Virtual power plant low-carbon operation method based on energy-carbon comprehensive price
By using the adaptive ADMM algorithm to build the Stackelberg game model in a multi-energy system, the problem of accurate consideration of carbon emission responsibilities in VPP in a multi-energy system is solved, the effective application of comprehensive energy-carbon prices is achieved, and the carbon emissions of VPP are reduced And improve the low-carbon and economical nature of the system.
Patent Information
- Application Number
- CN202411033233.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-15
AI Technical Summary
The existing technology is difficult to accurately consider the carbon emission responsibilities of VPP in multi-energy systems, especially the indirect carbon emissions caused by purchasing electricity and gas from the superior power grid online, and the existing algorithms cannot meet the requirements of the privacy of multi-subject transaction information, which affects the Effective application of comprehensive energy-carbon prices.
Adaptive alternating direction multiplier method (ADMM) is used to construct a multi-subject master-slave game trading model under the Stackelberg game framework. Through the coordinated optimization of distribution network, gas distribution network and VPP, the comprehensive energy-carbon price is formulated, and the energy-carbon comprehensive price is used. Carbon emission flow theory stimulates VPP to adjust energy usage strategies to achieve distributed solutions.
It effectively reduces the carbon emissions of VPP, improves the low-carbon and economics of multi-energy systems, and meets the information privacy requirements of each transaction entity, and improves the accuracy and efficiency of trading strategies.
Smart Images

Figure CN118965762A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of optimized operation of virtual power plants coupled with distribution networks and gas distribution networks, and in particular to a method for constructing and solving a multi-agent master-slave game trading model based on an adaptive step-size alternating direction multiplier method and taking into account multi-agent interactive transactions and considering VPP participation in a multi-energy system under an energy-carbon comprehensive price response mechanism. Background Art
[0002] Due to the strong uncertainty, decentralization and heterogeneity of distributed energy resources (DER), their large-scale grid connection poses a serious threat to the security of the power system. Virtual power plants have been widely used to effectively aggregate massive, decentralized and diverse DERs through advanced control, metering, communication and other technologies, and to achieve precise control of internal resources, providing an effective way to achieve low-carbon development. In addition, in the context of interactive integration of multiple energy systems, the study of multi-agent trading strategies in which VPPs participate in both the power distribution system and the gas distribution system can not only utilize the complementary characteristics of electricity and gas to achieve the coordinated operation of multiple energy networks, but also improve the economy and low-carbon nature of each entity.
[0003] At present, most of the low-carbon operation research on VPP carbon emission measurement focuses on the power supply side, and rarely extends to the user side, which makes it difficult to guide the low-carbon electricity consumption behavior on the load side. As an effective analysis method for the low-carbon development of the power system, the carbon emission flow theory provides a new idea for the allocation of carbon emission responsibilities on the load side, and also provides convenience for low-carbon demand response. In addition, the measurement methods of VPP carbon emissions in existing studies are not accurate enough. Most studies only consider the carbon emissions within the VPP, and do not consider the indirect carbon emissions caused by purchasing electricity from the superior power grid and purchasing gas from the gas grid; and the purchase and sale energy prices usually adopt fixed prices or time-of-use energy prices, without considering the huge potential of energy-carbon comprehensive prices to reduce VPP carbon emissions. Tracking the carbon flow of distributed VPP energy transactions and taking into account the carbon emission responsibilities assumed by VPP from multi-energy systems are also one of the issues that the present invention focuses on. Given that the multi-subject transaction modeling under the energy-carbon comprehensive pricing method considering carbon emission flows has certain complexity, how to achieve distributed solution while protecting the privacy of information of each transaction subject is also one of the challenges faced by traditional algorithms. Therefore, it is necessary to study the impact of the comprehensive energy-carbon price on the multi-agent master-slave game transactions of VPPs participating in multi-energy systems. Summary of the invention
[0004] The purpose of the present invention is to provide a method for constructing and solving a multi-agent master-slave game transaction model between VPP and multi-energy system based on energy-carbon comprehensive price, focusing on the energy-carbon comprehensive pricing Stackelberg game problem of VPP participating in the electricity-gas multi-energy coupling network, the mutual influence between the distribution network, the gas distribution network and the VPP, considering the game between the providers and responders of energy prices in multiple subjects, taking DSO and GSO as the two leaders of the master-slave game, and taking VPPO as the only follower of the game. Among them, the distribution network and the gas distribution network optimize the optimal power flow and carbon emission flow according to the energy purchase demand and internal optimization results of the VPP, and transmit the energy-carbon comprehensive price and node carbon potential obtained by the solution to the VPP, and the VPP formulates the latest internal optimization scheduling strategy and energy purchase demand according to the information transmitted by the distribution network and the gas distribution network.
[0005] In combination with the characteristics of the research problem, the present invention introduces the adaptive alternating direction multiplier method, and proposes a Stackelberg game model for comprehensive energy-carbon pricing taking into account the participation of VPP in the electricity-gas multi-energy coupling network. Considering the complex internal information of each subject and the large transaction volume, the traditional centralized optimization algorithm can no longer meet the information privacy requirements of each trading subject under this model. Therefore, it is proposed to use an adaptive ADMM algorithm to perform distributed solution to the proposed multi-subject master-slave game model.
