The Research Of Optimal Operation For A Micro-grid Based On Gray Neural Network Power Prediction

Micro grid optimization is still a problem at present.Its economic benefits,environmental benefits,as well as to the energy management is still difficult to be solved.The study of micro grid optimization can improve the quality of power and promote the application of micro power grid in power industry.Power prediction of pv and wind machine can ease influence on the stability of distribution network because of uncontrolled micro power fluctuation,improve given ability of pv and wind machine,improve the interconnection security of network and distribution network.This paper begins with the prediction model of the output power of photovoltaic and wind machine and studies environmental protection cost,operation maintenance cost and comprehensive benefits the micro grid.There are many factors that result in the power output of pv and wind machine which appear to be irregular,so power prediction of pv and wind machine can be regarded as a grey system.Grey prediction model choses history output as the input value generation,it can accurately grasp the potential trend of original data sequence by weakening sequence randomness and enhance the regularity of historical data.This paper adopts grey prediction method which suitable for "poor information,small sample" system.It leads to larger forecast errors because there is no basis for parameters selection of original gray GM(1,1)model.In this paper,the selection of background values adopts the optimization method and the selection of initial values adopts the least square method according to the results of prediction.The modified grey prediction GM(1,1)model improves the accuracy of micro grid power prediction.The grey prediction GM(1,1)model considers the impact of historical data on power prediction.The effects of temperature,solar radiation,wind speed,wind direction and other meteorological factors are not considered.So this article introduces the BP neural network,using the additional momentum method and adaptive vector to solve convergence problems of photovoltaic power,using the Hilbert huang transform to solve influence on prediction precision because of the wind speed fluctuation In the end,this paper puts forward the prediction model of grey neural network optimization combination to apply power prediction for wind power and photovoltaic by analysing prediction error.The grey neural network has little power fluctuation and small prediction error compared with the gray model and neural network model.So the prediction of power generation by using the combination of grey neural network is feasible and effective.Combined with the power prediction of micro grid fan and photovoltaic,the paper puts forward the environmental protection cost and operation maintenance cost and comprehensive benefit of multi-objective optimization model,.the method of the application of maximum membership degree changes the multi-objective function into a single objective function,and use self-adapting crossover mutation operation and individual optimal retention policy optimization genetic algorithm to obtain optimization model.According to the different grid of micro grid and large power grid trading patterns using four different micro power grid operation plan,the paper selects a typical summer day to analyse a certain area of micro grid simulation with different objective function and operation strategy,the simulation shows that multi-objective optimization can obtain higher profits than single objective optimization,and use small loss as far as possible to obtain high efficiency and environmental protection.According to different operation strategy simulation,it shows that the time-sharing electricity price under the micro grid and bidirectional energy distribution network communication scheme can achieve higher output result.