| Virtual Power Plant(VPP)has advanced communication,measurement,regulation technology,and market policies and mechanisms.It is an effective way to achieve efficient aggregation and optimal control of large-scale distributed energy under the "dual carbon",and is also a feasible strategy for new power systems to achieve interaction and intelligence on the energy supply side.How to coordinate various aggregated resources of VPPs to meet the operational requirements of VPPs under the constraints of safe operation of the power grid is also a key issue that needs to be urgently studied.This paper focuses on three periods of VPP including early planning,planning,and operation,and conducts in-depth research on the internal configuration and optimal operation of VPP.The dynamic characteristics of key elements are analyzed with researching on the flexibility and overall randomness of aggregation resources,providing theoretical support for aggregators to choose elements for VPP.On the basis of internal optimization configuration,optimal the external trading or interactive power of the virtual power plant involved,to eliminate power deviation and fluctuations caused by distributed energy;Further,consider the application situation of multiple virtual power plants coexistence in the same regional distribution network in the future,and carry out distributed coordinated optimization of multiple virtual power plant(MVPP)considering privacy;Finally,a VPP operation and control scheme for practical engineering applications is designed,and an operation and control platform system interface is constructed to promote new energy consumption and assist the development of new power systems.The main content and achievement of the paper are as follows:Based on the overall architecture and aggregation elements of VPP,starting from the power side and the demand side,the output characteristics and resource characteristics of different aggregation elements are analyzed and compared,and the similarities and differences between commercial VPP(CVPP)and technical VPP(TVPP)are elaborated.Specify the appropriate control architecture and configuration method by considering different types and application scenarios of VPP,at the meantime propose the deployment scheme of VPP operation platform in power system.A general model of VPP in power systems is constructed by researching on the composition conditions,control architecture,configuration methods,and mathematical models of internal resources of virtual power plants.,general models for different virtual power plants with multiple access modes and components have been built in DIgSILENT.The above lays the foundation for the research of virtual power plants in the following chapters.Aiming at the difficulties of dynamic aggregation and quantitative evaluation with different resources in VPP,a scheme for dynamic optimal configuration of internal resources is proposed.Starting with modeling the flexibility and randomness of various types of resources in a generic model,built a VPP optimal configuration system including "flexible adjustable resources,aggregation adjustment characteristics,optimization target modeling".Firstly,analyze the adjustable flexibility and randomness of different distributed generation,traditional generators,controllable loads,and energy storage,establishing a flexibility and randomness model for each aggregation element at the level of flexible adjustable resources.Secondly,At the level of aggregation adjustment characteristics,on the one hand,the flexibility of various resources is integrated to obtain the external adjustable flexibility of VPP,on the other hand,the randomness model of VPP is obtained by convolving the probability density function of the randomness of various resources.Finally,based on the portfolio theory,establish an optimal configuration model considering the randomness of VPP,and achieve dynamic matching of internal unit capacities through sensitivity analysis of key parameters,which provide a theoretical basis for the internal resource configuration of virtual power plant.At the external optimization level of VPP,a multi-level optimization architecture and model including day-ahead planning,rolling optimization,and real-time dispatch is proposed through Markov Decision Process(MDP)to address the problems that traditional optimization models may encounter when there are many aggregated elements in VPP,such as cumbersome calculation processes and longer solution time,And achieve optimization robustness through the idea of "multi-level optimization,step-by-step refinement".In the day-ahead plan,establish a VPP optimization model aimed at maximizing market benefits;In rolling optimization,the MDP model is established to describe stochastic dynamic programming problems,and the entropy function is constructed to characterize the magnitude and volatility of power deviation,and write it as a rolling optimization objective function into the feedback equation of MDP,effectively suppressing power fluctuations and peak valley differences;In real-time dispatch,fuzzy probability strategy is used to write it into the MDP strategy set,and realtime feedback is used to adjust the balanced energy storage charging and discharging capabilities,improve the tracking effect of the day-ahead plan,and maximize the economic benefits of the virtual power plant.Based on the Markov decision process,the two-layer optimization of 30-min scale rolling plan and 5-min scale real-time dispatch are combined to reduce the modeling level unify solution ideas,and improve computational efficiency.To solve the problem of coordination and mutual assistance among multiple virtual power plants in regional distribution networks,based on the idea of "information separation and decision coordination",an optimization mechanism for distributed coordination of multiple virtual power plants based on the decentralized optimization architecture is proposed.Firstly,pointed at distributed coordinated optimization mechanism for MVPP,a multi-period coordinated optimization dispatch model for multiple virtual power plants is established,building a transaction price function between multiple virtual power plants based on supply and demand relationships to enhance the willingness to interact between virtual power plants..Then,the optimization model is relaxed using Lagrange dual relaxation theory,transforming the original problem into a distributed optimization problem for MVPP,and reconstructing the day-ahead multiperiod coordinated optimal dispatch problem into a real-time optimal dispatch problem based on DEC-POMDP,effectively addressing optimization dispatch deviations caused by prediction errors while ensuring reasonable allocation of resources and output plans within the VPP within one day.Finally,solve the optimization problem based on the improved quantum genetic algorithm.Using coordinated optimization is more economical than independent optimization.At the same time,build the distributed coordinated optimization model based on the dual relaxation principle,global optimization can be achieved by exchanging only a small amount of information between VPPs,protecting the privacy of each entity while improving computational efficiency.Based on the functional deployment scheme and data path design of VPP,a VPP operation and control scheme for practical engineering applications is designed.Using configuration ideas,a VPP operation and control platform is built to achieve collaborative management and control of distributed resources,improve new energy consumption capacity,and promote friendly coordinated regulation of source network load storage. |
PDF Full Text Request