A method and system for electric vehicle participation in demand response control

Abstract

The present application relates to the technical field of load demand response control, and particularly relates to a method and system for electric vehicles participating in demand response control, wherein the control method introduces a system response load total power deviation ΔP, improves a cost incremental rate model of a leading intelligent agent and a following intelligent agent, calculates a cost incremental rate η of the leading and following intelligent agents in the k+1th iteration according to the improved cost incremental rate model, calculates a response load index of each intelligent agent in the k+1th iteration from η, calculates a response load total power deviation by the leading intelligent agent, and if the deviation exceeds an allowable error range, enters a next iteration cycle, and if the response load total power deviation ΔP does not exceed the allowable error range, the iteration is ended. i[k+1] i[k+1] The present application can quickly eliminate the response load total power deviation, improve peak shaving and valley filling effect, and make the system dynamic control performance better and the response rate faster.​

Description

Technical Field

[0001] This invention relates to the field of load demand response control technology, and in particular to a method and system for electric vehicles participating in demand response control. Background Technology

[0002] With the rapid development of electric vehicle technology and charging infrastructure, and in order to achieve the "dual carbon" goal and reduce fossil fuel extraction, the number of electric vehicles will inevitably increase dramatically in the future. The operation of a large number of electric vehicle loads charging may cause multi-node load power disturbances, affecting the voltage stability of the common coupling point and impacting the coordinated control of hybrid microgrid groups. To address these issues, some technologies utilize household electricity loads such as air conditioners in demand response control based on discrete master-slave consensus algorithms, and establish a multi-agent energy management system.

[0003] However, the above algorithm cannot quickly eliminate the dynamic deviation of total power, resulting in unsatisfactory peak shaving and valley filling, poor dynamic control performance of the system, and slow response rate. Summary of the Invention

[0004] In view of this, the purpose of this invention is to propose a method for electric vehicles to participate in demand response control, so as to solve the problem that traditional consensus algorithms cannot quickly eliminate dynamic deviations in total power, resulting in unsatisfactory peak shaving and valley filling, poor dynamic control performance of the system and slow response rate.

[0005] To achieve the above objectives, the present invention provides a method for electric vehicles to participate in demand response control, comprising:

[0006] S101: Begin;

[0007] S102: Calculate the incremental cost η of the leading agent and multiple follower agents based on the initial load. i Initial values ​​are determined, and the row random matrix Dn is determined based on the response load limit;

[0008] S103: The real-time response load total power deviation ΔP[k] is calculated by the dominant agent;

[0009] S104: Determine whether the agent to be computed is the dominant agent. If not, proceed to S107; if yes, proceed to S111.

[0010] S105: Distribute response load indicators according to their limit values;

[0011] S106: The agent exits, and Dn is updated and recalculated;

[0012] S107: Calculate the (k+1)th η of the following agent according to formula (26). f[k+1] ;

[0013]

[0014] Where σ is the convergence coefficient of the following agent, d ij Let be the (i,j)th element in the row random matrix Dn, where n represents the number of agents;

[0015] S108: Calculate the response load of the electric vehicle cluster under the following agent;

[0016] S109: Determine if the response load limit has been exceeded. If yes, execute S105; otherwise, execute S110.

[0017] S110: Determine whether the analysis and calculation of all following agents have been completed. If not, execute S107; if yes, execute S114.

[0018] S111: Calculate the (k+1)th η of the dominant agent according to formula (25) l[k+1]

[0019]

[0020] Where μ is the convergence coefficient of the dominant agent, and d ij Let be the (i,j)th element in the row random matrix Dn, where n represents the number of agents;

[0021] S112: Calculate the response load of the electric vehicle cluster under the dominant agent;

[0022] S113: Determine if the response load limit is exceeded. If yes, execute S105; otherwise, execute S114.

[0023] S114: Calculate the response load index of each agent and sum it up;

[0024] S115: Calculate the total load power deviation ΔP of the (k+1)th real-time response. [k+1] ;

[0025] S116: Determine ΔP [k+1] Is it within the allowable error range? If yes, proceed to S117; otherwise, proceed to S104.

[0026] S117: Output the response load indicators of each agent, then end.

[0027] Optionally, in both S108 and S112, the electric vehicle cluster response load is calculated using formula (23):

[0028]

[0029] Among them, P Di Let α be the response load of the i-th electric vehicle cluster.i β i These are the coefficients of the quadratic and linear terms of the demand-side load response cost function.