[0006] Firstly, the distribution system operator and the gas distribution system operator are regarded as leaders, and the virtual power plant operator is regarded as a follower, and a multi-agent master-slave game trading model considering VPP participating in the electricity-gas multi-energy system is constructed. Secondly, an energy-carbon comprehensive pricing method is formulated based on the carbon emission flow theory to encourage VPP to actively adjust its energy consumption strategy and trading strategy with the multi-energy system to realize multi-agent interactive trading; based on the previous analysis, considering the complex internal information of each subject and the large transaction volume, in order to meet the information privacy requirements of each trading subject, the adaptive ADMM algorithm is used to distribute the proposed multi-agent master-slave game model. Finally, the effectiveness of the proposed model and method is verified through examples. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] Figure 1 A two-tier framework for electricity-gas multi-energy system and VPP transactions;
[0008] Figure 2 Run the framework for VPP;
[0009] Figure 3 This is the topology diagram of the multi-energy VPP E33-G7 test system;
[0010] Figure 4 For wind power forecasting in distribution networks;
[0011] Figure 5It is the basic data for prediction within VPP;
[0012] Figure 6 Electricity prices in different scenarios;
[0013] Figure 7 Carbon potential variation diagram of distribution network nodes under Case 4;
[0014] Figure 8 The amount of electricity and gas purchased by VPP in different scenarios;
[0015] Fig. 9 Comparison chart of VPP energy purchase demand and carbon emissions under different scenarios;
[0016] Fig.10 VPP internal optimization result diagram;
[0017] Fig.11 Iterative convergence of multi-agent games;
[0018] Fig.12 This is the algorithm flow chart. DETAILED DESCRIPTION
[0019] For the convenience of explanation, the specific implementation modes of the present invention are further described according to embodiments.
[0020] The present invention provides a method for constructing and solving a master-slave game trading model of VPP participating in a multi-energy system under an energy-carbon comprehensive price response mechanism taking into account multi-agent interactive transactions based on an adaptive step-size alternating direction multiplier method. The specific steps are as follows:
[0021] S1: First, a two-layer model of multi-energy system-VPP collaborative optimization considering the energy-carbon comprehensive price is established to achieve the overall optimal operation of the distribution network, gas distribution network and VPP. Among them, the transaction behavior between the distribution system operator, the gas distribution system operator and the virtual power plant operator constitutes a multi-agent master-slave game behavior. This paper assumes that DSO and GSO are the two leaders of the master-slave game and regard VPPO as the only follower of the game. Its two-layer transaction framework is as follows Figure 1 shown.
[0022] S2: According to the optimization objectives and their operation constraints, the energy flow model and carbon emission flow distribution model of the distribution network and gas distribution network are established respectively. Among them, the optimal power flow model of the distribution network and gas distribution network is relaxed by the second order cone. In addition, DSO and GSO calculate the optimal power flow of the distribution network and gas distribution network according to the energy purchase demand of VPP to minimize their respective total operating costs, and solve the carbon flow distribution in combination with the distribution network power flow distribution, so as to formulate the energy-carbon comprehensive price, as follows:
[0023] Distribution Network Modeling:
[0024] The present invention adopts formulas (1)-(11) to establish a distribution network optimization operation model based on Distflow.
[0025] DSO takes minimizing the distribution network operation cost as the objective function and establishes the optimal power flow model of the distribution network including VPP energy purchase demand. The power flow should meet the following power balance constraints, line power limit constraints, SOC relaxation constraints, node voltage constraints, and generator active and reactive output constraints:
[0026]
[0027] Where: g∈n, w∈n represent the coal-fired unit g and fan w connected to node n respectively; b(n,·)∈n represents the line from node n to other nodes, denoted as b; l(·,n)∈n represents the branch l from other nodes to node n; N DN is the node set of the distribution network; is the active power output of wind turbine w at time t; is the power purchased by VPP from the distribution network at time t; are the active and reactive loads connected to node n at time t respectively; are the active and reactive powers flowing through lines b and l at time t respectively; a l,t represents the square of the current in branch l at time t; r l 、x l are the resistance and reactance of branch l respectively; is the dual variable corresponding to the active power balance constraint; Q g,t is the reactive power output of coal-fired unit g at time t; is the reactive power purchased by the distribution network from the upper grid at time t; η represents the power factor of the load; U n,t is the square of the voltage amplitude at node n at time t; S l is the upper limit of the apparent power of line l; U min , U max are the lower and upper limits of the node voltage amplitude respectively; P g,min , P g,max , Q g,min , Q g,max They are respectively the lower and upper limits of the active and reactive output of unit g.
[0028] The goal of DSO is to make the total operating cost of the distribution network f DSO Minimum, including the cost of coal-fired units and the cost of purchasing electricity from the upper grid
[0029]
[0030] Where: T is the total number of dispatching periods; G is the set of coal-fired units in the distribution network; a g , b g 、c g are the power generation cost coefficients of coal-fired unit g; P g,t is the active power output of unit g at time t; is the electricity selling price of the upper power grid at time t; It is the active power purchased by DSO from the upper grid at time t.
[0031] Gas distribution network modeling:
[0032] The present invention aims to minimize the operating cost of the gas distribution network and constructs a gas distribution network pricing model based on second-order cone programming. The gas distribution network consists of gas sources, natural gas pipelines, air compressors and gas loads. It is assumed that the flow direction of natural gas in the natural gas pipeline is determined. The goal of GSO is to make the operating cost of the gas distribution network GSO Minimum, that is, minimize the cost of natural gas purchased by GSO from natural gas companies. The present invention uses formula (12) to represent the natural gas node flow balance constraint, formula (13) defines the pipeline natural gas flow Weymouth equation, formula (14) represents the natural gas network pipeline flow restriction, formula (15)-(16) define the natural gas node pressure restriction, formula (17) represents the gas well supply constraint, formula (18) establishes the gas distribution network operation cost function, and the established gas distribution network optimal power flow mathematical model considering the second-order cone constraint is as follows:
[0033]
[0034]
[0035] Where: s∈j represents the natural gas source s connected to node j; z(j) is the pipeline set of the end node j; v(j) is the pipeline set of the head node j; N GN is the node set of the gas distribution network; w ij,t 、w jk,t are the amount of natural gas flowing through pipelines ij and jk at time t respectively; is the gas load at node j of the gas distribution network at time t; is the amount of natural gas purchased by VPP from the gas distribution network at time t; is the dual variable corresponding to the natural gas node flow balance constraint; π i,t , π j,t are the pressures of nodes i and j at time t; C ij is the pipeline transmission characteristic parameter; They are the upper and lower limits of the amount of natural gas transmitted through pipelines; are the upper and lower limits of the air pressure at node j, respectively; k is the air compressor coefficient; are the upper and lower limits of the output of natural gas source s, respectively; are the gas price and gas volume purchased by GSO from gas source s at time t; N s A collection of natural gas sources.
