Research On Short-term Forecasting Of Multi-element Loads In Integrated Energy Systems Based On Quantum Weighted Gated Recurrent Unit Neural Networ

In recent years,energy consumption has continued to grow,and energy problems have become increasingly prominent.The traditional single energy system can no longer meet the development of today’s economy and society,and the Integrated Energy System(IES)is in the ascendant.The integrated energy system takes the power system as the core and realizes the coordinated planning and operation of multiple energy sources through energy conversion equipment and storage equipment.In recent years,it has become an important means to deal with energy and environmental problems.IES multi-load short-term forecasting is the premise and basis for the safe and reliable operation of the system,and it must properly deal with the problems of strong volatility and high coupling caused by the relatively small scale of IES and the interconnection of multiple energy sources.First,aiming at the short-term multivariate load forecasting for integrated energy systems,this paper proposes a short-term forecasting model for multiple loads based on the Quantum Weighted Gate Recurrent Unit(QWGRU)neural network.The maximum information coefficient theory,which has better performance for nonlinear variables,is used for correlation analysis,and the input quantity of the model is selected;then multi-task learning architecture is used to construct multi-load forecasting model based on quantum weighted GRU neural network,and the quantum information processing mechanism is used to make the multivariate load forecasting model have strong better nonlinear approximation and generalization capabilities.The simulation results show that the short-term prediction accuracy of the multivariate load prediction model of the quantum weighted GRU neural network proposed in this paper is better.Secondly,in order to improve the information processing ability of the multivariate load forecasting model and improve the accuracy of the multivariate load short-term forecasting of the integrated energy system,a short-term prediction model of multiple loads based on Quantum Weighted Multi Hierarchy Gated Recurrent Unit(QWMHGRU)neural network is proposed.In this paper,the gating structure of GRU is improved to form a Multi Hierarchy Gated Recurrent Unit(MHGRU),which has two reset gates and two update gates,and the quantum weighted neurons are used to form the MHGRU,and the quantum weighted multi hierarchy gated recurrent unit neural network multivariate short-term load forecasting model is formed.The simulation results show that the weighted average accuracy of the QWMHGRU multivariate load forecasting model in summer and winter can reach more than 97%,which is higher than the MHGRU,QWGRU and GRU models.Finally,in order to further improve the short-term forecasting accuracy of multiple loads,a multistage multiple load short-term forecasting model considering high-dimensional eigenvectors is proposed.A quantum weighted denoising autoencoder(QWDAE)is used to extract the potential high-dimensional feature vector in the multi-load,and it is used as the input feature of the model;then the QWGRU model and the QWMHGRU model are used to build a multi-stage prediction model.The simulation results show that the high-dimensional vector extracted by the quantum weighted denoising autoencoder contains the potential characteristics of multiple loads,which can improve the load prediction accuracy,and the multi-stage load prediction model can obtain better load prediction results.