Multiple Load Short-term Forecasting For Integrated Energy System And Its Application

Integrated Energy System(IES)is development direction of energy consumption mode in the future.It realizes the collaborative planning and scheduling of multiple energy systems,such as electricity,cold and heat,which can effectively improve energy efficiency and promote the modular development of renewable energy.In the stage of planning and design of IES and its optimal energy dispatching,the forecasting task of multiple loads is very important.However,the current researches mainly focused on the prediction of single load,and failed to consider the coupling characteristics of multiple loads in IES.This paper proposes a short-term multiple load forecasting method based on deep learning.In order to consider the coupling influence relationship among multiple loads,Pearson method is firstly used to analyze the correlation between multiple loads and between multiple loads and other factors,so as to simplify the input dimension of network model.In order to fully explore the interaction among multiple load series and the dependence of each series data on the time dimension,this paper selects the electrical,cold and heat historical load data as well as the related weather factors such as temperature and humidity as the input,and uses the short and long time memory network to carry out the prediction modeling.In order to improve the prediction accuracy and reduce the complexity of model calculation,a particle swarm optimization method based on chaotic algorithm is proposed to optimize the structural parameters of network model.The actual data of a comprehensive energy system in an industrial park are used to carry out simulation experiments.The results show that the proposed method can achieve a good prediction effect for multiple load prediction of different typical days including cooling season and heating season,which can provide an effective reference for subsequent optimization scheduling schemes.In order to apply the method presented in this paper to engineering projects,a multiple load short-term forecasting application system is designed.The system has built modules of user rights management,data preprocessing and load prediction,and realized the functions of algorithm packaging and encryption as well as offline training and online prediction.