Research On Demand Forecasting Factors For Energy Planning

Urban is one of the main carriers of human life and social development,consuming nearly 80% of the global energy consumption and producing a large amount of greenhouse gas emissions.China’s current urbanization and low-carbon development have entered a critical stage,and the “double carbon” goal further highlights the importance of urban energy issues in future social development.Urban energy planning can be regarded as the top-level design of low-carbon city construction.At present,most studies on urban energy issues focus on the energy consumption analysis of individual buildings or building stocks,while the composition of the energy system at the district scale and urban scale is more complex,and the influencing factors are more diverse.Energy demand forecasting and key factors identification is more difficult.The land parcel is the smallest land unit that controls the intensity of land development in the regulatory detailed planning,and the city is the implementation object of the master planning.This work introduces the spatial division methods of regulatory detailed planning and master planning,taking the urban-scale and parcel-scale energy demands as the research objects.The methods,including data survey,statistical analysis,model construction,and parameter sensitivity analysis,are used to establish an energy demand forecasting model,which can provide a theoretical basis and calculation method for urban energy planning.The main content of this article is as follows:Firstly,taking residential parcels as the research object,a method for obtaining big-data information about urban residential parcels and buildings is proposed to establish a database including nearly two thousand residential communities in Shanghai.Though combining field survey and geographic information system technologies,a series of land-use parameters(including floor area ratio,green area ratio,building density,maximum building height,household density,high-rise building proportion,and standardized compactness index)are acquired,while several building shape parameters(including building height,building base area,building shape coefficient)for more than 36,000 residential buildings are obtained.Moreover,the statistical representativeness of the established database is verified,and eighty-six prototype residential buildings for energy simulation are developed.Based on this,the cooling and heating energy demands for residential parcels are investigated.Secondly,taking commercial or office parcels as the research object,this work collected the architecture design information of more than 180 office buildings in Shanghai.The external geometric parameters of the typical office buildings are extracted to develop a parcel-scale model of building stocks for conducting building performance simulation.Based on this,an automatic generation framework is proposed.A series of land-use parameters(including volume area ratio,building density,relative compactness,building height distribution,building form,frontal area density,canyon aspect ration,street ratio,etc.)are investigated according to their sensitivities.Thirdly,taking the urban as the research object,the survey collected the 2008-2018 statistical data of economic development and energy consumption for the main downtown districts of 46 large and medium Chinese cities,which are used to establish an urban-scale energy-demand databased containing twenty-two parameters and three energy consumption indicators.Based on this,an urban-scale energy demand prediction model based on grey system theory is established.By comparing the actual energy consumption data of 46 samples,the established model has a prediction error of 5.72%for annual comprehensive electricity consumption per capita,a prediction error of 8.53%for annual residential electricity consumption per capita,and a prediction error of 15.77%for annual central heating energy consumption per unit floorage.Finally,to identify and extract the critical parameters of large-scale energy demand forecasting models,this work introduces various machine learning algorithms to construct the data-driven models to predict heating and cooling demands of residential parcels and office parcels,urban-scale central heating energy consumption urban-scale comprehensive,and residential electricity consumption.Based on this,the parameter sensitivity is quantified in terms of Sobol indices and Morris measures,while the sensitivity sequence for each energy consumption index is achieved.Moreover,an acquisition approach of minimum parameter sets is developed to obtain urban-scale and parcel-scale energy-demand-prediction minimum parameter sets.In summary,the minimum parameter set for energy demand prediction proposed in this work identifies the critical factors of energy demand in different stages of energy planning,which can provide guidance for urban data collection and urban planning and has important reference significance for urban energy planning and management.