[0030] Optionally, formula (23) is derived by substituting formula (11) into formula (17):

[0031]

[0032]

[0033] Define C i For P Di The partial derivative η i Let C be the incremental cost rate of the load response of the i-th electric vehicle cluster. i Let γ be the load response cost of the i-th electric vehicle cluster. i This is the user revenue coefficient.

[0034] Optionally, formula (11) is derived from the peak-shaving and valley-filling electric vehicle load response cost model, which includes:

[0035] Before the reduction of electric vehicle load, the revenue on the supply side is defined as:

[0036]

[0037] Where, p r For demand-side retail electricity prices, the time-of-use pricing mechanism enables p r P changes with time. LB,i,t is the basic load of electric vehicles, and n is the total number of electric vehicle clusters;

[0038] After the load reduction of electric vehicles, the rewards received by users are equivalent to the demand response management costs on the demand side, expressed as follows:

[0039]

[0040] in, These are the coefficients of the quadratic and linear terms of the demand response management cost function, respectively. This is the revenue coefficient; if user enthusiasm is low and the demand for load reduction is high, then the value will be large. Let T be the load reduction of the i-th electric vehicle cluster based on incentive-driven demand response at time t;

[0041] After the reduction in electric vehicle load, the revenue on the supply side is:

[0042]

[0043] The demand response cost for system load reduction is:

[0044]

[0045] Substituting equations (1)-(3) into equation (4), we obtain the demand response cost for peak shaving of the system as follows:

[0046]

[0047] When the load on electric vehicles increases, the reward users receive is the demand response management cost on the demand side, expressed as:

[0048]

[0049] in, These are the coefficients of the quadratic and linear terms of the demand response management cost function, respectively. This is the profit coefficient, equivalent to the minimum profit for users participating in the valley filling activity. Increase the load on the i-th electric vehicle cluster at time t;

[0050] With the increase in electric vehicle load, the revenue on the supply side is:

[0051]

[0052] The demand response cost for increasing system load is:

[0053]

[0054] Substituting equations (1), (6), and (7) into equation (8), the demand response cost for valley filling in the system is obtained as follows:

[0055]

[0056] Combining equations (5) and (9), we can obtain the load response cost model for peak shaving or valley filling of the system, as shown in equation (10).

[0057]

[0058] Optionally, the total load power deviation ΔP is:

[0059]

[0060] Wherein, δP D The total load index for system response (kW) The cumulative value of real-time response load.

[0061] Optionally, the derivation process of formulas (25) and (26) includes:

[0062] Traditional discrete average consensus protocol:

[0063]

[0064] Where, x i Let k represent the consensus state variable of agent i, and k be the iteration number; Ni is the set of neighboring agents of agent i, and a ij The elements in the adjacency matrix A represent the weight coefficients of the communication connection between agent i and agent j. They are 1 when a communication connection exists and 0 otherwise.

[0065] The Laplace matrix L of the adjacency matrix A contains element l ij for:

[0066]

[0067] To facilitate the application of the discrete average consensus protocol, based on equations (19) and (20), the elements l in matrix L are used. ij The traditional discrete average consensus protocol expression is transformed into the discrete average consensus protocol expression shown in equation (21):

[0068]

[0069]

[0070] Where, d ij Let be the (i,j)th element in the row random matrix Dn;

[0071] Based on formulas (21) and (24), formulas (25) and (26) are derived.

[0072] Optionally, the method includes: based on the influence of the upper and lower limits of the agent's response load on the limit value of the agent's cost incremental rate, the cost incremental rate calculation formula of each agent is modified again, as shown in formula (27);

[0073]

[0074] in, and ε represents the upper and lower limits of the incremental cost rate of agent i, respectively, and ε is the convergence coefficient of the dominant or follower agent. The upper and lower limits of the response load of each agent under different load responses are determined by the load capacity of the charging station and the number of signed user agreements, and the values ​​are determined by estimation and prediction methods.

[0075] Optionally, after determining whether the condition is negative in S116, the method is further defined as follows: if the method is stuck in an infinite loop, if not, S104 is executed; if yes, an alarm is issued.

[0076] Based on the same invention, this invention also provides a system for electric vehicles participating in a demand response control method, comprising:

[0077] Dispatch Center Layer: Under the premise of ensuring dynamic power balance and stable node voltage, this layer collects resource information on the responsive load and dispatch cost of each intelligent agent, thereby deciding to issue peak shaving during peak electricity consumption periods and valley filling during off-peak periods.

