Analysis Of Window Opening Behavior In Tianjin Dormitories

It is urgent to reduce the building energy consumption according to the sharp increase of it.With the improvement of the national living standard,the indoor air quality gets more and more attention.Natural ventilation is one of the important means to improve indoor air quality.Window opening behavior not only has great impacts on indoor air quality and thermal environment,but also greatly changes building energy consumptions.In recent years,many researchers began to study it.They analyzed the influencing factors of window opening behavior and established the predict model.Most of the existing predict models are based on Logistic regression algorithm,which generally are too complex with low prediction accuracy,so their practical value is low.The total energy consumption of university in China is huge,and the per capita energy consumption of universities is far higher than that of China,which has a high potential for energy conservation.Most of the university dormitories don’t have a unified mechanical ventilation system.Natural ventilation is commonly used to improve the indoor air quality,and indoor air quality cannot be effectively guaranteed.In order to solve those mentioned problems,this paper conducts a complete heating season monitoring of window opening behavior,outdoor environmental parameters and indoor environmental parameters in dormitories of a university in Tianjin.The experiment data are used to improve the performance of the window prediction model.After analyzing the data,this paper improve the traditional Logistic window prediction model,adjusts the critical value of the model and proposes a new logistic regression model with high prediction accuracy.Meanwhile,this paper introduces average Bayesian network,LDA(Linear Dominant Analysis),QDA(Quadratic Dominant Analysis),and SVM(Support Vector Machine)into the study of window opening behavior for the first time.The results show that the average Bayesian network and SVM algorithm have very high application value in the prediction of window opening behavior,and the prediction accuracy of both two models are more than 80%.At the same time,this paper also uses K-means clustering and system clustering algorithm to propose a more suitable ventilation model for energy consumption simulation.The clustering model is introduced into the Energy Plus,and the energy consumptions of building heating under the condition of natural ventilation,heat recovery without mechanical ventilation and mechanical ventilation with heat recovery are analyzed and compared.Through the simulation results and the test data,it shows that natural ventilation can not effectively improve the air quality in the dorms.But window opening behavior greatly increases the heating energy consumption of the building.The mechanical ventilation system with heat recovery has the best ventilation results,with the lowest building heating energy consumption,which means mechanical ventilation system with heat recovery has a very high potential.