Research On The Short-term Forecast Of Multi-element Load Of The Integrated Energy System Based On The Theory Of Intelligence

As a technological innovation in the field of energy,the development of integrated energy system affects human social life.The dynamic balance between the energy supply and the various load needs is very important to the supply quality of integrated energy system.The multivariate load short-term forecasting of the integrated energy system is the main measure to solve the above dynamic problems.Therefore,it has become an important research field of integrated energy system.In order to improve the accuracy of multivariate load forecasting,it is necessary to analyze the characteristics of each load before forecasting.In this paper,firstly,the load characteristics and the load variation law of all kinds of loads are analyzed through different load indexes.Secondly,the Spearman coefficient derived from Copula theory is used to quantitatively analyze the nonlinear correlation between multivariate loads and meteorological factors.In order to improve the learning and generalization ability of the model,based on the nonlinear coupling between multivariate loads and weather factors,the kernel principal component analysis((KPCA)),which can deal with nonlinear data,is selected to reduce the dimensionality and decouple the original data,and the generalized regression neural network(GRNN)is selected to realize the short-term prediction of multivariate loads.Through the analysis of numerical examples,it can be seen that compared with PCA-GRNN model and GRNN model,KPCA-GRNN model can realize multivariate load forecasting more accurately.After studying and analyzing the GRNN model,in view of its shortcomings,an improved generalized regression neural network(IGRNN)model is constructed.On the basis of GRNN,the multi-bandwidth of the model layer is extended,and a screening layer is introduced between the pattern layer and the summation layer to enhance the discrimination ability of the model to the effective quantity and the interference quantity.After analyzing the mathematical principle of bat algorithm(BA),based on its shortcomings,it is improved by introducing inertia weight factor and learning factor and adding direction constraint,and the improved bat algorithm(IBA)is used as the optimization training method of IGRNN model to realize the multivariate short-term load forecasting of integrated energy system.Through the simulation of an example,compared with other models,the KPCA-IBA-IGRNN model proposed in this paper has higher prediction accuracy and has certain theoretical significance.