Power market demand response management triggering privacy protection method based on aggregation game

Abstract

The invention discloses an electricity market demand response management triggering privacy protection method based on an aggregation game, and belongs to the technical field of electricity market demand management. The method comprises the following steps: acquiring a real-time state of a generator by using a state-space equation; constructing an aggregation game model of the multi-turbine generator to optimize the profit of the generator; constructing a continuous time random noise generation model; transmitting a noise-added aggregation function estimation result and a noise estimation signal to other specified generators; determining an optimized output strategy of the generator; and constructing a state space expression of the turbine generator, and driving the generator to output power according to a predetermined strategy. According to the method, privacy protection is introduced to trigger dynamic average consistent control, time-varying Nash equilibrium point search and model prediction control, the total output electric power of the generator set in a dynamic environment meets the market power requirement, the strategy is adjusted to maximize the income of the generator set, communication resources are saved, and the power efficiency of the generator set is improved. And the privacy information of the generator is protected through the power value transmitted in the random noise interference network.

Description

Technical Field

[0001] This invention belongs to the field of electricity market demand management technology, specifically relating to a method for triggering privacy protection in electricity market demand response management based on aggregation game theory. Background Technology

[0002] With the development of smart grids and the energy internet, demand response management in the electricity market has become an important means to achieve supply and demand balance and improve system stability and economy. In recent years, convergent game theory has demonstrated powerful modeling capabilities in the field of demand response management, effectively characterizing the decision-making behavior of numerous participants in the electricity market, such as power companies and end-users, under competitive conditions. Market participants make optimal decisions that satisfy system constraints and obtain maximum benefits through game theory models. Therefore, research on demand response management based on convergent game theory provides an economical and effective platform for the development of the electricity market, which is of great significance.

[0003] In the turbine generator payoff function model, the profit obtained through power generation is related to the sum of the decisions made by all participants. How local turbine generators can obtain this sum of decisions in a fully distributed manner is key to solving the aggregation game problem. Traditional methods for estimating the sum of decisions are often designed based on a fixed environment, meaning the sum of decisions remains constant. This makes traditional methods unsuitable for dynamic environments because the strategies of turbine generators need to be adjusted according to changes in the environment, and the sum of decisions varies. The continuous-time dynamic averaging consensus algorithm can effectively estimate the average of dynamic decisions. By introducing the continuous-time dynamic averaging consensus algorithm, the inapplicability of traditional methods in dynamic environments is effectively overcome.

[0004] Continuous-time dynamic average consensus algorithms typically rely on multiple objects exchanging information in open networks. Preventing the theft of privacy information and ensuring continuous information exchange are crucial for secure and efficient task completion. Event-triggered mechanisms and privacy protection mechanisms are powerful tools for addressing these issues. Event-triggered mechanisms can avoid continuous information transmission, saving communication resources. Privacy protection mechanisms based on random noise injection can interfere with network eavesdroppers from stealing actual private information, preventing security incidents and economic losses caused by privacy leaks.

[0005] Therefore, the technical problem that this invention aims to solve is how to solve the demand response management problem through dynamic aggregation game theory, and achieve privacy protection and trigger communication under the constraint of meeting market power demand. Summary of the Invention

[0006] The purpose of this invention is to provide a privacy protection method for demand response management in the power market based on aggregation game theory, so as to solve the problems mentioned in the background art.