[0036] Carbon emission flow modeling:
[0037] The research focus of the present invention is how to price carbon emissions from power generation to end users so that the correct incentives can be provided between the electricity-gas multi-energy system. The carbon emission flow (CEF) theory is established based on the system energy flow, which can intuitively characterize the flow direction of carbon emissions during system operation and provide a new analytical perspective for low-carbon economic scheduling. In the CEF model, the carbon flow index is usually used to describe the carbon emission allocation. The present invention mainly needs to obtain the carbon potential of each node in the electricity-gas multi-energy system as a carbon signal, and establish a connection between the carbon price and the carbon potential of the node connected to the VPP through a carbon tax. The CEF model is established as follows.
[0038] 1) Distribution network CEF model
[0039] The node carbon potential represents the carbon emission value equivalent to the power generation side caused by the unit electricity consumption at the node. The calculation formula is as follows:
[0040]
[0041] Where: ρ l,t They represent the carbon potential of node n and the carbon flow density of branch l at time t respectively; g is the carbon emission intensity of coal-fired unit g.
[0042] The branch carbon flow density represents the carbon emission value equivalent to the power generation side caused by the branch transmission of unit electricity. According to the proportional sharing principle, the carbon flow density of all branches flowing out of node n is equal to the carbon potential of the node, and its expression is as follows:
[0043]
[0044] 2) Gas distribution network CEF model
[0045] The CEF model of the gas distribution network is similar to that of the distribution network. The calculation formulas for the carbon potential of each node and the carbon flow density of the branch are as follows:
[0046]
[0047] Where: is the carbon potential of gas grid node j at time t; ρ ij,t is the carbon flow density of the pipeline connected to node j at time t when there is airflow flowing into pipeline ij; is the carbon emission intensity of the gas source s connected to node j.
[0048] 3) Energy-carbon integrated pricing method based on CEF theory
[0049] As a consumer, VPP needs to bear certain carbon emission responsibilities when purchasing electricity and gas from the distribution network and gas distribution network. After calculating the node carbon potential of the distribution network and gas distribution network using the CEF model, the connection between carbon price and node carbon potential is established through carbon tax, and an electricity / gas energy-carbon comprehensive pricing method combining node marginal electricity price / node marginal gas price with carbon price is formulated, as follows:
[0050]
[0051]
[0052] Where: is the electricity-carbon comprehensive price of distribution network node n at time t; is the gas-carbon comprehensive price of node j in the gas distribution network at time t; τ is the carbon tax.
[0053] S3: The VPP operation framework established by the present invention is as follows Figure 2 As shown. These include gas turbine CHP units, gas boilers, distributed wind power and energy storage units. Flexible loads include electrical loads and thermal loads, and the response to electrical and thermal load demand is considered to guide users to make reasonable adjustments to their energy demand. VPP exchanges electrical power with the distribution network through interconnection lines, and transmits natural gas with the gas distribution network through natural gas pipelines.
[0054] VPP objective function:
[0055] The goal of the VPP energy optimization model is the total operating cost f VPPO Minimum:
[0056]
[0057]
[0058] Where: The electricity purchase cost for the interaction between VPP and distribution grid; The cost of natural gas purchased from the distribution grid for the VPP; is the comprehensive demand response cost for electricity and heat within the VPP; They represent the load transfer and load reduction in VPP at time t respectively; represents the reduction of heat load in VPP at time t; are the compensation unit prices for electricity load reduction and transfer respectively; The unit price for compensation of heat load reduction.
[0059] VPP constraint model:
[0060] 1) Power balance constraints
[0061] The power balance constraints of various energy flows within the VPP are as follows:
[0062]
[0063] Where: are the power supply power and heat generation power of the gas turbine at time t respectively; is the actual output of renewable energy at time t; are the charging and discharging power of the energy storage at time t respectively; They represent the loads after the electricity and heat load demands are responded to at time t respectively; is the heating power of the gas boiler at time t; are the amounts of natural gas input to the gas turbine and gas boiler at time t respectively.
[0064] 2) Gas turbine CHP unit constraints
[0065]
[0066] Where: are the power supply efficiency and heat production efficiency of the gas turbine respectively; is the calorific value of natural gas per unit volume; Q EH The heat energy converted into unit electrical energy; They are the upper and lower limits of the gas turbine power supply respectively.
[0067] 3) GB output constraints for gas boilers
[0068]
[0069] Where: η GB The electric heat conversion efficiency of the gas boiler; It is the upper and lower limits of the thermal output of the gas boiler in the VPP.
[0070] 4) Energy storage unit constraints
[0071]
[0072] Where: S t is the capacity of the energy storage device in VPP at time t; η loss , η ES,cha , η ES,dis are the energy loss coefficient, energy charging and discharging efficiency of the energy storage device respectively; loss <<1; are the maximum charging and discharging power of the energy storage device respectively; S min , S max are the minimum and maximum storage capacities of the energy storage device, respectively; are binary variables, representing the charging and discharging states of the energy storage device at time t.
[0073] 5) Constraints on comprehensive electricity and heat demand response
[0074]
[0075] Where: represents the initial load of the electrical load in the VPP at time t; It indicates the upper limit of the amount of electrical load that can be transferred; Indicates the upper limit of the amount of electrical load that can be reduced; Indicates the initial load of heat load at time t; Indicates the maximum amount of heat load that can be reduced.