[0078] Multi-agent collaborative control layer: Composed of load agents, different electric vehicle clusters are regulated by corresponding load agents. The leading load agent receives load reduction or increase instructions and response load total indicators from the dispatch center, collects electric vehicle cluster resource information and uploads it to the dispatch center layer, and collects the real-time response load of each electric vehicle cluster in the electric vehicle response layer for calculation. The following load agents cooperate with the leading load agent to complete the demand response collaborative task and realize peak shaving and valley filling.

[0079] The load agent sends load reduction or increase control signals to the electric vehicle response layer, which is an electric vehicle cluster composed of multiple electric vehicles in a region.

[0080] Optionally, the electric vehicle cluster adopts a centralized communication method.

[0081] Beneficial effects: This control method introduces the total power deviation ΔP of the system response load to improve the incremental cost model of the dominant agent and the follower agent, obtaining formulas (25) and (26). Based on formulas (25) and (26), the incremental cost η of the dominant and follower agents in the (k+1)th iteration is calculated respectively. i[k+1] Then by η i[k+1] Calculate the response load index of each agent in the (k+1)th iteration, and then the leading agent calculates the total power deviation of the response load. If the deviation exceeds the allowable error range, proceed to the next iteration cycle. If the total power deviation of the response load ΔP does not exceed the allowable error range, the iteration ends.

[0082] As described above, this control method can quickly eliminate the total power deviation of the response load, improve peak shaving and valley filling effects, and make the system's dynamic control performance better and the response rate faster. The maximum steady-state total power deviation calculated by this control method is 0.0001%, and the algorithm has high calculation accuracy. It can quickly calculate the incremental cost rate and response load components of each agent. The algorithm has fast convergence and good steady-state convergence. Through calculation, the total power deviation can be quickly reduced to zero. When the discrete simulation step size is set to 0.07s, the maximum calculation completion time of the algorithm is about 3s, and the commonly used calculation time is about 0.8s. Attached Figure Description

[0083] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0084] Figure 1 This is a flowchart of the demand response control method according to an embodiment of the present invention;

[0085] Figure 2 This is a diagram of a three-layer control architecture for electric vehicles participating in demand response based on multiple agents, according to an embodiment of the present invention.

[0086] Figure 3 This is a topology diagram of the multi-agent cooperative control layer in the simulation model system of this invention embodiment;

[0087] Figure 4 This is a graph showing the incremental cost rate and real-time load reduction curves of each agent in the simulation results of this embodiment of the invention.

[0088] Figure 5 This is a comparison chart of the steady-state load reduction of each agent during different response processes in the simulation results of this embodiment of the invention;

[0089] Figure 6 This is a graph showing the incremental cost rate and real-time load increase of each agent in the simulation results of this embodiment of the invention.

[0090] Figure 7 This is a comparison chart of the steady-state load increase of each agent during different response processes in the simulation results of this embodiment of the invention;

[0091] Figure 8 This is a structural diagram of an AC / DC electric vehicle charging station according to an embodiment of the present invention;

[0092] Figure 9 This is a voltage variation curve of the charging side of the AC electric vehicle charging station 1 according to an embodiment of the present invention;

[0093] Figure 10 This is a graph showing the change in total power load on the charging side of the AC electric vehicle charging station 1 according to an embodiment of the present invention. Detailed Implementation

[0094] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0095] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0096] like Figure 1 As shown, a method for electric vehicles to participate in demand response control includes:

[0097] S101: Begin;

[0098] S102: Calculate the incremental cost η of the leading agent and multiple follower agents based on the initial load. i Initial values ​​are determined, and the row random matrix Dn is determined based on the response load limit;

[0099] S103: The real-time response load total power deviation ΔP[k] is calculated by the dominant agent;

[0100] S104: Determine whether the agent to be computed is the dominant agent. If not, proceed to S107; if yes, proceed to S111.

[0101] S105: Distribute response load indicators according to their limit values;

[0102] S106: The agent exits, and Dn is updated and recalculated;

[0103] S107: Calculate the (k+1)th η of the following agent according to formula (26). f[k+1] ;

[0104]

[0105] Where σ is the convergence coefficient of the following agent, d ij Let be the (i,j)th element in the row random matrix Dn, where n represents the number of agents;

[0106] S108: Calculate the response load of the electric vehicle cluster under the following agent;

[0107] S109: Determine if the response load limit has been exceeded. If yes, execute S105; otherwise, execute S110.