[0007] The objective of this invention is achieved as follows: a privacy protection method for demand response management in the electricity market based on aggregation game theory, characterized by the following steps:

[0008] Step S1: Construct a continuous-time dynamic system model of the turbine generator, determine the state-space equation of the turbine generator's power output, and obtain the generator's real-time state;

[0009] Step S2: Construct an aggregate game model for multi-turbine generators to optimize generator profits, and use the aggregate function estimation model to assist in solving the problem;

[0010] Step S3: Construct a continuous-time random noise generation model, superimpose its output onto the aggregation function estimation result to achieve privacy protection, and superimpose noise to generate a noise estimation signal;

[0011] Step S4: Introduce an event-triggered mechanism to avoid continuous transmission, and transmit the noisy aggregation function estimation result together with the noise estimation signal to other designated generators;

[0012] Step S5: Construct a noise estimation model, estimate the original noise based on the noise estimation signal, and obtain the input of the aggregation function estimation model;

[0013] Step S6: Construct a time-varying Nash equilibrium point search model to determine the optimal power output strategy for the generator;

[0014] Step S7: Construct the state-space representation of the turbine generator and generate power control inputs to drive the generator to output power according to a predetermined strategy.

[0015] Preferably, the construction of the continuous-time dynamic system model of the turbine generator in step S1 specifically involves:

[0016] In an undirected, connected communication network, consider N turbine generators. The dynamic model of the i-th turbine generator is as follows:

[0017] ;

[0018] in, It is the output power of the i-th generator system. It is the steam valve opening degree of the i-th generator. It is the relative speed of the i-th generator. The time constant of the i-th mechanical turbine It is the time constant of the speed controller of the i-th machine. It is the turbine gain of the i-th machine. It is the adjustment constant of the i-th machine. It is the speed of the synchronizer. It is the inertial constant. It is the unit damping constant. It is the power control input of the i-th generator.

[0019] Preferably, the construction of the aggregation game model for the multi-turbine generator in step S2 is specifically as follows:

[0020] In an undirected, interconnected communication network, considering N turbine generators operating in a profit-maximizing mode, the aggregation game model is defined as follows: ;

[0021] The constraints are: ;

[0022] in, Let be the power generation cost function of the i-th generator. This represents the cost coefficient. ; It is the aggregation function of aggregation game; Indicates electricity price, This indicates the degree to which local energy prices are affected by total power output; This represents the total power demand in the market.

[0023] Preferably, in step S2, an aggregation function estimation model is used to assist in the solution, and the aggregation function estimation model is:

[0024] ;

[0025] in, It is the aggregation function of generator i on the market. Local estimates, yes To his neighbors aggregate functions The estimate, It is a scalar and satisfies , , , The aggregation function estimation model depends only on information from its neighbors.

[0026] Preferably, in step S3, a continuous-time random noise generation model is constructed, and its output is superimposed on the aggregation function estimation result to achieve privacy protection. Furthermore, noise is superimposed to generate a noise estimation signal. Specifically:

[0027] The continuous-time random noise generation model is as follows: ;

[0028] in, It is a one-dimensional standard Brownian motion over a complete probability space; and Indicates the specified function; Represents a constant; The output of the random noise generation model represents the noise signal that is injected into privacy information to protect it from eavesdropping.

[0029] Will This is superimposed on the aggregation function estimation result to ensure that the true information is not intercepted during transmission over an open network. The noise estimation signal is obtained by superimposing noise on it, and its expression is:

[0030] .

[0031] Preferably, in step S4, an event-triggered mechanism is introduced to avoid continuous transmission, and the noisy aggregation function estimation result is transmitted together with the noise estimation signal to other designated generators, specifically as follows:

[0032] The event triggering mechanism for the i-th generator is determined by the following expression:

[0033] ;

[0034] in, It is a trigger function. and These are all measurement errors. , , It is a constant; At the k-th trigger moment of generator i, only when triggered will generator i transmit the information sequence. Send it to the designated generator j, thereby avoiding continuous communication and saving communication resources.

[0035] Preferably, in step S5, a noise estimation model is constructed, and the original noise is estimated based on the noise estimation signal to obtain the input of the aggregation function estimation model, specifically as follows:

[0036] The random noise estimation model for the j-th generator is determined by the following expression:

[0037] ;

[0038] in, and These represent the noise levels of generator j relative to its neighbor i. and The estimate, For continuously differentiable bounded time-varying gain, Satisfying the Lipschitz condition , ;

[0039] Generator j obtains the aggregation function estimate for i by subtracting the estimated noise from the noisy aggregation function estimate from its neighbor i:

[0040] ;

[0041] in, It is the input to the aggregation function estimation model of j.