[0076] S4. In summary, considering the master-slave game relationship between the leader DSO, GSO and the follower VPP in the multi-agent master-slave game transaction model, each participant optimizes its own operating state based on its own interest goals. The present invention uses the adaptive ADMM algorithm to perform distributed solution on the proposed multi-agent master-slave game model. The details are as follows:
[0077] Master-slave game transaction model:
[0078] The game model includes three elements: participant set, strategy set and benefit set. The established master-slave game model is as follows:
[0079]
[0080] 1) Participant set: DSO, GSO, VPPO represent the set of all participants, where DSO and GSO are leaders and VPPO is a follower.
[0081] 2) Strategy set: The strategy of the leader DSO is the amount of electricity purchased from each distributed power generator and the upper grid at each moment, as well as the electricity-carbon comprehensive price set, expressed as The strategy of the leader GSO is to purchase natural gas from each gas distribution station and set a gas-carbon comprehensive price, expressed as The strategy of the follower VPPO is to purchase electricity from DSO and gas from GSO, expressed as
[0082] 3) Benefit: The benefit of each participant is the objective function, which can be expressed as f DSO 、f GSO and f VPPO .
[0083] Distributed solution of the master-slave game trading model based on the adaptive ADMM algorithm:
[0084] Based on the principle of ADMM algorithm, the auxiliary conditions are introduced as shown in the following formula:
[0085]
[0086] Where: are the electric energy that VPPO expects to trade with DSO and the electric energy that DSO expects to trade with VPPO at time t, respectively; They are the amount of natural gas that VPPO expects to purchase from GSO and the amount of natural gas that GSO expects to sell to VPPO at time t, respectively.
[0087] Based on the ADMM principle, the distributed models of optimal pricing of distribution network, gas distribution network and VPP energy optimization are obtained respectively:
[0088] 1) Distributed Optimization Model for Optimal Pricing of Distribution Network
[0089]
[0090] Where: and are the Lagrange multiplier and penalty factor corresponding to DSO respectively.
[0091] 2) Distributed Optimization Model for Optimal Pricing of Gas Distribution Network
[0092]
[0093] 3) Distributed Model of VPP Energy Optimization
[0094]
[0095] Where: are the corresponding Lagrange multipliers when VPPO expects to trade with DSO and GSO respectively.
[0096] Distributed optimization solution process:
[0097] Step 1: Initialization: Initialize parameters; record the number of iterations k = 1 and start iteration;
[0098] Step 2: The coupling variables and Lagrange multipliers are updated as follows:
[0099]
[0100] Where: z represents the type of energy; k is the number of distributed solution iterations.
[0101] Step 3: Calculation method of original residual and dual residual:
[0102]
[0103] Where: are the primal residual and dual residual in the k+1th iteration respectively.
[0104] Step 4: Determine convergence. Determine whether the following conditions are met:
[0105]
[0106] Where: ε pri , ε dual are the convergence thresholds of the primal and dual residuals, respectively.
[0107] If satisfied, the iteration converges; otherwise, update k=k+1 and return to step 2.
[0108] Implementation of simulation examples
[0109] The present invention is further described below with an example.
[0110] Figure 3 This is a topological diagram of the IEEE 33-node distribution network, 7-node gas distribution network and 1 VPP coupling network used in the implementation of the present invention. In the figure: W is a wind turbine installed at node 31 of the distribution network; G1~G5 represent 5 coal-fired units, which are installed at nodes 3, 9, 29, 14 and 21 of the distribution network respectively. Their operating parameters are shown in Table 1 and Figure 4 W1 and W2 represent gas distribution stations, which are connected to gas distribution network nodes 6 and 7. The forecast curves of renewable energy output and load within VPP are shown in Figure 5 The parameters of each unit are shown in Table 2.
[0111] Assume that the power factor η of the node load in the distribution network is 0.85. The price of electricity purchased by the distribution network from the upper power grid is set at 160$ / MWh, and the price of gas purchased by GSO from the gas source s is set at 0.52$ / m 3 , the carbon tax is set to 45$ / tCO2. In the adaptive step size ADMM algorithm, the initial step size ρ is set to 1, μ is set to 10, τ incr and τ decr Set to 2, the convergence threshold ε pri , ε dual Both are 10 -3 This paper builds a simulation model based on the Matlab 2018b platform and uses the Gurobi solver to solve it.
[0112] Table 1 Parameters of coal-fired units in the distribution network
[0113]
[0114] Table 2 VPP related parameters
[0115]
[0116] In order to verify the impact of multi-agent master-slave game on electricity-natural gas transactions and the energy purchase costs of each entity under the energy-carbon comprehensive price response mechanism, the following four energy settlement scenarios are set up.
[0117] Case 1: The settlement electricity price adopts a fixed time-of-use electricity price, and the settlement gas price adopts a fixed natural gas price. Carbon emission flows are not considered. VPPO conducts electricity-natural gas transactions as a complete price receiver, and the flexible load in the VPP is not considered for optimal scheduling.
[0118] Case 2: Based on Case 1, the flexible load in the VPP considers the comprehensive demand response of electricity and heat.
[0119] Case 3: The settlement electricity price of the distribution network and the settlement gas price of the gas distribution network adopt the node marginal electricity price (LMEP) and node marginal gas price (LMGP) obtained after the game equilibrium of each subject, respectively. Carbon emission flow is not considered, and the flexible load in the VPP is considered to be optimally dispatched;
[0120] Case 4: The settlement electricity price of the distribution network and the settlement gas price of the gas distribution network are respectively the electricity-carbon comprehensive price and gas-carbon comprehensive price obtained after the game equilibrium of each subject, taking into account the carbon emission flow and the optimal scheduling of the flexible load in the VPP, which is the model proposed in this paper.
[0121] The simulation results of the above four scenarios are shown in Table 3.
