Research On Thermal Comfort Of Rural Residences Based On Machine Learning

The research goal of this thesis is to improve the current situation of poor thermal performance and poor living comfort of rural residences,and improve the living quality of villagers.Taking the thermal comfort of residences as the starting point,select rural residences in southwest of Hubei province as the research object.Based on the theoretical analysis of geographical conditions and climate characteristics in southwest of Hubei province,field investigations were carried out on the situation of rural residences.Using the method of combining on-site measurement and questionnaire survey,a more comprehensive analysis of the residential thermal comfort of local residents.According to statistical calculations,the thermal neutral temperature of local residents in summer is 25.3℃,and the acceptable temperature range for 80% of residents is 22.0～28.6℃;the thermal neutral temperature in winter in this area is 13℃,and the acceptable temperature range for 80% of residents is 8.4～ 17.6℃.In addition,the survey results show that local houses have poor thermal insulation performance and poor airtightness.Thermal comfort is difficult to meet the needs of residents,and there is a lot of room for optimization design.In order to make the research more in-depth and more scientific,this thesis used Energy Plus software to evaluate the indoor thermal comfort rate.This article combined local regional characteristics and selected eight thermal environment optimization strategies,including building orientation,outer wall insulation thickness,roof insulation thickness,roof slope,area ratio of window to wall and exterior window glass material and thickness,shading method and summer ventilation strategy.Then simulated the thermal comfort improvement effect under different optimization strategy combinations.Using the simulation results as training data and using MATLAB to built a platform,and then established a thermal comfort prediction model for rural residences in southwest of Hubei province based on BP neural network.It has been verified that the correlation of this model can reach 96.9%,and the relative error between the training sample and the output of the BP neural network is within 0.6%.And it can be used to evaluate the impact of residential design schemes on indoor thermal comfort.In order to further explore the thermal comfort potential of rural residences in southwest of Hubei province,this thesis adopted the particle swarm algorithm with strong optimization ability to establish a thermal comfort optimization model based on the prediction model.Through the restriction of the constraint conditions,the best optimization strategy combination of traditional residential and newly-built residential thermal comfort was obtained.Subsequently,the results was analyzed and evaluated from four aspects:theory,comparison of two types of housing optimization strategies,comparison of orthogonal test results,and combined with the regional characteristics of southwest of Hubei province.It is proved that the model is an optimized model with ideal effect.