[0042] Preferably, step S6, which involves constructing a time-varying Nash equilibrium point search model to determine the optimal power output strategy for the generator, specifically includes:

[0043] The time-varying Nash equilibrium point search model for the i-th turbine generator is:

[0044] ;

[0045] in, , ; It is the control gain, which satisfies ; It is a parameter and , It is a parameter that satisfies ; and It is a constant; a set Let i represent the set of i's neighbors, and let i exchange information with its neighbors.

[0046] Preferably, the construction of the state-space representation of the turbine generator in step S7 specifically involves:

[0047] The state-space representation of the i-th turbine generator is: ;

[0048] in, , , and ;

[0049] remember It is the power control input sequence and It is the output sequence; in order to make the i-th generator output power Reaching the time-varying Nash equilibrium point Power control input for:

[0050] ;

[0051] in, , It is a diagonal matrix, where the diagonal elements are predetermined weight parameters. It is the discrete-time reference sequence at time k. and It is a matrix, and its specific form is: , .

[0052] Compared with the prior art, the present invention has the following improvements and advantages:

[0053] 1. By introducing a privacy-preserving continuous-time dynamic average consensus algorithm, the time-varying decision sum of a multi-turbine generator system is estimated. At the same time, a constrained aggregation game dynamic Nash equilibrium point search algorithm is proposed to be combined with it, so that the dynamic Nash equilibrium point can be searched with only the information of the neighbors.

[0054] 2. By introducing a continuous-time random noise injection mechanism, the privacy of continuous-time signals is protected, ensuring that the output power value of the turbine generator is not directly stolen by malicious eavesdroppers; at the same time, an event triggering mechanism is introduced to avoid continuous information exchange and save system resources. Attached Figure Description

[0055] Figure 1 This is an overall flowchart of the method of the present invention.

[0056] Figure 2 This is a schematic diagram of the communication network for four turbine generators.

[0057] Figure 3 For each turbine generator, apply the aggregation function separately. A schematic diagram of the estimation results.

[0058] Figure 4 A schematic diagram illustrating the result of injecting privacy information into continuous-time random noise to prevent eavesdropping.

[0059] Figure 5 This is a schematic diagram showing the power output of each turbine generator.

[0060] Figure 6 A schematic diagram illustrating the effectiveness of the dynamic Nash equilibrium search model is shown.

[0061] Figure 7 A schematic diagram showing the total output power of all turbine generators and the market power demand. Detailed Implementation

[0062] The invention will be further summarized below with reference to the accompanying drawings.

[0063] like Figure 1 As shown, a privacy protection method for demand response management in the electricity market based on aggregation game theory is proposed. The method includes the following steps:

[0064] Step S1 involves constructing a continuous-time dynamic system model for the turbine generator, specifically as follows:

[0065] In an undirected, connected communication network, consider N turbine generators. The dynamic model of the i-th turbine generator is as follows:

[0066] ;

[0067] in, It is the output power of the i-th generator system. It is the steam valve opening degree of the i-th generator. It is the relative speed of the i-th generator. The time constant of the i-th mechanical turbine It is the time constant of the speed controller of the i-th machine. It is the turbine gain of the i-th machine. It is the adjustment constant of the i-th machine. It is the speed of the synchronizer. It is the inertial constant. It is the unit damping constant. It is the power control input of the i-th generator.

[0068] In step S2, an aggregation game model is constructed to optimize generator profits, and an aggregation function estimation model is constructed to assist in solving it, specifically as follows:

[0069] N turbine generators work together to meet market power demand. To maximize the profit of each generator, a power output strategy is formulated for the generators using a convergent game model. The convergent game model is defined as follows: ;

[0070] The constraints are: ;

[0071] in, Let be the power generation cost function of the i-th generator. This represents the cost coefficient. ; It is the aggregation function of aggregation game; Indicates electricity price, This indicates the degree to which local energy prices are affected by total power output; This represents the total power demand in the market.