[0122] Table 3 Comparison of operating costs and carbon emissions of each entity under different scenarios
[0123]
[0124] As shown in Table 3, compared with Case 1 without considering the comprehensive demand response of electricity and heat, the operating costs of each subject in Cases 2, 3 and 4 considering the comprehensive demand response are reduced, and the carbon emissions of VPP are effectively reduced. The demand response of flexible loads reduces the dependence of VPP on the distribution network and gas distribution network, and reduces the amount of electricity and gas purchased. Compared with the fixed energy settlement price in Case 2, it can be seen that the total energy purchase cost and carbon emissions of multiple subjects in Case 3 considering the master-slave game interaction between VPPO and DSO and GSO are reduced, which are reduced by 400.9$ and 3.25tCO2 respectively. It shows that the way of settling energy prices by LMEP and LMGP is conducive to guiding VPP energy optimization more economically, and effectively improves the economy and low carbon of VPP.
[0125] By comparing Case 3 and Case 4, it can be seen that compared with the single LMEP and LMGP, after considering the energy-carbon comprehensive price, Case 4 takes the flexible load in the VPP into consideration, responds to the demand according to the energy-carbon comprehensive price, reduces the purchase of energy during the period of high carbon potential at the node, and the VPP carbon emissions are reduced by 7.29% compared with Case 3. However, due to the increase in carbon price in the energy-carbon comprehensive pricing method, the VPP energy purchase cost and total cost of Case 4 are increased compared with Case 3. In addition, since the carbon emission density of electricity in the distribution network is higher than that of natural gas in the gas distribution network, the inclusion of carbon prices will encourage VPP to use more natural gas and reduce the demand for electricity purchase, that is, compared with Case 3, the GSO operating cost of Case 4 is increased and the DSO operating cost is reduced. It can be seen that the proposed multi-agent master-slave game trading strategy is conducive to improving the economy and low-carbon benefits of each subject.
[0126] In order to analyze the impact of different pricing methods on VPP power purchase, Figure 6 The distribution network settlement electricity price and carbon price change curves of the VPP node in Case 3 and 4 are given. Figure 7 The carbon potential of all nodes in the distribution network under Case 4 is given.
[0127] Depend on Figure 6 It can be seen that considering the carbon emission tax on the electricity purchase demand of VPP from the distribution network will increase the electricity purchase price. The electricity-carbon comprehensive price is always higher than LMEP, and its increment varies with time. At certain times (such as 05:00-09:00), when the carbon emission density of the node is large, the difference between the electricity-carbon comprehensive price and LMEP is also large. In contrast, the natural gas-carbon comprehensive price is also rising, but its changes are small. The carbon potential of each node in the gas distribution network is basically the same, and the carbon emissions taken into account by the VPP when purchasing gas from the gas distribution network only depend on the gas consumption. Therefore, the natural gas-carbon comprehensive price basically does not change with time. The LMGP of the VPP and the gas distribution network coupling node calculated by the present invention is 0.5120$ / m 3 The combined price of natural gas and carbon with carbon tax is 0.6174$ / m 3 .
[0128] pass Figure 7 By comparing the carbon potential of each node with the IEEE 33-node system topology, it can be seen that the carbon potential of the nodes equipped with wind turbines and nearby nodes is relatively low, because wind turbines are close to zero carbon emissions; while the nodes where the VPP is coupled to the distribution network are close to coal-fired units with high carbon emissions, and their node carbon potential is affected by thermal power units, but the carbon emission intensity of each thermal power unit is relatively close, so Figure 6 As shown, the carbon prices at the coupling nodes of the VPP and the distribution network mostly change slightly at all times.
[0129] In order to further explore the impact of carbon price on the specific electricity and gas purchases of VPP, Figure 8 The comparison results of the electricity and gas purchased at each moment of Case 3 and 4 VPP are given.
[0130] Depend on Figure 8 It can be seen that during the period from 05:00 to 10:00, the VPP purchased less electricity in Case 4 considering the carbon price than in Case 3 not considering the carbon price. At this time, the VPP purchased more natural gas because the carbon potential of the coupling node between the VPP and the distribution network is higher. Figure 6 From the carbon price change curve, we can see that the carbon potential of the node is higher during the 05:00-10:00 period, while the carbon potential of the node coupled with the gas distribution network is lower, that is, the carbon density of electricity is higher than that of natural gas. Therefore, considering the carbon price will incentivize VPP to use more natural gas and reduce the amount of electricity purchased.
[0131] In order to more intuitively see the energy purchase demand and carbon emissions of VPP under the energy-carbon integrated price response mechanism, a comparative analysis is conducted on Case 3 and Case 4. The comparison results are as follows: Fig. 9 shown.
[0132] Depend on Fig. 9 It can be seen intuitively that compared with Case 3 without considering the carbon price, natural gas becomes competitive under Case 4 with carbon price, and VPP is more inclined to purchase natural gas with low carbon potential, that is, the total gas purchase volume has increased, thereby reducing the purchase of electricity from the distribution network with high carbon potential, and the total electricity purchase volume of VPP has decreased by 1.91MWh. In addition, after considering the carbon price, the total carbon emissions of VPP decreased from 29585.8kg to 27436.6kg. Based on the energy-carbon comprehensive price response mechanism, VPP has greatly reduced carbon emissions. This result shows that the proposed multi-agent master-slave game energy settlement price can effectively promote carbon emission reduction and improve low-carbon benefits.