[0108] S110: Determine whether the analysis and calculation of all following agents have been completed. If not, execute S107; if yes, execute S114.

[0109] S111: Calculate the (k+1)th η of the dominant agent according to formula (25) l[k+1]

[0110]

[0111] Where μ is the convergence coefficient of the dominant agent, and d ij Let be the (i,j)th element in the row random matrix Dn, where n represents the number of agents;

[0112] S112: Calculate the response load of the electric vehicle cluster under the dominant agent;

[0113] S113: Determine if the response load limit is exceeded. If yes, execute S105; otherwise, execute S114.

[0114] S114: Calculate the response load index of each agent and sum it up;

[0115] S115: Calculate the total load power deviation ΔP of the (k+1)th real-time response. [k+1] ;

[0116] S116: Determine ΔP [k+1] Is it within the allowable error range? If yes, proceed to S117; otherwise, proceed to S104.

[0117] S117: Output the response load indicators of each agent, then end.

[0118] This control method introduces the total power deviation ΔP of the system response load and improves the incremental cost rate model of the dominant agent and the follower agent, obtaining formulas (25) and (26). Based on formulas (25) and (26), the incremental cost rate η of the dominant and follower agents in the (k+1)th iteration is calculated respectively. i[k+1] Then by η i[k+1] Calculate the response load index of each agent in the (k+1)th iteration, and then the leading agent calculates the total power deviation of the response load. If the deviation exceeds the allowable error range, proceed to the next iteration cycle. If the total power deviation of the response load ΔP does not exceed the allowable error range, the iteration ends.

[0119] As described above, this control method can quickly eliminate the total power deviation of the response load, improve peak shaving and valley filling effects, and make the system's dynamic control performance better and the response rate faster. The maximum steady-state total power deviation calculated by this control method is 0.0001%, and the algorithm has high calculation accuracy. It can quickly calculate the incremental cost rate and response load components of each agent. The algorithm has fast convergence and good steady-state convergence. Through calculation, the total power deviation can be quickly reduced to zero. When the discrete simulation step size is set to 0.07s, the maximum calculation completion time of the algorithm is about 3s, and the commonly used calculation time is about 0.8s.

[0120] In some embodiments, the electric vehicle cluster response load is calculated using formula (23) in both S108 and S112:

[0121]

[0122] Among them, P Di Let α be the response load of the i-th electric vehicle cluster. i β i The coefficients of the quadratic and linear terms of the demand-side load response cost function are (¥ / (kW·h)).

[0123] In some embodiments, formula (23) is derived by substituting formula (11) into formula (17):

[0124]

[0125]

[0126] Define C i For P Di The partial derivative η i Let C be the incremental cost rate of the load response of the i-th electric vehicle cluster. i Let γ be the load response cost of the i-th electric vehicle cluster. i This is the user revenue coefficient.

[0127] In some embodiments, Formula 11 is derived from a peak-shaving and valley-filling electric vehicle load response cost model, wherein the peak-shaving and valley-filling electric vehicle load response cost model includes:

[0128] Before the reduction of electric vehicle load, the supply-side revenue (unit: ¥) is defined as follows:

[0129]

[0130] Where, p r The demand-side retail electricity price (¥ / kW·h) and the time-of-use pricing mechanism enable p r Changes over time; P LB,i,tis the base load of electric vehicles (kW); n is the total number of electric vehicle clusters;

[0131] After the load reduction of electric vehicles, the reward received by users is equivalent to the demand response management cost on the demand side (unit: ¥, the same below), expressed as follows:

[0132]

[0133] in, These are the quadratic term coefficients (¥ / (kW·h)) and linear term coefficients (¥ / (kW·h)) of the user reward model, respectively. This is the revenue coefficient (¥). If user enthusiasm is low and the demand for load reduction is high, then the value will be large. Let t be the load reduction (kW) of the i-th electric vehicle cluster based on incentive-driven demand response at time t, where T is the set of time periods;

[0134] After the reduction in electric vehicle load, the revenue on the supply side is:

[0135]

[0136] The demand response cost for system load reduction is:

[0137]

[0138] Substituting equations (1)-(3) into equation (4), we obtain the demand response cost for peak shaving of the system as follows:

[0139]

[0140] When the load on electric vehicles increases, the reward users receive is the demand response management cost on the demand side, expressed as:

[0141]

[0142] in, These are the quadratic coefficients (¥ / (kW·h)) and linear coefficients (¥ / (kW·h)) of the demand response management cost function, respectively. This is the profit coefficient (¥), equivalent to the minimum profit for users participating in the valley filling activity; Increase the load (kW) of the i-th electric vehicle cluster at time t;

[0143] With the increase in electric vehicle load, the revenue on the supply side is:

[0144]

[0145] The demand response cost for increasing system load is:

[0146]

[0147] Substituting equations (1), (6), and (7) into equation (8), the demand response cost for valley filling in the system is obtained as follows:

[0148]

[0149] Combining equations (5) and (9), we can obtain the load response cost model for peak shaving or valley filling of the system, as shown in equation (10).

