Study On Statistical Platform For Building Energy Consumption And Prediction Method Based On BP Neural Network

The statistics of buildings Energy consumption is the main ways of grasping building energyconsumption data, is basis for energy audit, only really possess the basic information of alltypes of building and the information on energy consumption can truly realize the energysaving potential of buildings to achieve building energy efficiency, while the governmentbuilding energy efficiency regulatory measures and economic incentive policies must rely onthe statistics of energy consumption. At present, the statistical work of building energyconsumption in China has carried out four years, the research is minimal on the applicationsoftware system of building energy consumption statistics, resulting in inefficient of energyconsumption statistics, statistical methods have to be improved, the current large amount ofaccumulated data on basic energy consumption statistics did not play the role expected. Theforecast of future energy consumption of building contribute to building energy efficiency.Forecast on the building energy consumption is little. Based on the above situation, this papercarried out the following work：At first, Study the statistic indicators system on the building energy consumption and thereporting system on the energy consumption of civil buildings, establish database o based onthe SQL server2000. The data table consists by three parts of the construction of basicinformation, the building energy consumption information and building prediction sampledata, multi-tables are applied between the various parts to achieve the inheritance of the childtable to parent table: the operation of adding, modification and deleting in the parent tablewill generate new changes in the child table accordingly; he operation of adding, modificationand deleting in the child table has no impact to the parent table. In addition, the establishmentof a reliable user access control to protect the database security.The second, In the analysis of the working mechanism of the BP neural network basedon its optimization, to determine the neural network architecture, including network layersand layers of neurons, hidden neuron layers and hidden neurons; the most value Thenormalized input sample data; the finalization of the neural network learning and trainingprocess. Based on theory of BP neural network to establish the forecasting system by usingDelphi6.0. Include the basic building information management, building energy consumptioninformation management, building energy consumption summary energy consumptionforecast sample data management and building energy consumption forecast management。At last, based on2010energy consumption statistics and this Guangzhou City,Dongguan City, the building energy consumption data, and analysis of building energy audits and energy-saving potential, validated the validity of improving data processing、assistingbuilding energy audit and forecasting building energy consumption via BP neural networkalgorithm by the software system.