Research On The Load Forecasting And Multi Time Scale Scheduling Optimization Of Integrated Energy System

The prediction of multiple loads of Integrated Energy System has a great significance to optimize energy structure,master demand fluctuation to achieve supply and demand balance,improve renewable energy consumption level and safe and stable operation of the system.Among them,short-term and ultra short-term forecasting of multiple load plays an important role in the optimization of system scheduling and other aspects.Therefore,the precision and speed of short-term and ultra short-term multiple load forecasting has been a hot topic in domestic and foreign research field.The traditional model of power load forecasting is linear and single.The traditional research methods often adopt linear system method,which can only reflect the single correlation between the related variables.For the integrated energy flow input of the integrated energy system,the linear coefficient prediction method can not characterize its nonlinear relationship and its strong nonlinear correlation coupling.Therefore,based on the coupling of multiple loads of integrated energy users,this paper considers the copula theory to analyze the nonlinear correlation between multiple loads and the loads and weather.The application of Copula theory requires the establishment of edge distribution function,estimation of unknown parameters,selection of optimal copula function and other key steps,which is closely related to the practical effect.Based on the nonlinear correlation of various variables,the dimension of input data is decoupled by kernel principal component analysis,and then the predictive output of generalized regression neural network under the optimal parameters is achieved,and the accurate prediction of multi load demand of comprehensive energy users can be realized.After the short-term accurate prediction of multiple load,the multi-time scale scheduling optimization of the integrated energy system in industrial park is considered.Based on the previous multiple load demand forecasting results,this paper establishes the daily economic scheduling model of integrated energy system based on the prediction results of multiple load demand,in the day ahead scale and based on the short-term forecast results,aiming at the lowest energy consumption cost of regional comprehensive energy,and considering the solution through Cauchy variation particle swarm optimization algorithm;secondly,rolling control is carried out on the basis of the results of the ultra short-term load forecasting in the day at 15 minutes in a day.The goal of energy consumption cost is the lowest in the time domain,and more accurate multiple load forecasting results and photovoltaic output prediction results are updated in the near time period,thus achieving higher source load matching degree;finally,real-time feedback optimization is carried out at the scale of 5 minutes,and the safety optimization is carried out with the minimum system adjustment as the target considering the uncertainty of source load caused by prediction error,so as to achieve the system fluctuation level.Moreover,the stable output of the unit can maintain the safe and stable operation of the system.