[0150]

[0151] In some embodiments, the total power deviation ΔP of the response load is:

[0152]

[0153] Wherein, δP D The total load index for system response (kW) The cumulative value of real-time response load;

[0154] In some embodiments, the derivation process of formulas (25) and (26) includes:

[0155] Traditional discrete average consensus protocol:

[0156]

[0157] Where, x i Let k represent the consensus state variable of agent i, and k be the iteration number; Ni is the set of neighboring agents of agent i, and a ij The elements in the adjacency matrix A represent the weight coefficients of the communication connection between agent i and agent j. They are 1 when a communication connection exists and 0 otherwise.

[0158] The Laplace matrix L of the adjacency matrix A contains element l ij for:

[0159]

[0160] To facilitate the application of the discrete average consensus protocol, based on equations (19) and (20), the elements l in matrix L are used. ij The traditional discrete average consensus protocol expression is transformed into the discrete average consensus protocol expression shown in equation (21):

[0161]

[0162]

[0163] Where, d ijLet be the (i,j)th element in the row random matrix Dn;

[0164] Based on formulas (21) and (24), formulas (25) and (26) are derived.

[0165] In some embodiments, the method includes: modifying the cost increment rate calculation formula of each agent again based on the influence of the upper and lower limits of the agent's response load on the limit value of the agent's cost increment rate, as shown in formula (27);

[0166]

[0167] in, and ε represents the upper and lower limits of the incremental cost rate of agent i, respectively, and ε is the convergence coefficient of the dominant or follower agent. The upper and lower limits of the response load of each agent under different load responses are determined by the load capacity of the charging station and the number of signed user agreements, and the values ​​are determined by estimation and prediction methods.

[0168] Since the upper and lower limits of the agent's response load determine the limit of the agent's incremental cost rate, the problem that arises when the calculated real-time response load of the agent exceeds the limit is solved by further improving the incremental cost rate model.

[0169] In some embodiments, after determining whether the condition is negative in S116, the method is further defined as follows: if the method is stuck in an infinite loop, if not, S104 is executed; if yes, an alarm is issued.

[0170] like Figure 2 As shown, to further implement the present invention, the present invention also provides a system for electric vehicles participating in a demand response control method, comprising:

[0171] Dispatch Center Layer: Under the premise of ensuring dynamic power balance and stable node voltage, this layer collects resource information such as the responsive load and dispatch cost of each intelligent agent, and then decides to issue peak shaving during peak electricity consumption and valley filling during off-peak periods.

[0172] Multi-agent collaborative control layer: Composed of load agents, different electric vehicle clusters are regulated by corresponding load agents. The leading load agent receives load reduction or increase instructions and response load total indicators from the dispatch center, collects electric vehicle cluster resource information and uploads it to the dispatch center layer, and collects the real-time response load of each electric vehicle cluster in the electric vehicle response layer for calculation. The following load agents cooperate with the leading load agent to complete the demand response collaborative task and realize peak shaving and valley filling.

[0173] Electric vehicle response layer: Multiple electric vehicles in a region form an electric vehicle cluster, which executes load reduction or increase control signals issued by the load agent. The electric vehicle cluster adopts a centralized communication method.

[0174] During operation, the scheduling center first calculates the total response load index based on information such as system power flow and total responsive load, and then sends the demand response command and the total response load index to the dominant load agent in the multi-agent collaborative control layer.

[0175] Subsequently, the leading load agent receives the issued response instructions and total response load indicators, as well as collects real-time response load information from each electric vehicle cluster. It calculates the total power deviation required for demand response and then sends the response instructions and total power deviation information to neighboring follower load agents. The agents communicate with each other and utilize the cost increment rate update criterion for the leading and follower agents in the improved discrete master-slave consensus algorithm based on cost increment rate to calculate the cost increment rate and response load components. Then, each agent follows the equal increment rate criterion to convert the response load components into control signals required by each electric vehicle charging station to execute the response load indicators, and sends these signals to each electric vehicle charging station.

