Virtual power plant cooperative frequency modulation control method based on multi-agent system

Abstract

The invention discloses a virtual power plant cooperative frequency modulation control method based on a multi-agent system. The method comprises the steps of building a cooperative control architecture based on the multi-agent system; obtaining a power regulation requirement of an upper-level power grid automatic generation control system; based on the power regulation requirement, calculating the optimal reference regulation power of the virtual power plant through a virtual power plant cooperative power control and information interaction mechanism joint design scheme; and based on the optimal reference adjustment power of the virtual power plant, controlling an inverter through a dynamic power control method to adjust the output of renewable energy sources in the virtual power plant so as to meet the power adjustment requirement of a superior power grid, thereby realizing the frequency modulation target of the power grid. According to the method, the operation economy can be effectively improved while the virtual power plant is utilized to assist in supporting power grid frequency modulation, the information interaction safety is ensured, the inverter control stability is ensured, and the frequency modulation reliability is improved.

Description

Technical Field

[0001] The present invention relates to a virtual power plant collaborative frequency regulation control method based on a multi-agent system, and belongs to the technical field of power system operation and control. Background Art

[0002] In recent years, more and more renewable energy sources are being connected to the power system, and the synchronous inertia of the power system is constantly decreasing, which limits the regulation capacity of traditional generators and may not be able to support the frequency regulation of the power grid under the large-scale access of renewable energy. At the same time, since there are a large number of small-capacity distributed resources for frequency regulation on the distribution side, such as renewable energy wind power and photovoltaics, using virtual power plants to aggregate and coordinate these distributed resources to support grid frequency regulation has become an emerging alternative solution.

[0003] At present, the research on virtual power plants participating in frequency regulation mainly focuses on designing resource optimization scheduling and bidding and auction strategies to improve the stability of system frequency, while there are few studies on supporting system frequency regulation through collaborative power control of virtual power plants based on multi-agent systems. In addition, during the collaborative frequency regulation process, virtual power plants may be affected by diversified communication network problems, such as network attacks, privacy leaks, etc., which brings huge challenges to the frequency regulation of virtual power plants. However, existing studies rarely consider the impact of diversified communication network problems on the collaborative frequency regulation of virtual power plants. Therefore, the present invention proposes a collaborative frequency regulation control method for virtual power plants based on multi-agent systems that takes into account diversified communication network problems. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a virtual power plant collaborative frequency regulation control method based on a multi-agent system, which can effectively improve the economy of its operation while utilizing the virtual power plant to assist in supporting the frequency regulation of the power grid, ensure the security of its information interaction, ensure the stability of inverter control, and improve the reliability of its frequency regulation.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] The present invention provides a virtual power plant collaborative frequency regulation control method based on a multi-agent system, comprising:

[0007] Build a collaborative control architecture based on a multi-agent system;

[0008] Obtain the power regulation requirements of the upper-level power grid automatic power generation control system;

[0009] Based on the power regulation requirements, the optimal reference regulation power of the virtual power plant is calculated through the joint design of the virtual power plant collaborative power control and information interaction mechanism;

[0010] Based on the optimal reference regulation power of the virtual power plant, the inverter is controlled by the dynamic power control method to adjust the output of renewable energy inside the virtual power plant to meet the power regulation needs of the upper power grid, thereby achieving the grid frequency regulation target.

[0011] Furthermore, the collaborative control architecture based on the multi-agent system includes a physical grid and an information grid. The physical grid includes multiple virtual power plants composed of renewable energy sources such as wind power and photovoltaics. The information grid includes a virtual power plant control agent for implementing collaborative power control and a renewable energy unit agent for performing inverter dynamic power control.