[0133] The optimization results of VPP internal electrical power and thermal power are as follows Fig.10As shown. For the internal units of VPP, the consumption of new energy has the highest priority, and the phenomenon of wind abandonment is minimized. VPP completes the internal optimal scheduling according to the comprehensive price of energy-carbon. The gas turbine CHP unit generates electricity and heat within the output range. The electric energy storage is mainly charged when the comprehensive price of energy-carbon is low, such as 01:00-04:00 and 07:00, and discharged at 06:00 and 12:00-14:00 when the comprehensive price of electricity-carbon is high, thereby reducing the purchase of electricity from the distribution network, reducing the total operating cost and carbon emissions. At the same time, the comprehensive demand response of electricity and heat loads can achieve peak shaving and valley filling, and alleviate the peak load regulation pressure of the power grid; only when the output of the internal units cannot meet their load requirements, they will purchase electricity from the distribution network and gas from the gas distribution network. Taking into account that the heat load demand in the VPP is met first by the gas turbine CHP unit, after considering the heat load demand response, most of the time it is heated by the gas turbine only, and only when the gas turbine is insufficient to supply heat from 23:00 to 24:00, the gas boiler GB unit will produce power.
[0134] The present invention analyzes the iterative convergence of the proposed multi-agent master-slave game transaction strategy. Fig.11 (a) shows the convergence iteration of the primal residual and the dual residual in Case 4. Fig.11 (b) Plot the game interaction iterative convergence of DSO, GSO and VPPO.
[0135] according to Fig.11 (a) Residual convergence It can be seen that the proposed algorithm meets the convergence accuracy requirement after 45 iterations and converges to 10 -3 The calculation time is 329s. Fig.11 (b) It can be seen intuitively that the energy purchase costs of the leaders DSO and GSO converge to $4776.7 and $18079.3 respectively, and the energy purchase cost of the follower VPPO converges to $13009.4. At this time, the master-slave game between the leader DSO, GSO and the follower VPPO reaches the Stackelberg equilibrium, and each subject cannot reduce its own operating cost by adjusting its own trading strategy alone. Therefore, it shows that the distributed optimization algorithm proposed in this paper has good convergence performance and computational efficiency.
[0136] In order to further verify the effectiveness of the adaptive ADMM algorithm proposed in this paper, a comparative analysis is performed with the fixed step size ADMM. Table 4 lists the solution performance of the fixed step size and adaptive step size ADMM. As shown in Table 4, compared with the traditional ADMM, the adaptive step size ADMM reduces the dependence on the initial value selection by correcting the step size, thereby reducing the number of iterations and solution time, and improving the solution efficiency.
[0137] Table 4 Comparison between traditional ADMM and adaptive step size ADMM
[0138]
[0139] The above is the content of the present invention. Any changes made according to the technical solution of the present invention, as long as the functional effects produced do not exceed the scope of the technical solution of the present invention, belong to the protection scope of the present invention.
Claims
1. A method for constructing and solving a multi-agent master-slave game transaction model for VPPs participating in a multi-energy system under an energy-carbon integrated price response mechanism, characterized in that: The following steps are involved: S1: From the perspective of multi-agent master-slave game, a two-layer master-slave game transaction model considering VPP participation in the electricity-gas multi-energy system is constructed to achieve the coordinated operation of the distribution network, gas distribution network and VPP; S2: According to the optimization objectives and their operation constraints, energy flow models and carbon emission flow distribution models of the distribution network and gas distribution network are established respectively; S3: Based on the distribution of energy and carbon flows, the coupling between the distribution network, gas distribution network and VPP is established, and an energy-carbon integrated pricing strategy model, which is essentially a Stackelberg game problem, is proposed to guide VPP to actively adjust its energy consumption strategy and trading strategy with multiple energy systems to achieve the goal of carbon emission reduction; S4: Taking the privacy issues of each subject into consideration, the adaptive step-size alternating direction multiplier method is used to perform a distributed solution to the transaction problem between multiple subjects in step 3, and the final solution is the equilibrium value of the multi-agent master-slave game.
2. The method for constructing and solving a two-layer model of multi-agent master-slave game transactions of VPPs participating in multi-energy systems under the energy-carbon comprehensive price response mechanism according to claim 1 is characterized in that: The distribution system operator (DSO) and the gas system operator (GSO) are the two leaders of the master-slave game, and the virtual power plant operator (VPPO) is the only follower of the game. The game is repeated until the transactions of all entities in the two-layer model are agreed upon.
3. The method for constructing and solving a multi-agent master-slave game transaction model for VPP participation in a multi-energy system according to claim 1 is characterized in that: The optimal power flow model of the distribution network is specifically as follows: 1) The objective function of DSO is to make the total operating cost of the distribution network DSO Minimum, including the cost of coal-fired units and the cost of purchasing electricity from the upper grid Where: T is the total number of dispatching periods; G is the set of coal-fired units in the distribution network; a g 、b g 、c g are the power generation cost coefficients of coal-fired unit g; P g,t is the active power output of unit g at time t; P is the electricity selling price of the upper power grid at time t; t grid It is the active power purchased by DSO from the upper grid at time t. 2) The constraints of the model include power balance constraint, line power limit constraint, SOC relaxation constraint, node voltage constraint, generator active and reactive output constraint, and the following equation is obtained: Where: g∈n, w∈n represent the coal-fired unit g and fan w connected to node n respectively; b(n,·)∈n represents the line from node n to other nodes, denoted as b; l(·,n)∈n represents the branch l from other nodes to node n; N DN is the node set of the distribution network; is the active power output of wind turbine w at time t; P t buy is the power purchased by VPP from the distribution network at time t; are the active and reactive loads connected to node n at time t respectively; are the active and reactive powers flowing through lines b and l at time t respectively; a l,t represents the square of the current in branch l at time t; r l 、x l are the resistance and reactance of branch l respectively; is the dual variable corresponding to the active power balance constraint; Q g,t is the reactive power output of coal-fired unit g at time t; is the reactive power purchased by the distribution network from the upper grid at time t; η represents the power factor of the load; U n,t is the square of the voltage amplitude at node n at time t; S l is the upper limit of the apparent power of line l; U min , U max are the lower and upper limits of the node voltage amplitude respectively; P g,min , P g,max , Q g,min , Q g,max They are respectively the lower and upper limits of the active and reactive output of unit g.