[0072] The aggregation function estimation model is as follows: ;

[0073] in, It is the aggregation function of generator i on the market. Local estimates, yes To his neighbors aggregate functions The estimate, It is a scalar and satisfies , , , The aggregation function estimation model relies solely on information from its neighbors, thus the method is fully distributed.

[0074] In step S3, a continuous-time random noise generation model is constructed, and its output is superimposed on the aggregation function estimation result to achieve privacy protection. Furthermore, noise is superimposed to generate a noise estimation signal, specifically as follows:

[0075] The continuous-time random noise generation model is as follows: ;

[0076] in, It is a one-dimensional standard Brownian motion over a complete probability space; and It is the specified function; It is the output of the random noise generation model;

[0077] Will This is superimposed on the aggregation function estimation result to prevent the interception of real information during transmission over open networks, thus achieving privacy protection. Simultaneously, noise is superimposed to obtain the noise estimation signal, expressed as: ;

[0078] Step S4 introduces an event-triggered mechanism to avoid continuous transmission, transmitting the noisy aggregation function estimation result along with the noise estimation signal to other designated generators, specifically:

[0079] The event triggering mechanism for the i-th generator is determined by the following expression:

[0080] ;

[0081] in, It is a trigger function. and These are all measurement errors. , , It is a constant; At the k-th trigger moment of generator i, only when triggered will generator i transmit the information sequence. Send it to the designated generator j, thereby avoiding continuous communication and saving communication resources.

[0082] In step S5, a random noise estimation model is constructed, and the noise estimation result is subtracted from the received noisy aggregation function estimation result to obtain the input of the aggregation function estimation model, specifically:

[0083] The random noise estimation model for the j-th generator is determined by the following expression:

[0084] ;

[0085] in, and These represent the noise levels of generator j relative to its neighbor i. and The estimate, For continuously differentiable bounded time-varying gain, Satisfying the Lipschitz condition , ;

[0086] Generator j obtains the aggregation function estimate for i by subtracting the estimated noise from the noisy aggregation function estimate from its neighbor i:

[0087] ;

[0088] in, It is the input to the aggregation function estimation model of j.

[0089] Step S6 involves constructing a time-varying Nash equilibrium search model to determine the optimal power output strategy for the generator, specifically as follows:

[0090] The time-varying Nash equilibrium point search model for the i-th turbine generator is:

[0091] ;

[0092] in, , ; It is the control gain, which satisfies ; It is a parameter and , It is a parameter that satisfies ; and It is a constant; a set Let i represent the set of neighbors of i, and i exchanges information with its neighbors; this model can solve the time-varying Nash equilibrium point of constrained aggregation games, providing an optimal power output strategy for generators.

[0093] In step S7, the state-space representation of the turbine generator is constructed, and power control inputs are generated to drive the generator to output power according to a predetermined strategy, specifically as follows:

[0094] The state-space representation of the i-th turbine generator is: ;

[0095] in, , , and ;

[0096] remember It is the power control input sequence and It is the output sequence; it generates the power control input. The generator is driven to output power according to an optimized strategy to maximize profits (objective); power control input. for:

[0097] ;

[0098] in, , It is a diagonal matrix, where the diagonal elements are predetermined weight parameters. It is the discrete-time reference sequence at time k. and It is a matrix, and its specific form is: , .

[0099] Consider demand response management for four turbine generators in the electricity market; communication networks between generation systems, such as... Figure 2 As shown; the parameters of the power generation system are shown in Table 1, and , ,as well as The relevant parameters of the algorithm and noise estimation model are shown in Table 2; for the i-th turbine generator, there is random noise. and In the noise estimation model, parameters are selected. and Noise estimation gain The initial state is set as follows: For system 1, , For system 2, , For system 3, , For system 4, , In the aggregation function estimation model, an adaptive scalar is selected. , as well as In the dynamic Nash equilibrium search model, the following settings are made: Adaptive variables The initial value is selected as , Select the prediction time domain Control time domain Weight matrix , ,in yes An identity matrix of order 1.