[0012] Furthermore, the virtual power plant collaborative power control and information interaction mechanism joint design scheme includes:

[0013] Power control optimization modeling, including: objective function and constraints;

[0014] Objective function:

[0015]

[0016] Among them, V is the set of virtual power plants, a i ,b i ,c i is the cost coefficient of virtual power plant i, P i v (t) is the power output of virtual power plant i at time t;

[0017] Constraints:

[0018]

[0019] Among them, P i v,l (t) is the power loss of virtual power plant i at time t, P a (t) is the power regulation requirement of the upper power grid automatic power generation control system, and is the upper and lower limits of the power output of virtual power plant i at time t, ΔP i v (t) is the change in power output of virtual power plant i at time t, and are the upper and lower limits of the power output ramp rate of virtual power plant i at time t;

[0020] Coordinated power control strategies, including:

[0021] Update the state λ of the virtual power plant control agent i as follows: i :

[0022]

[0023] Where k is the number of iterations, For collection The base number, in, and is the state information range of virtual power plant control agent i, is the state sent by the neighbor agent j of the virtual power plant control agent i for the kth time, N i is the neighbor set of virtual power plant control agent i, β k is the step length, f i ′(k) is the objective function f of the virtual power plant control agent i i (k) is the gradient, μ is the deviation coefficient, ξ i (k) is the deviation of the virtual power plant control agent i;

[0024] Update the power output P of the virtual power plant control agent i as follows: i v (k+1):

[0025]

[0026] Information exchange mechanisms include:

[0027] Initialize the interactive state information λ of the virtual power plant control agent i(j) i(j) , randomly generate weight parameter d i(j)→j(i) ;

[0028] Virtual power plant control agent i sends its encrypted negative state E to virtual power plant control agent j i (-λ i ) and the public key E i ;

[0029] Then the virtual power plant control agent j calculates the weight difference E i (d j→i (λ j -λ i )),include:

[0030] Using the virtual power plant to control the public key E of agent i i Encryption Lambda j , that is, j →E i (λ j );

[0031] Compute the difference in the ciphertext:

[0032] E i (λ j -λi )=E i (λ j +(-λ i ))=E i (λ j )*E i (-λ i ) (7);

[0033] Compute the weighted difference in the ciphertext:

[0034]

[0035] Then the virtual power plant control agent j sends E i (d j→i (λ j -λ i )), the virtual power plant control agent i accepts and uses the private key D i Decrypt it and compare it with the weight d i→j Multiply them together to get the weighted difference d i→j d j→i (λ j -λ i );

[0036] After each VPP control agent collects the weighted differences from all its neighbors, it updates its state accordingly using (5).

[0037] Furthermore, the dynamic power control method comprises:

[0038] Constructing a renewable energy state space model:

[0039]

[0040] Δy c =C c Δx c (10);

[0041] Where Δx c , Δy c is the incremental state and output vector, Δu c is the incremental input vector, A c , B c , C c They are the state transfer matrix, input control matrix and output matrix respectively;

[0042] Using the sampling time T s Discretizing the model, we get:

[0043] Δx m (k+1)=A m Δx m (k)+Bm Δu(k) (11);

[0044] Δy m (k) = C m Δx m (k) (12);

[0045] Where Δx m , Δy m is the incremental state and output vector after discretization, Δu is the incremental input vector after discretization, is the discretized state transfer matrix, e is the base of the natural logarithm, is the input control matrix, I is the identity matrix, C m =C c is the discretized output matrix;

[0046] The model is further rewritten as:

[0047] l(k+1)=Al(k)+BΔu(k) (13);

[0048] Δy(k)=Cl(k) (14);

[0049] Among them, l is the new state vector, Δy is the new incremental output vector, C=[C m 0] are the new state transfer matrix, input control matrix and output matrix respectively;

[0050] The dynamic power control target J is:

[0051]

[0052] Among them, T is the matrix transpose, n is the time, N p is the prediction period, l(k+n|k) is the state quantity predicted at time k+n, Δu(k+n|k) is the incremental input quantity predicted at time k+n, and W′ and V′ represent the corresponding weight matrices;

[0053] The constraints are:

[0054]

[0055] Among them, Δu max With y max are the maximum ranges of incremental input and control output respectively;

[0056] Then the renewable energy reference power control law Δu is calculated by the following inequality:

[0057]

[0058] Where K = ηH(k) -1 is the auxiliary matrix, H(k) is a positive definite matrix, and η is the maximum range of the objective function J;

[0059]

[0060] Wherein, L=G(k)K is the auxiliary matrix, and G(k) is the feedback gain matrix;

[0061]

[0062] Among them, M and N are symmetric matrices, and finally G(k)=LK is calculated by the obtained K and L -1 , thus obtaining the power control law Δu(k+n|k)=G(k)l(k+n|k).

