Research On Heating Room Temperature Based On Functional Data Analysi

According to the 2022 Building Energy Consumption and Carbon Emissions Research Report,the total carbon emissions of the whole process of buildings in China accounted for 50.9% of the national carbon emissions in 2020.Due to the cold weather in the northeast,central heating carbon emissions have exceeded the total carbon emissions of buildings half.Compared with similar developed countries,energy consumption of building heating per square meter is high,but thermal comfort is low,so reducing the energy consumption of heating system is the most potential direction for energy conservation and emission reduction in the building field.With the improvement of data collection and storage technology,more and more data are presented in the form of functional curves and images.The traditional statistical method has many assumptions in the analysis,which is not suitable for this kind of high-dimensional data.Functional data is the observation sequence of random variables in the function space,and each random object is a curve,in which ”functional” refers to the generation process of data is a functional process.The basic idea of functional data analysis is to treat the observed data as a whole,functionalize the discrete data,and then use non-parametric statistical methods to process the data.Under fewer assumptions,the analysis of high-dimensional data can be realized and the internal structural characteristics of the data can be obtained.The room temperature data during heating period shows obvious functional characteristics with time,so this paper discusses the heating room temperature from the perspective of functional data analysis.Based on the functional data analysis method,this paper analyzes the hourly room temperature data of 27 households in a community in Changchun for 102 days from November 25,2020 to March 6,2021,including two aspects of heating room temperature difference and heating room temperature prediction.When studying the difference of heating room temperature,the hourly temperature data of each household in the heating period is taken as a research object.Through functional principal component clustering analysis,27 households in the residential area are grouped into two categories,roughly grouping families in the same building into one category.After clustering,the average temperature difference between the two types of buildings is about 3 degrees,and the functional ANOVA proves that the difference is statistically significant,so that the heating company can find the causes of uneven heating and timely adjust.When predicting the heating room temperature,the daily hourly average room temperature of 27 households in the residential area during the heating period is regarded as a functional data object,and the daily hourly average indoor temperature data from November 25,2020 to February 28,2021 is used to predict the average indoor temperature data from March 1,2021 to March 6,2021.The indoor temperature was predicted from two perspectives of functional time series analysis and functional regression analysis,and the mean square error was used to evaluate the model.The results show that the prediction of the two models can reflect the change trend of the room temperature,and the functional regression model has smaller prediction error in this study.Therefore,the further modification of the regression model in practical application is expected to conduct reasonable regulation on the heating system in time according to the regression prediction results,so as to reduce the heating energy consumption.