4. The method for constructing and solving a multi-agent master-slave game transaction model for VPPs participating in a multi-energy system based on an energy-carbon integrated price response mechanism according to claim 1 is characterized in that: The optimal power flow model of the gas distribution network is as follows: 1) The gas distribution network consists of gas sources, natural gas pipelines, air compressors and gas loads. The present invention assumes that the natural gas flow direction in the natural gas pipeline is determined. The goal of GSO is to reduce the operating cost of the gas distribution network to GSO Minimum, that is, minimize the cost of natural gas purchased by GSO from natural gas companies. The specific expression is: Where: are the gas price and gas volume purchased by GSO from gas source s at time t; N s A collection of natural gas sources. 2) The gas distribution network needs to meet the corresponding network constraints, including natural gas node flow balance constraints, gas network pipeline flow constraints, node pressure constraints and natural gas source output constraints, as follows: π j,t ≤kπ i,t Where: s∈j represents the natural gas source s connected to node j; z(j) is the pipeline set of the end node j; v(j) is the pipeline set of the head node j; N GN is the node set of the gas distribution network; w ij,t 、w jk,t are the amount of natural gas flowing through pipelines ij and jk at time t respectively; is the gas load at node j of the gas distribution network at time t; is the amount of natural gas purchased by VPP from the gas distribution network at time t; is the dual variable corresponding to the natural gas node flow balance constraint; π i,t , π j,t are the pressures of nodes i and j at time t; C ij is the pipeline transmission characteristic parameter; They are the upper and lower limits of the amount of natural gas transmitted through pipelines; are the upper and lower limits of the air pressure at node j, respectively; k is the air compressor coefficient; are the upper and lower limits of the output of natural gas source s respectively.
5. The method for constructing and solving a multi-agent master-slave game transaction model for VPPs participating in a multi-energy system based on an energy-carbon integrated price response mechanism according to claim 1 is characterized in that: The multi-energy system energy-carbon pricing model based on carbon emission flow theory is as follows: 1) Carbon dioxide in the energy industry is mostly produced on the source side, while the ultimate driving force of carbon emissions is on the load side. How to price carbon emissions from power generation to end users so that the correct incentives can be provided between the electricity-gas multi-energy system is the focus of this invention. The carbon emission flow (CEF) theory is established based on the system energy flow, which can intuitively characterize the flow direction of carbon emissions during system operation, and provide a new analytical perspective for low-carbon economic scheduling. In the CEF model, the carbon flow index is usually used to describe the carbon emission allocation. The research of this invention mainly needs to obtain the carbon potential of each node in the electricity-gas multi-energy system as a carbon signal, and establish a connection between the carbon price and the carbon potential of the node to which the VPP is connected through carbon tax. The CEF model of its distribution network is established as follows: Where: ρ l,t They represent the carbon potential of node n and the carbon flow density of branch l at time t respectively; g is the carbon emission intensity of coal-fired unit g. 2) The CEF model of the gas distribution network is similar to that of the distribution network. The calculation formulas for the carbon potential of each node and the carbon flow density of the branch are as follows: Where: is the carbon potential of gas grid node j at time t; ρ ij,t is the carbon flow density of the pipeline connected to node j at time t when there is airflow flowing into pipeline ij; is the carbon emission intensity of the gas source s connected to node j. 3) As a consumer, VPP needs to bear certain carbon emission responsibilities when purchasing electricity and gas from the distribution network and gas distribution network. After calculating the node carbon potential of the distribution network and gas distribution network using the CEF model, the connection between carbon price and node carbon potential is established through carbon tax, and an electricity / gas energy-carbon comprehensive pricing method combining node marginal price and carbon price is formulated, as follows: Where: is the electricity-carbon comprehensive price of distribution network node n at time t; is the gas-carbon comprehensive price of node j in the gas distribution network at time t; τ is the carbon tax. The above-mentioned energy-carbon integrated pricing method encourages VPP to actively adjust its energy purchasing demand and internal optimal scheduling strategy, thereby reducing its carbon emissions and operating costs.
6. The method for constructing and solving a multi-agent master-slave game transaction model for VPPs participating in a multi-energy system based on an energy-carbon integrated price response mechanism according to claim 1 is characterized in that: The follower VPP energy optimization model is as follows: 1) The goal of the VPP energy optimization model is the total operating cost f VPPO Minimum: Where: The electricity purchase cost for the interaction between VPP and distribution grid; The cost of natural gas purchased from the distribution grid for the VPP; P is the comprehensive demand response cost of electricity and heat in VPP; t tran , P t cut They represent the load transfer and load reduction in VPP at time t respectively; represents the reduction of heat load in VPP at time t; are the compensation unit prices for electricity load reduction and transfer respectively; The unit price for compensation of heat load reduction. 2) The constraints of the model include power balance constraints, gas unit output constraints, gas boiler output constraints, energy storage unit constraints, and electric and thermal comprehensive demand response constraints. The specific expressions are as follows: (1) Power balance constraints The power balance constraints of various energy flows within the VPP are as follows: Where: P t CHP , are the power supply power and heat generation power of the gas turbine at time t respectively; P t Wind is the actual output of renewable energy at time t; P t ES,cha , P t ES,dis are the charging and discharging power of the energy storage at time t; P t load , They represent the loads after the electricity and heat load demands are responded to at time t respectively; is the heating power of the gas boiler at time t; are the amounts of natural gas input to the gas turbine and gas boiler at time t respectively. (2) Constraints of gas turbine CHP units The mathematical model of power supply and heat supply of CHP unit is as follows: Where: are the power supply efficiency and heat production efficiency of the gas turbine respectively; is the calorific value of natural gas per unit volume; Q EH The heat energy converted into unit electrical energy; They are the upper and lower limits of the gas turbine power supply respectively. (3) Gas boiler GB output constraints Where: η GB The electric heat conversion efficiency of the gas boiler; They are the upper and lower limits of the thermal output of the gas boiler in the VPP respectively. (4) Energy storage unit constraints The introduction of energy storage devices can further improve the operational flexibility of VPP. The energy storage device is modeled as follows: Where: S t is the capacity of the energy storage device in VPP at time t; η loss , η ES,cha , η ES,dis are the energy loss coefficient, energy charging and discharging efficiency of the energy storage device respectively; loss <<1; are the maximum charging and discharging power of the energy storage device respectively; S min , S max are the minimum and maximum storage capacities of the energy storage device, respectively; are binary variables, representing the charging and discharging states of the energy storage device at time t. (5) Constraints on comprehensive electricity and heat demand response Where: represents the initial load of the electrical load in the VPP at time t; It indicates the upper limit of the amount of electrical load that can be transferred; Indicates the upper limit of the amount of electrical load that can be reduced; Indicates the initial load of heat load at time t; Indicates the maximum amount of heat load that can be reduced.