[0100] Table 1 shows the parameters of the power generation system.

[0101]

[0102] Table 2 shows the parameters of the algorithm and noise estimation model.

[0103]

[0104] Assume the initial market power demand is 200 kilowatts; time The unit is hours ;when At that time, power demand ;when At that time, the market added 400 kW of power demand. ;when At that time, the market will have an additional power demand of 200 kW. ;when At that time, the market reduced its load by 300kW. ;when At that time, the market reduced the load by another 100 kW. .

[0105] To verify the effectiveness of the dynamic Nash equilibrium search model, the Lagrange multiplier method is introduced: by introducing Lagrange multipliers... Construct the Lagrange function:

[0106] ;

[0107] According to the Lagrange multiplier method, when and satisfy At this time It is the Nash equilibrium point of the aggregation game.

[0108] In this invention application, to avoid the complexity of solving the above equations, a dynamic Nash equilibrium point search model is constructed, using the Lyapunov function. Verifiable It converges to a bounded region of the origin, i.e. hour, The system converges to a bounded region of the origin, and after stabilization, it satisfies supply and demand equilibrium. Therefore, when At that time, estimated It is the Nash equilibrium point of the aggregation game.

[0109] like Figure 3 As shown, the aggregation function estimation model configured for each turbine generator can estimate the average power output. That is, the aggregation function, and the estimation error is bounded.

[0110] like Figure 4 As shown, by using continuous-time random noise Injecting actual privacy signals In the process, the signal transmitted in the network is obtained. Malicious eavesdroppers on the network can only steal information during transmission. ,and and The signals are inconsistent in size, so eavesdroppers cannot obtain actual private signals. .

[0111] like Figure 5 As shown, when market power demand changes, the turbine generator adjusts its power output. .

[0112] like Figure 6 and 7 As shown, once the system stabilizes, tending to zero and It approaches zero.

[0113] This invention application introduces privacy-protected triggering dynamic average consistency control, time-varying Nash equilibrium point search, and model predictive control to ensure that the total power output of the generator set meets market power demand in a dynamic environment, and adjusts the strategy to maximize the unit's revenue; at the same time, it saves communication resources and protects the privacy information of the turbine generator by interfering with the power value transmitted in the network through random noise.

[0114] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A privacy protection method for demand response management in the electricity market based on aggregation game theory, characterized in that: The method includes the following steps: Step S1: Construct a continuous-time dynamic system model of the turbine generator, determine the state-space equation of the turbine generator's power output, and obtain the generator's real-time state; Step S2: Construct an aggregate game model for multi-turbine generators to optimize generator profits, and use the aggregate function estimation model to assist in solving the problem; Step S3: Construct a continuous-time random noise generation model, superimpose its output onto the aggregation function estimation result to achieve privacy protection, and superimpose noise to generate a noise estimation signal; Step S4: Introduce an event-triggered mechanism to avoid continuous transmission, and transmit the noisy aggregation function estimation result together with the noise estimation signal to other designated generators; Step S5: Construct a noise estimation model, estimate the original noise based on the noise estimation signal, and obtain the input of the aggregation function estimation model; Step S6: Construct a time-varying Nash equilibrium point search model to determine the optimal power output strategy for the generator; Step S7: Construct the state-space representation of the turbine generator and generate power control inputs to drive the generator to output power according to a predetermined strategy.

