Research On Integrated Energy System Modeling And Muliti-Energy Cooperative Optimal Control

Integrated energy system is a new type of energy system that integrates multiple energy sources,equipment and technologies,and the research on its modeling and multi-energy cooperative optimal control has important theoretical significance and application value for realizing the functions of multi-energy complementarity,cooperative management and optimal operation.In terms of modeling,how to seek a balance between accuracy and practicability through factors such as ’model order,number of parameters,data dimensions,and data quantity’,and how to seek reasonable data processing methods through intelligent algorithms are urgent problems that need to be solved in the current research.In terms of control,methods such as adapting to external changes,reducing prediction errors,eliminating measurement errors,and optimizing variables are all important components of control strategies,and are effective ways to improve control performance in accordance with indicators such as energy saving and efficiency enhancement.In this regard,this paper takes the simplified model as the starting point,determines the model parameters,data dimensions and data volume through theoretical analysis,builds the model with intelligent search and machine learning algorithms,and conducts modeling research from three perspectives of system-level static,dynamic and device-level.With the goal of realizing the functions of the integrated energy system,the control strategy is constructed using adaptive control,parameter update,interference suppression,optimizer control and other methods,and the research on multi-energy cooperative optimal control is carried out.The main Innovative work is as follows:(1)In the research of system-level static modeling,the energy relationship of the system is reflected as a coupling matrix,and a coupling matrix intelligent modeling method based on directed multigraph is proposed.Select nodes in the form of energy to reduce data dimension.Build a directed multi graph and an adjacency matrix.Reduce modeling difficulty by iteratively searching data.Build a coupling matrix by combining energy paths.This method has the characteristics of generalizability,low data dimension,small amount of computation,low modeling difficulty and fast modeling speed.(2)In the research of system-level dynamic modeling,considering adding dynamic links to make up for the deficiency that the coupling matrix only contains static characteristics,an intelligent modeling method of multi-layer network structure of integrated energy system is proposed.Based on the directed multigraph,it is expanded into multiple layers by duplicating energy nodes,and the device dynamic model is set between the layers,forming an improved form of the coupling matrix.This method establishes a structure combining the dynamic characteristics of the equipment with the energy relationship of the system,which can be used in various comprehensive energy systems and the equipment can be arranged flexibly and has high practicability.(3)In the research of device-level modeling,considering that the device model is suitable for the multi-layer network structure,a machine learning modeling method based on Koopman space up-dimensionality is proposed.Parameters that reflect the amount of data and the dynamic performance mode are set.The Koopman operation is used to upgrade the space,and the linear relationship of the data is found through training,then the parameter values are obtained.Equipment polynomial model is fitted based on the parameter values.The data dimension and volume are limited in the modeling process,so practicality is improved,while accuracy is improved by identifying patterns.This method is versatile,can fully simplify the model and flexibly change the model form,and is suitable for combining with system-level models.(4)In the research of multi-energy collaborative optimal control,consider the rational use of existing models,an adaptive model predictive control method for data-driven online model reconstruction is proposed.Initiate control in an adaptive manner.Obtain data through collection and prediction,and update parameters in combination with machine learning algorithms to achieve online reconstruction.Set measurement parameters to suppress interference.Estab an optimizer with the goal of realizing functions,and perform optimal control.Through the advanced control strategy,this method improves the control speed by maintains the minimum amount of data,reducing forecasting errors by ensuring optimal forecasting mode,improve control accuracy by fully excavate data information.From the perspective of comprehensive energy system function realization,this paper combines system-level static model,system-level dynamic model and device-level model with data-driven online model reconstruction adaptive model predictive control method,optimization control method,PID control and other methods.Applied to the design optimization,energy management and optimal control of different types of integrated energy systems,verifies the availability of the proposed modeling and multi-energy collaborative optimal control method,and achieves the effects of multi-energy complementation,energy saving,and efficiency enhancement.