7. The method for constructing and solving a multi-agent master-slave game transaction model for VPPs participating in a multi-energy system based on an energy-carbon integrated price response mechanism according to claim 1 is characterized in that: The multi-agent master-slave game model is solved as follows. 1) Considering the master-slave game relationship between the leader DSO, GSO and the follower VPP in the multi-agent master-slave game transaction model of VPP participating in the electricity-gas multi-energy system, each participant optimizes its own operating status based on its own interests and formulates corresponding energy trading strategies. Due to the complex internal information of each subject and the large transaction volume, the traditional centralized optimization algorithm can no longer meet the information privacy requirements of each trading subject under this model. Therefore, the adaptive ADMM algorithm is used to perform distributed solution on the proposed multi-agent master-slave game model. The master-slave game transaction model is: (1) Participant set: DSO, GSO, and VPPO represent the set of all participants, where DSO and GSO are leaders and VPPO is a follower. (2) Strategy set: The strategy of the leader DSO is the amount of electricity purchased from each distributed power generator and the upper-level power grid at each moment, as well as the electricity-carbon comprehensive price set, expressed as P g,t , The strategy of the leader GSO is to purchase natural gas from each gas distribution station and set a gas-carbon comprehensive price, expressed as The strategy of the follower VPPO is to purchase electricity from DSO and gas from GSO, expressed as P t buy , (3) Benefit: The benefit of each participant is the objective function, which can be expressed as f DSO 、f GSO and f VPPO . 2) The present invention adopts the adaptive ADMM algorithm to perform distributed solution to the proposed multi-agent master-slave game model, which can ensure that the interaction between the participating entities does not involve the privacy of their respective internal information. Only the boundary information of each entity during the transaction is required, thereby achieving distributed and efficient solution while protecting the privacy of the information of each transaction entity. Based on the principle of ADMM algorithm, the auxiliary conditions are introduced as shown in the following formula: Where: P t buy , P t ex are the electric energy that VPPO expects to trade with DSO and the electric energy that DSO expects to trade with VPPO at time t, respectively; They are the amount of natural gas that VPPO expects to purchase from GSO and the amount of natural gas that GSO expects to sell to VPPO at time t, respectively. Based on the ADMM principle, the distributed models of optimal pricing for distribution network, optimal pricing for gas distribution network and energy optimization for VPP are obtained respectively. (1) Distributed Optimization Model for Optimal Pricing of Distribution Networks Where: and ρ are the Lagrange multiplier and penalty factor corresponding to DSO, respectively. (2) Distributed Optimization Model for Optimal Pricing of Gas Distribution Network Where: Lagrange multiplier for GSO distributed solution. (3) Distributed Model of VPP Energy Optimization Where: are the corresponding Lagrange multipliers when VPPO expects to trade with DSO and GSO respectively.
8. The method for constructing and solving a multi-agent master-slave game transaction model for VPPs participating in a multi-energy system based on an energy-carbon integrated price response mechanism according to claim 7 is characterized in that: The specific steps of the adaptive ADMM algorithm are as follows: (1) Initialization: Initialize parameters; record the number of iterations k = 1 and start iteration; (2) Repeat: ① The coupling variables and Lagrange multipliers are updated as follows: Where: z represents the type of energy; k is the number of distributed solution iterations. ②Calculation method of original residual and dual residual: Where: are the primal residual and dual residual in the k+1th iteration respectively. ③Judgment of convergence: If the following conditions are met: The iteration converges and outputs the final solution to the problem; otherwise, update k=k+1 and return to step ①; (3) Step size selection: Since the step size ρ has a direct impact on the ADMM solution speed, even improper step size may lead to non-convergence of the solution process. Therefore, the present invention adopts an adaptive ADMM algorithm to automatically update the step size through the relationship between the original residual and the dual residual, improve the algorithm convergence, and reduce the iteration time. The step size update mechanism is as follows: Where: μ is the proportional coefficient between the original residual and the dual residual; τ incr , τ decr They are the acceleration and deceleration factors of step size change respectively.
9. A multi-agent master-slave game transaction model construction and solution method system for VPP participating in multi-energy systems under the energy-carbon comprehensive price response mechanism, characterized in that: The method comprises a memory, a processor and computer program instructions stored in the memory and capable of being executed by the processor. When the processor executes the computer program instructions, the method steps as claimed in any one of claims 1 to 8 can be implemented.
10. A computer-readable storage medium having stored thereon computer program instructions that can be executed by a processor, and when the processor executes the computer program instructions, the method steps according to any one of claims 1 to 8 can be implemented.
Citation Information
Cited By
Multi-virtual power plant master-slave game collaborative optimization method considering carbon transaction
CN119205182A
Electric smelting magnesium load peak regulation control method based on price guidance and master-slave game
CN120341891A
A load peak shaving control method for electric smelting magnesium based on price guidance and master-slave game
CN120341891B
