Load Forecasting And Operation Optimization Of An Radiant Heating/Cooling Air-Conditioning System With Fresh Air Supply Based On Data Mining

In order to achieve the goal of carbon neutralization,the task of building energy conservation is becoming more and more urgent.At the same time,with the improvement of people’s requirements for indoor thermal comfort,radiant cooling/heating and fresh air conditioning systems have attracted more and more attention and research.Compared with traditional air conditioning,radiant cooling/heating and fresh air conditioning have obvious advantages of energy saving,comfort and health.However,due to the complexity of the system itself and the system application is still in the preliminary promotion stage,there are some problems,such as insufficient research on load and energy consumption level,high proportion of energy consumption of fresh air system,mismatch between operation characteristics and operation control,and a large number of actual operation data to be further analyzed and excavated.Taking the radiant cooling/heating and fresh air conditioning system as the research object,this paper makes an in-depth analysis by using data mining technology,and puts forward and verifies the application of data mining technology in load forecasting,energy consumption and load influencing factors and operation optimization of radiant cooling/heating and fresh air conditioning system.This paper mainly carries out the following research:Firstly,the comprehensive energy consumption level and energy efficiency level of fresh air unit in the actual operation of the system are evaluated,and the variation laws of climate characteristics,unit water temperature and other parameters under continuous operation are analyzed.The annual comprehensive operation energy consumption per unit area of the system is 47.40 kWh/m~2.The fresh air energy consumption in the air conditioning season accounts for 42%of the total energy consumption of the HVAC system.The dehumidification COP_Dof the fresh air unit is concentrated between 1.2-1.6 and the refrigeration COP_Cis concentrated between 1.1-1.8,which is quite different from the rated value of COP.Secondly,a zone-level artificial neural network load forecasting model is constructed to predict the moisture load and cooling load in summer.In order to improve the prediction accuracy of the fresh air load forecasting model and consider the impact of personnel activities in the building on the air conditioning load,a new air load forecasting model considering the division of thermal zones is established.The results show that the prediction accuracy of the zoning artificial neural network model has obvious advantages.The coefficient of variation of the root mean square error(CV-RMSE)of the prediction results of the four zoning models ANN3 moisture load and cooling load are 8.72%and 9.98%respectively,which is 14.27%and 14.61%lower than that of the model without zoning.The ANN model for load forecasting of cold source units has good prediction accuracy,and the CV-RMSE of the prediction result is 6.86%.Then,the relationship between the influencing factors and the load and energy consumption of the fresh air system is analyzed.Based on the grey correlation analysis method,the relationship between the fresh air load at time t and the influencing factors at time T-n is analyzed to explore the impact of the dynamic delay effect on the fresh air load.Based on the results,the input layer of the ANN model is optimized,and an improved ANN prediction model considering the dynamic delay effect is established.The correlation degree of time T moisture load with T-1 moisture load and T-6 outdoor temperature is relatively large,and the correlation degrees are 0.896 and 0.808 respectively.The outdoor temperature at T-6and moisture load at T-1 are introduced as the input parameters of the improved Ann model.Among the evaluation indexes,the improved ANN model is greatly improved,and the CV-RMSE is reduced from 22.99%to 9.33%.Finally,based on the load forecasting results,the system operation optimization strategy is formulated.The energy storage optimization control strategy of fresh air system is applied for typical load day and the whole air conditioning season.The optimization results reduce the operation energy consumption and operation cost by 27.2%and 29.2%respectively.Applying the load distribution optimization strategy in the typical working conditions of the air conditioning season,the overall partial load rate(PLR)of each unit is improved.Under the typical working conditions,the total energy consumption of the system fresh air unit and cold source machine room in 6 days is 291662kWh.The optimized load operation strategy saves11.2%of the total energy consumption compared with that before optimization.