The Smart Grid Power Configuration Software And Load Forecasting Algorithm

Compared with traditional power grid, Smart Grid possess more advantages, suchas strong, interactive, safe, efficient, compatibility and self-healing, could providequick decision for grid staff, and optimize power resource configuration.Configuration Software is the intelligent core of Smart Grid, which played a decisiverole, so a Power Configuration Software is designed with electric consumptionprediction in this paper, which has significant meaning to improve the powerdeployment precision of Smart Grid. This paper mainly studies the prediction methodfor power comsuption, and designs a Power Configuration Software. The maincontents are as following:This paper analyzes the development and research status of ConfigurationSoftware, introduces the overall design process of Power Configuration Software,discusses various issues in software designing process in detail, such as the databasemodeling, data collection and communication protocol. The system development toolis confirmed, so the basic function of graphics rendering system is completed.The basic principle of grey theory and the modeling process of GM(1,1) isstudied, and the power consumption is predicted and simulated through GM(1,1).Aiming at the defect of low prediction precision in GM(1,1) model, a second orderpolynomial optimized GM(1,1) model is present-HGM(1,1). Simulation results showthat the prediction precision of HGM model is better than that of GM model.As HGM(1,1) model only considers the power consumption, and neglects otherouter factors, a HGMBP model is proposed combining gray model with BP neuralnetwork model based on the fundmental principles of BP neural network model. Itserves the predicted value of HGM(1,1), weather, and the quantitative results oftemperature as inputs, trains the process with BP algorithm, and synthesizes thefactors affecting power consumption. Simulation tests show that HGMBP modelimproves greatly in power prediction, especially in certain period when powerconsumption is distinctly affected by meteorological factor. The global optimization characteristics of genetic algorithm is studied becauselocal minimum and slow convergence speed appears when weight threshold is trainedin HGMBP model. With genetic algorithm to optimize the weight threshold ofHGMBP model, a HGMHBP model is proposed to achieve appropriate threshold.Simulation tests show that the prediction precision of HGMBP model decreases withless training data. So the designed Power Configuration Software adopts HGMBPmodel to predict power conumption in Smart Grid.