Short Term Load Prediction Of Pelagic Island Microgrid And Optimal Scheduling Of Discrete Energy

As the pivot and platform for ocean development,the pelagic islands possess strategic values both in economy,politics and military affairs.Sufficient and reliable energy supply is the foundation and guarantee for the realization of island multi-function development,the promotion of marine economic development and the implementation of the strategy of "Maritime Power".Considering the abundant renewable energy stored in the island and its surrounding waters,it can effectively guarantee the long-term reliability of power and resources supply,by improving the renewable energy generation capacity,and allocating various production and living loads.In the future development scene of the pelagic island,there will be a large number of new electric loads,which are characterized by diversification and complexity.Therefore,this paper proposes a load clustering method based on improved K-means algorithm.User types are delicately classified according to the shape characteristics of load curve.Meanwhile,the information theory is introduced into the predictability evaluation of historical load data,and the predictability index is constructed by calculating the information entropy of the sample,in this way,a priori sample evaluation scheme is established.Moreover,to qualify the large volatilities and strong uncertainties of some loads,this paper implements a prediction intervals method based on the lower upper bound theory.On the one hand,existing forecast evaluation index are improved from two aspects of interval width and cumulative error.On the other hand,by using neural network and particle swarm optimization algorithm,the prediction model is constructed and the structure parameters are acquired.The historical load data with strong uncertainty is used to validate the proposed method,the prediction and analysis verified the superiority of the method compared with the traditional point to point prediction and the existing interval prediction method.Finally,this paper proposes an energy scheduling scheme for the pelagic islands by using electric vessel(EV),and the scheme is divided into three different modes according scheduling pattern.On the basis of EV number,vessel design loading,and island predefined time window,an improved ant colony algorithm is used to search the optimal scheduling scheme under different modes.In the simulation,based on the actual geographic data of islands in the South China Sea,ship configuration and cumulative navigation distance of the different modes are presented,which demonstrate the feasibility and rationality of the proposed scheme.