Research On Optimal Design Of Wind Environment In Block Based On Parameterization And Machine Learning

With the rapid development of our country's economy and the continuous increase of urban population,the number and scale of high-density residential areas are also increasing rapidly.Environmental problems in residential areas are becoming increasingly prominent,and the wind environment is one of the main problems.The wind speed at pedestrian height plays an important role in the daily life of residents.It will affect the indoor and outdoor thermal comfort,energy consumption and pollutant diffusion,and then threaten the health and comfort of residents.Therefore,how to optimize the design of the wind environment in residential areas is an urgent problem to be solved at this stage.This paper proposes a residential area wind environment optimization design method based on the "parametric modeling-Open FOAM simulation-machine learning" framework to study the influence of residential area morphology on summer ventilation and winter wind protection.Machine learning is used to establish a predictive model.The research is mainly divided into three parts:First,a residential area model is established based on the parametric platform Grasshopper.This paper selects five morphological elements(land orientation,building height,building length,building staggered and building overhead),and generates several residential area cases through the value changes of morphological elements.The value range of each morphological element is determined by the statistical results of the sample data,and conforms to the technical regulations of Wuhan local construction project planning and management.Secondly,using Open FOAM software to simulate the outdoor wind environment of the residential area model,using four evaluation indicators(summer-suitable wind speed area proportion,summer-wind speed dispersion,winter-quiet wind area ratio and winter-wind speed dispersion Degree)to evaluate the outdoor ventilation efficiency of residential areas and obtain a data set containing morphological elements and evaluation indicators.Finally,the data in the database is divided into two parts: training set and test set.Machine learning methods are used to learn the data in the training set.Then the test set is used to verify the learning results,and the error evaluation index is used to evaluate the difference between the predicted value of the prediction model and the actual value in the database,and the model with the best predictive performance is selected.The results show that:(1)After optimization,the ventilation in summer and windbreak in winter are improved greatly.Compared with the maximum(small)value obtained by the first generation optimization objective,the maximum value of summer suitable wind speed area proportion increases by 16.8%,the minimum value of summer wind speed dispersion decreases by 0.345,the maximum value of winter calm area proportion increases by17.6%,and the minimum value of winter wind speed dispersion decreases by 0.328.(2)Machine learning method can be used to improve the efficiency and accuracy of wind environment prediction.Among the six machine learning models,limit gradient lifting has the best prediction performance.(3)Based on the five types of morphological factors that affect the wind environment of residential areas,this paper puts forward the optimal design strategies of wind environment from three aspects: comprehensive consideration of summer ventilation and winter wind protection,focusing on summer ventilation,focusing on winter wind protection.Designers can choose according to the actual needs.