Research On Power Flow Optimization Method For Hybrid Electricity,Gas,and Heat Energy System

With the development of the society and the progress of technology,the structure of the energy has transformed from single traditional energy to multi-source clean energy which means that the energy will transmit in multiple types.Therefore,the network has become a hybrid energy system with multi-energy coupling form.In order to optimize the system operation,power flow calculation and optimization analysis for hybrid energy system has become more important.This thesis proposes a Newton-Raphson method for hybrid power flow model considering initial sensitivity based on the mathematical model of hybrid energy system and build a regional hybrid system optimization model using reinforcement learning algorithm.Meanwhile,a distributed reinforcement learning methods is put forward to solve the hybrid energy system optimal power flow problem containing multiple We-energy.The specific research contents are as follows:1.The topology of hybrid energy system is researched including the mathematical model of power system,natural gas system and centralized thermal system.And the energy hub model is established according to the coupling characteristics of multi-energy flow.By considering the operating conditions of various energy equipment and analyzing the transmission mode of multi-energy coupling in energy hub,the foundation of hybrid power flow calculation and optimization analysis is built.2.Based on the correlation method of power flow in the power system,the node type of hybrid energy system is determined,and a new model of the Newton-Raphson method is established considering the electric-gas-heat hybrid energy.On the basis of the energy flow and node state in the hybrid system,an initial Guess Estimation Newton method is proposed to solve the Newton's initial sensitivity problem.The validity of the new method is proved by the simulation,which provides a theoretical basis for the initial selection of the Newton method for the hybrid power flow and reduce power flow calculation time.3.Considering the topology model of hybrid energy system,a system optimization model is established with energy loss,economy and sociality.Based on the Markov decision process and the model of reinforcement learning optimization,a detailed optimal flow model of regional hybrid energy system in Q-learning method is proposed through system variable discretization,reward function design,system optimization process.By simulation analysis in a regional hybrid energy system with smaller nodes,the model method is verified to be feasible.4.According to the multi We-energies composition of the hybrid energy system structure model,a double-deck optimal power flow model is established and distributed reinforcement learning algorithm is used to realize distributed parallel optimization of system.Average-consensus-based information discovery algorithm is proposed to realize the coordinated interconnection between We-energies and this method is carried out to design the distributed optimization model for hybrid energy system including the reward function and the system framework design.The validity and practicability of the method are verified by simulation.Last but not least,the thesis concludes the main research works and discusses the further work.