[0176] Finally, the controllers of each AC and DC electric vehicle charging station in the electric vehicle response layer receive control signals, execute response commands, and complete the assigned peak shaving and valley filling tasks. Each AC and DC electric vehicle charging station consists of a power conversion device, electric vehicle loads, a central controller, a charging controller, energy storage, and an energy storage controller, and each employs different control strategies to execute load response tasks. AC electric vehicle charging stations use switching charging modes for load response, utilizing AC / DC converters to charge electric vehicles and employing both fast and slow charging methods. DC electric vehicle charging stations use a power-on / off operating mode and a fast charging method for load response.

[0177] To verify the superiority of this system, the present invention is based on Figure 2 A simulation model of a multi-agent electric vehicle cluster participating in demand response system was constructed using Matlab / Simulink software. The simulation study investigated the system's peak shaving and valley filling modes, as well as the operational characteristics of the electric vehicle response layer. Figure 3 The figure shown is a topology diagram of the multi-agent cooperative control layer in the simulation model system.

[0178] Simulation Content 1: Study on System Response under Load Reduction Command.

[0179] Let the initial total load reduction be 1050kW, and the consistent incremental cost rate be 8.7028¥ / (kW·h). At this point, the total cost of system demand response is at its lowest, and the system reaches the optimal load reduction state.

[0180] If the dispatch center sends load reduction targets of 1250kW and 1350kW to the leading agent at 0s and 4s respectively, then... Figure 4 The figure shows the cost increment rate and real-time load reduction curves for each agent. From... Figure 4 (a) It can be seen that after receiving the two load reduction indicators, the cost increment rate of each agent increased to 8.845¥ / (kW·h) and 8.916¥ / (kW·h) at 0.65s and 4.66s, respectively, due to the increase in the total load reduction task. Figure 4 In (b), PD represents the amount of load reduction by the agent; the greater the load reduction of the electric vehicle cluster, the larger the PD. Figure 4 (b) It can be seen that the real-time load reduction of each agent increases with the increase of the total load reduction index, and is rationally allocated according to the capacity ratio. For example... Figure 5 The figure shown is a comparison of the steady-state load reduction of each agent during different response processes in the simulation results. Figure 5 It can be seen that after the initial system stabilizes, the steady-state total power deviation is 9.9564 × 10⁻⁶. -7 After the second indicator system stabilizes, the steady-state total power deviation is 9.9566 × 10 kW. -7 The kW indicates that the multi-agent collaborative control layer has a high degree of accuracy in achieving the overall load reduction target.

[0181] Therefore, when this control method is applied to electric vehicle clusters to participate in demand response, each agent satisfies the equal incremental rate criterion, the system is stable, the total cost of load reduction is minimized, and the optimal allocation of load reduction is achieved.

[0182] Simulation Content 2: Study on System Response under Increased Load Command.

[0183] Assuming the initial load increase target is 750kW, if the dispatch center sends load increase targets of 850kW and 950kW to the dominant agent at 0s and 4s respectively. Figure 6 The figure shows the cost increment rate and real-time load increase curves for each agent. Figure 6 (a) It can be seen that at 0.68s, the incremental cost rate of each agent converges to a consistent value of 8.506¥ / (kW·h). The incremental cost rate of each agent increases with the increase of the total load index, eventually converging to a consistent value, thus minimizing the total cost of the system's load increase. At 4.73s, the incremental cost rate of each agent converges to a consistent value of 8.580¥ / (kW·h). Figure 6 In (b), PI represents the increase in load by the agent; the greater the increase in load of the controlled electric vehicle cluster, the larger the PI. Figure 6 (b) It can be seen that the real-time increase in load for each agent increases with the increase in the total increase in load index. For example... Figure 7The figure shown is a comparison of the steady-state load increase of each agent during different response processes in the simulation results. Figure 6 and Figure 7 It can be seen that after the system stabilizes, each agent can reasonably allocate the increased load according to the capacity ratio, achieving the optimal allocation state of the increased load. Moreover, the algorithm converges quickly, and the time from the start of calculating the incremental cost rate and response load of the agents to the realization of system stability does not exceed 0.8 seconds.

[0184] Simulation Content 3: Study on the operating characteristics of the electric vehicle response layer.

