Research On Cooling Load Correction Coefficient Of Capillary Radiation Air Conditioning Room

With the continuous development of science and technology,human beings have higher and higher requirement for living environment.Compared with convection air conditioning system,radiation air conditioning system can provide a more comfortable indoor environment.In this case,this kind of air conditioning system have gradually begun to be known by people.Temperature and humidity independence control air conditioning system has been proposed by Chinese scholar for a long time,and many areas have also been adopted this type of air conditioning system.Due to the different ways of heat exchange,both the methods of calculation and results of cooling load of convection air conditioning and radiation air conditioning have differences.However,the relevant design standard of radiation air conditioning system only provide interior design temperature for reference,and the calculation method is based on the convection way,which means that the difference between the two kinds of air conditionings has not been taken into consideration and will bring troubles for designer.In order to study the difference in cooling load between the two systems and facilitate the calculation of cooling load of radiant air-conditioning rooms,this paper proposes a cooling load correction coefficient₈（9） and a lot of researches of it are carried.At first,a capillary radiation ceiling panel+displacement ventilation air-conditioning system is established in TRNSYS,the accuracy of model are verified and a reasonable simulation time is analyzed;six factors which affect the value of cooling load correction coefficient are selected from perspective of heat balance equation group,the correlation analysis is also carried out.Then the influence methods and processes of different influencing factors on cooling load are analyzed in view of heat gain in TRNSYS;at last,launching the mathematical models which include grey box model and black box model of the cooling load correction coefficient,the multiple linear regression equation of the cooling load correction coefficient and the BP neural network model are obtained.The results show that the six factors that affect the cooling load correction coefficient are location,room temperature,orientation,room type,window-to-wall ratio and roof.The order of influence is as follows:roof>orientation>area>window-to-wall ratio>room temperature>room type.At the same time,a new correlation factor"Area×Orientation"interactive variable was discovered.This interactive variable can be described by the variable of the maximum radiation intensity of a certain direction in a certain area.Orientation,area,window-to-wall ratio,and room temperature affect the cooling load correction coefficient by affecting the radiant heat of the room,while the roof affects the cooling load correction coefficient by affecting the heat conduction of the radiant ceiling and roof.In the design and load calculation of radiant air-conditioning systems,more attention should be paid to the above-mentioned influencing factors.The accuracy of multiple linear regression equation is related to the number of variables.Reducing the number of variables can simplify the equation form and decrease the work of design,but at same time the accuracy will be lowed.A grey box model which takes room temperature,window-to-wall ratio,maximum radiation intensity,and roof as variables is the optimal model.The mean square error MSE is 5.43×10^-4,the average absolute error MAE is 0.0188,and² is 0.943,and this multiple linear regression equation can be considered for practical application for design work.Two BP neural network models both have high model accuracy and prediction accuracy.The mean square error MSE of neural network training model is 3.8×10^-4 and 1.7×10^-3,the correlation coefficients R of the training results are 0.988 and 0.961 respectively,and the correlation coefficients R of the prediction results reach 0.9816 and 0.9382.It demonstrates that BP neural network can reflect the internal relationship of cooling load correction coefficient and influence factors well.The BP neural network with the influence factors in the article as input have high prediction accuracy and strong nonlinear mapping ability.BP neural network can reliably predict the cooling load correction coefficient.