Research On Distributed Energy Management Strategy In Energy Local Area Network

The energy source of the global energy interconnection covers a wide range of energy fields.Through the comprehensive utilization of various energy sources,to pursuit the greatest value of clean energy in social and economic development and environmental protection.The energy local area network,which consists of renewable energy power generation equipment,distributed controllable power generation equipment,energy storage systems,electric vehicles and smart loads,is a subnet of the energy interconnection.It is necessary to optimize its energy management to operate economically and safely and to work in synergy with other energy local area networks to improve the overall performance of the energy interconnection.In the case of a variety of distributed energy sources within the energy local area network,PV(photovoltaic)and wind power,which are the main power generation method of the future energy Internet,are more environmentally and the structure is more easier to construct than other power generation methods.But its existence is affected by the environment,and it can not ensure stable energy supply because of uncertainty,volatility and other issues.This article evaluates the energy forms within the energy local area network from three aspects: energy output,economic benefits and environmental impact.Analyzes the non-linear characteristics of distributed energy output of energy local area network based on photovoltaic and wind power.Because of distributed energy output needs a large number of qualifications which is unnecessary for neural network.The prediction methods based neural network are used,because the model predicts it can not be well adapted to the new environment and has not fitting ability for nonlinear problems.Aiming at the fluctuation of electricity load and analyzing the changing rule of electricity load,a method based on NARX neural network for energy load and distributed energy prediction is proposed.In training sample data processing,the principal component analysis is used to select training sample.In the process of NARX neural network prediction,genetic algorithm is used to avoid the neural network from falling into the local minimum and the data is optimized through gray relation to reduce the randomness of parameters.Predict power consumption and distributed energy output by using improved NARX neural network.The weather features such as clear,cloudy,rain,snow and other abstract elements for quantification.Through the actual example,the energy dispatching optimization model is validated.According to the prediction results,the energy storage device is simulated and the energy efficiency of distributed energy is improved.