[0185] like Figure 8 The diagram shows the structure of AC and DC electric vehicle charging stations. Figure 9 The figure shown is a voltage variation curve on the charging side of AC electric vehicle charging station 1. Figure 10 The figure shown is a curve illustrating the total power variation of the charging side load at AC electric vehicle charging station 1. Figure 9 and Figure 10 It can be seen that charging station 1 can respond quickly to commands. When reducing the load, the charging side voltage first rises and then falls, finally maintaining at around 750V, and can quickly execute the target of reducing the load by 234.14kW. The electric vehicle response layer using this control method can execute the overall peak shaving task relatively accurately, and the dynamic deviation of the total power can be reduced rapidly with the system response, realizing precise power control of each electric vehicle charging station's intelligent body.

[0186] In summary, the maximum steady-state total power deviation calculated by this control method is 0.0001%, the algorithm has high calculation accuracy, and can quickly calculate the incremental cost rate and response load component of each agent. The algorithm has fast convergence and good steady-state convergence. Through calculation, the total power deviation can be quickly reduced to zero. When the discrete simulation step size is set to 0.07s, the maximum calculation end time of the algorithm is about 3s, and the commonly used calculation time is about 0.8s.

[0187] The distributed three-layer control architecture reduces the adverse effects of large communication data volume and communication latency. This architecture is very suitable for application scenarios in which a large number of electric vehicles are involved in demand response and are widely distributed. It is also beneficial for the power system to optimize the operation of distributed multi-source and multi-load attributes in the future.

[0188] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention (including the claims) is limited to these examples; within the framework of the invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of the different aspects of the invention as described above, which are not provided in the details for the sake of brevity.

[0189] This invention is intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for electric vehicles participating in demand response control, characterized in that, include: S101: Beginning; S102: Calculate the incremental cost ƞ of the leading agent and multiple follower agents based on the initial load. i Initial values ​​are set, and the row random matrix Dn is determined based on the response load limit; S103: The real-time response load total power deviation ΔP[k] is calculated by the dominant agent; S104: Determine whether the agent to be computed is the dominant agent. If not, proceed to S107; if yes, proceed to S111. S105: Distribute response load indicators according to their limit values; S106: The agent exits, and Dn is updated and recalculated; S107: Calculate the (k+1)th ƞ of the following agent according to formula (26). f [k+1]; ， (26) Where σ is the convergence coefficient of the following agent, d ij Let be the (i, j)th element in the row random matrix Dn, where n represents the number of agents; S108: Calculate the response load of the electric vehicle cluster under the following agent; S109: Determine if the response load limit has been exceeded. If yes, execute S105; otherwise, execute S110. S110: Determine whether the analysis and calculation of all following agents have been completed. If not, execute S107; if yes, execute S114. S111: Calculate the (k+1)th ƞ of the dominant agent according to formula (25). l [k+1] ， (25) Where μ is the convergence coefficient of the dominant agent, and d ij Let be the (i, j)th element in the row random matrix Dn, where n represents the number of agents; S112: Calculate the response load of the electric vehicle cluster under the dominant agent; S113: Determine if the response load limit is exceeded. If yes, execute S105; otherwise, execute S114. S114: Calculate the response load index of each agent and sum it up; S115: Calculate the total load power deviation ΔP[k+1] of the (k+1)th real-time response; S116: Determine whether ΔP[k+1] is within the allowable error range. If yes, execute S117; otherwise, execute S104. S117: Output the response load indicators of each agent, then end.

2. The method for electric vehicle participation in demand response control according to claim 1, characterized in that, In both S108 and S112, the electric vehicle cluster response load is calculated using formula (23): ， (23) Among them, P Di Let α be the response load of the i-th electric vehicle cluster. i β i These are the coefficients of the quadratic and linear terms of the demand-side load response cost function.

3. The method for electric vehicle participation in demand response control according to claim 2, characterized in that, The formula (23) is derived by substituting formula (11) into formula (17): ， (11) (17) Define C i For P Di The partial derivative η i Let C be the incremental cost rate of the load response of the i-th electric vehicle cluster. i Let γ be the load response cost of the i-th electric vehicle cluster. i This is the user revenue coefficient.