2. The method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: The construction of the continuous-time dynamic system model of the turbine generator in step S1 is specifically as follows: In an undirected, connected communication network, consider N turbine generators. The dynamic model of the i-th turbine generator is as follows: ； in, It is the output power of the i-th generator system. It is the steam valve opening degree of the i-th generator. It is the relative speed of the i-th generator. The time constant of the i-th mechanical turbine It is the time constant of the speed controller of the i-th machine. It is the turbine gain of the i-th machine. It is the adjustment constant of the i-th machine. It is the speed of the synchronizer. It is the inertial constant. It is the unit damping constant. It is the power control input of the i-th generator.

3. The method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: The construction of the aggregation game model for the multi-turbine generator in step S2 is specifically as follows: In an undirected, interconnected communication network, considering N turbine generators operating in a profit-maximizing mode, the aggregation game model is defined as follows: ; The constraints are: ; in, Let be the power generation cost function of the i-th generator. This represents the cost coefficient. ; It is the aggregation function of aggregation game; Indicates electricity price, This indicates the degree to which local energy prices are affected by total power output; This represents the total power demand in the market.

4. The method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 3, characterized in that: In step S2, an aggregation function estimation model is used to assist in the solution. The aggregation function estimation model is as follows: ； in, It is the aggregation function of generator i on the market. Local estimates, yes To his neighbors aggregate functions The estimate, It is a scalar and satisfies , , , The aggregation function estimation model depends only on information from its neighbors.

5. The method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: In step S3, a continuous-time random noise generation model is constructed, and its output is superimposed on the aggregation function estimation result to achieve privacy protection. Furthermore, noise is superimposed to generate a noise estimation signal. Specifically: The continuous-time random noise generation model is as follows: ; in, It is a one-dimensional standard Brownian motion over a complete probability space; and Indicates the specified function; Represents a constant; The output of the random noise generation model represents the noise signal that is injected into privacy information to protect it from eavesdropping. Will This is superimposed on the aggregation function estimation result to ensure that the true information is not intercepted during transmission over an open network. The noise estimation signal is obtained by superimposing noise on it, and its expression is: 。 6. The method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: In step S4, an event-triggered mechanism is introduced to avoid continuous transmission. The noisy aggregation function estimation result, along with the noise estimation signal, is transmitted to other designated generators. Specifically: The event triggering mechanism for the i-th generator is determined by the following expression: ； in, It is a trigger function. and These are all measurement errors. , , It is a constant; At the k-th trigger moment of generator i, only when triggered will generator i transmit the information sequence. Send it to the designated generator j, thereby avoiding continuous communication and saving communication resources.

7. The method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: In step S5, a noise estimation model is constructed. The original noise is estimated based on the noise estimation signal to obtain the input of the aggregation function estimation model. Specifically: The random noise estimation model for the j-th generator is determined by the following expression: ； in, and These represent the noise levels of generator j relative to its neighbor i. and The estimate, For continuously differentiable bounded time-varying gain, Satisfying the Lipschitz condition , ; Generator j obtains the aggregation function estimate for i by subtracting the estimated noise from the noisy aggregation function estimate from its neighbor i: ； in, It is the input to the aggregation function estimation model of j.

8. A method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: Step S6 involves constructing a time-varying Nash equilibrium point search model to determine the optimal power output strategy for the generator, specifically as follows: The time-varying Nash equilibrium point search model for the i-th turbine generator is: ； in, , ; It is the control gain, which satisfies ; It is a parameter and , It is a parameter that satisfies ; and It is a constant; a set Let i represent the set of i's neighbors, and let i exchange information with its neighbors.

9. A method for triggering privacy protection in electricity market demand response management based on aggregation game theory as described in claim 1, characterized in that: The construction of the state-space representation of the turbine generator in step S7 is specifically as follows: The state-space representation of the i-th turbine generator is: ; in, , , and ; remember It is the power control input sequence and It is the output sequence; in order to make the i-th generator output power Reaching the time-varying Nash equilibrium point Power control input for: ； in, , It is a diagonal matrix, where the diagonal elements are predetermined weight parameters. It is the discrete-time reference sequence at time k. and It is a matrix, and its specific form is: , .

Images