Building Energy Efficient Optimization Study Based The Sensitivity Analysis Of Design Parameters

China’s energy and environmental problems are increasingly prominent,building energy consumption accounts for 21% of the total social energy consumption,and green sustainable development has become the mainstream direction of social development.The traditional architectural design is subjective and lack of science,and the final architectural design scheme can not guarantee the energy-saving demand.Therefore,in order to improve the efficiency of building energy conservation,it is necessary to carry out efficient energy-saving design based on parametric analysis combined with local climate and environmental resources.In this paper,combined with the method of sensitivity analysis,using machine learning agent model,taking office buildings in Tianjin as the research object,this paper puts forward the optimization design strategy based on sensitivity analysis in the preliminary design stage of buildings,and compares with the overall optimization,which verifies the necessity of sensitivity analysis and comparison.First of all,through reading the relevant literature of building energy-saving optimization design,summed up the variables affecting energy-saving in the preliminary design stage of the building,and constructed the energy consumption model of office buildings in two cold regions.Then,the parametric model of building energy consumption is established based on the grasshopper platform of parametric design in the preliminary design stage of each building.Combined with Monte Carlo method,super Latin square sampling is carried out for building design variables,and BP neural network is used to train the acquired data set to obtain a high fitting building energy consumption prediction model.Then,using the established energy consumption prediction model,the sensitivity analysis of passive design variables related to energy conservation of buildings is carried out under different global sensitivity analysis methods.Finally,the influence rules of various parameters of buildings and their interaction on building performance are obtained,which provides theoretical basis for the key optimization design direction of the final building scheme.The top three variables of building energy consumption model are window type,floor height and lighting density,but the influence of air tightness in total building load is greater than that in peak load.In the building energy consumption model B,the length width ratio,height and the thickness of roof insulation layer are the most important parameters.The cost and sensitivity ranking of different sensitivity analysis methods are comprehensively analyzed and compared,and Morris method is preferred for sensitivity analysis of office building energy consumption model in cold area.Finally,the NSGA2 multi-objective genetic algorithm combined with the prediction model is applied to optimize the design strategy based on sensitivity analysis,and compared with the overall optimization design strategy.Through the analysis of the optimization process and Pareto optimal solution set,the distribution degree of the solution set of the former will be reduced,but the optimization results are almost the same,and the optimization cost is greatly reduced,which provides a new and effective reference for the future building performance optimization design based on parametric analysis.