4. The method for electric vehicle participation in demand response control according to claim 3, characterized in that, Formula (11) is derived from the peak-shaving and valley-filling electric vehicle load response cost model, which includes: Before the reduction of electric vehicle load, the revenue on the supply side is defined as: ， ， (1) Where, p r For demand-side retail electricity prices, the time-of-use pricing mechanism enables p r Changes with time, is the basic load of electric vehicles, and n is the total number of electric vehicle clusters; After the load reduction of electric vehicles, the rewards received by users are equivalent to the demand response management costs on the demand side, expressed as follows: ， ， (2) in, , These are the coefficients of the quadratic and linear terms of the demand response management cost function, respectively. This is the revenue coefficient; if user enthusiasm is low and the demand for load reduction is high, then the value will be large. Let T be the load reduction of the i-th electric vehicle cluster based on incentive-driven demand response at time t; After the reduction in electric vehicle load, the revenue on the supply side is: ， ， (3) The demand response cost for system load reduction is: ， ， (4) Substituting equations (1) to (3) into equation (4), we obtain the demand response cost for peak shaving of the system as follows: ， ， (5) When the load on electric vehicles increases, the reward users receive is the demand response management cost on the demand side, expressed as: ， ， (6) in, , These are the coefficients of the quadratic and linear terms of the demand response management cost function, respectively. This is the profit coefficient, equivalent to the minimum profit for users participating in the valley filling activity. Increase the load on the i-th electric vehicle cluster at time t; With the increase in electric vehicle load, the revenue on the supply side is: ， ， (7) The demand response cost for increasing system load is: ， ， (8) Substituting equations (1), (6), and (7) into equation (8), the demand response cost for valley filling in the system is obtained as follows: ， ， (9) Combining equations (5) and (9), we can obtain the load response cost model for peak shaving or valley filling of the system, as shown in equation (10): ， ， 。 (10) 5. The method for electric vehicle participation in demand response control according to claim 1, characterized in that, The total load power deviation ΔP is: (24) Wherein, δP D The total system response load index, The cumulative value of real-time response load.

6. The method for electric vehicle participation in demand response control according to claim 5, characterized in that, The derivation process of formulas (25) and (26) includes: Based on the traditional discrete average consensus protocol: (19) Where, x i N represents the consensus state variable of the i-th agent, k is the iteration number; i Let a be the set of neighboring communicating agents of agent i. ij The elements in the adjacency matrix A represent the weight coefficients of the communication connection between agent i and agent j. They are 1 when a communication connection exists and 0 otherwise. Based on the elements l in the Laplace matrix L of the adjacency matrix A ij : (20) Based on equations (19) and (20), using the elements l in matrix L ij The traditional discrete average consensus protocol expression is transformed into the discrete average consensus protocol formula shown in formula (21): (21) (22) Where, d ij Let be the (i, j)th element in the row random matrix Dn; Based on formulas (21) and (24), formulas (25) and (26) are derived.

7. The method for electric vehicle participation in demand response control according to claim 6, characterized in that, The method includes: based on the influence of the upper and lower limits of the agent's response load on the limit value of the agent's cost incremental rate, the cost incremental rate calculation formula of each agent is modified again, as shown in formula (27); (27) Where, η u i and η l i ε represents the upper and lower limits of the incremental cost rate of agent i, respectively, and ε is the convergence coefficient of the dominant or follower agent. The upper and lower limits of the response load of each agent under different load responses are determined by the load capacity of the charging station and the number of signed user agreements, and the values ​​are determined by estimation and prediction methods.

8. A method for electric vehicles participating in demand response control according to claim 6, characterized in that, The step S116 after determining whether the condition is negative includes: determining whether the method has entered an infinite loop; if not, executing S104; if yes, issuing an alarm.

9. A system for an electric vehicle participating in a demand response control method according to any one of claims 1-8, characterized in that, include: Dispatch Center Layer: Under the premise of ensuring dynamic power balance and stable node voltage, this layer collects resource information on the responsive load and dispatch cost of each intelligent agent, thereby deciding to issue peak shaving during peak electricity consumption periods and valley filling during off-peak periods. Multi-agent collaborative control layer: Composed of load agents, different electric vehicle clusters are regulated by corresponding load agents. The leading load agent receives load reduction or increase instructions and response load total indicators from the dispatch center, collects electric vehicle cluster resource information and uploads it to the dispatch center layer, and collects the real-time response load of each electric vehicle cluster in the electric vehicle response layer for calculation. The following load agents cooperate with the leading load agent to complete the demand response collaborative task and realize peak shaving and valley filling. The load agent sends load reduction or increase control signals to the electric vehicle response layer, which is an electric vehicle cluster composed of multiple electric vehicles in a region.

10. The system for an electric vehicle participating in a demand response control method according to claim 9, characterized in that, The electric vehicle cluster uses a centralized communication method.

Images