Research And Application Of Multivariate Statistical Analysis In Building Climate Division

With the rapid development of current economy,the problem of building energy consumption has become increasingly prominent,and the green building and building energy saving technology has been given high staus by government.In order to design a green building,building climate zoning has become the focus of current research.From the perspective of multivariate statistical analysis,this thesis uses a variety of statistical methods to study the climate of China's building.The meteorological data(4~8 times daily)from 476 representative meteorological stations around the climate characteristics in China in past 33 years are selected as the research object,the abnormal data were excluded,and the missing data is replenished by interpolation and elimination.Furthermore,the data are smoothly processed to remove the noise,as a result,a database that reflects the nature of the problem is built.Nine partition variables are constructed which are latitude,longitude,altitude,heating days,heating control temperature,cooling period of stay,cooling control temperature,solar radiation and dew point temperature.Using statistical regression analysis method,the 9 partition variables at certain years(as of 2017)are analysed.Moreover,the values of regression prediction are given,and the results are prospective.The definition of weighted mixed distance is given,which reflects the weather,time as well as space.This definition not only reflects the meteorological distance,but also reflects the spatial and temporal distance.Cluster analysis was used to cluster the 476 stations in the whole country,as a result,the different partition of building climate is acquired.Forthermore,the first principal component deviation analysis method is established,which provides the criterion for the determination of the best class number.By analysing and clustering more than 100 million data collected on the experimental,the national climate is divided into six regions,which are temperate very cold and humid region,warm temperate cold semi-arid and semi-humid region,temperate cold and arid region,subtropical mild humid region,temperate and subtropical four seasons region,subtropical warm winter moist region.It is shown that the classification results are basically consistent with the corresponding regional environment,and can provide a reliable gauge for the design of building energy-saving.