification model.The identification model was applied to the actual heat network.Taking a heat exchange station in Harbin as an experimental object,the thermal parameters of the heat exchange station and the 18 heating buildings in the heat exchange station were identified.Reliability.The identification and analysis of the model under different training periods and verification periods are also conducted,and the influence of the two on the model identification results and model selection is given.The research shows that when identifying the model,the presence or absence of noise terms and the outdoor air temperature fluctuation range will affect the identification results.The difference between the identification effects of the gray-box model and the black-box model decreases first and then increases as the noise term increases.When the noise is disturbed,the identification effect of the black-box model is better.As the noise term increases,the identification effects of the two types of models tend to be consistent.When the noise term is large enough,the gray-box model identification effect is better.When the verification period is the same and the training period is different,the blackbox model identification effect is better.The training period does not have much influence on the identification effect of the black-box model,but has a greater influence on the graybox model.When the training period is the same and the verification period is different,the identification effects of the two types of models are not the same.For the model itself,the gray-box model has better identification effect when the verification period is shorter,and the black-box model has better identification effect when the verification period is longer,and the longer the verification period,the better the identification effect.Accurately model and identify the system based on the historical operating data of a specific heating system,and predict the indoor air temperature data for a certain period of time according to the identification model It also provides a reasonable direction for the operation and management of heating systems.