[0063] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the collaborative frequency control method of a virtual power plant based on a multi-agent system when executing the program.

[0064] A computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the virtual power plant collaborative frequency control method based on a multi-agent system.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The present invention provides a virtual power plant collaborative frequency regulation control method based on a multi-agent system. The optimal reference regulation power of the virtual power plant is calculated through a joint design scheme of collaborative power control of the virtual power plant and an information interaction mechanism. The dynamic power control method is used to control the inverter to adjust the output of renewable energy in the virtual power plant. While using the virtual power plant to assist in supporting the frequency regulation of the power grid, the economy of its operation can be effectively improved, the security of its information interaction can be guaranteed, the stability of the inverter control can be guaranteed, and the reliability of its frequency regulation can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of a virtual power plant collaborative frequency regulation control method based on a multi-agent system provided by an embodiment of the present invention;

[0068] Figure 2 is a power deviation comparison diagram provided by an embodiment of the present invention;

[0069] Figure 3 is a comparison chart of total operating costs provided by an embodiment of the present invention;

[0070] Figure 4It is a comparison diagram of dynamic response results provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0071] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0072] Embodiment 1: This embodiment introduces a virtual power plant collaborative frequency regulation control method based on a multi-agent system, including:

[0073] Build a collaborative control architecture based on a multi-agent system;

[0074] Obtain the power regulation requirements of the upper-level power grid automatic power generation control system;

[0075] Based on the power regulation requirements, the optimal reference regulation power of the virtual power plant is calculated through the joint design of the virtual power plant collaborative power control and information interaction mechanism;

[0076] Based on the optimal reference regulation power of the virtual power plant, the inverter is controlled by the dynamic power control method to adjust the output of renewable energy inside the virtual power plant to meet the power regulation needs of the upper power grid, thereby achieving the grid frequency regulation target.

[0077] like Figure 1 As shown, the virtual power plant collaborative frequency regulation control method based on the multi-agent system provided in this embodiment specifically involves the following steps:

[0078] (1) Build a collaborative control architecture based on a multi-agent system;

[0079] (2) Obtaining the power regulation requirements of the upper-level power grid automatic power generation control system;

[0080] (3) Based on the power regulation requirements, the optimal reference regulation power of the virtual power plant is calculated through the joint design of the virtual power plant collaborative power control and information interaction mechanism;

[0081] (4) Based on the optimal reference power regulation of the virtual power plant, the inverter is controlled by a dynamic power control method to adjust the output of renewable energy within the virtual power plant to meet the power regulation requirements of the upper power grid, thereby achieving the grid frequency regulation target.

[0082] The specific steps include:

[0083] Step 1: Build a collaborative control architecture based on a multi-agent system.

[0084] The collaborative control architecture based on the multi-agent system involved includes a physical grid and an information grid. The physical grid includes multiple virtual power plants composed of renewable energy sources such as wind power and photovoltaics. The information grid includes a virtual power plant control agent for implementing collaborative power control and a renewable energy unit agent for performing inverter dynamic power control.

[0085] Step 2: Obtain the power regulation requirements of the upper-level power grid automatic power generation control system.

[0086] Step 3: Based on the power regulation requirements, the optimal reference regulation power of the virtual power plant is calculated through a joint design scheme of virtual power plant collaborative power control and information interaction mechanism.

[0087] Step 3.1: Power control optimization modeling, including: objective function and constraints;

[0088] Objective function:

[0089]

[0090] Among them, V is the set of virtual power plants, a i ,b i ,c i is the cost coefficient of virtual power plant i, P i v (t) is the power output of virtual power plant i at time t;

[0091] Constraints:

[0092] Σ i∈V P i v (t)-Σ i∈V P i v,l (t) = P a (t) (2);

[0093]

[0094] Among them, P i v,l (t) is the power loss of virtual power plant i at time t, P a (t) is the power regulation requirement of the upper power grid automatic power generation control system, and is the upper and lower limits of the power output of virtual power plant i at time t, ΔP i v (t) is the change in power output of virtual power plant i at time t, and are the upper and lower limits of the power output ramp rate of virtual power plant i at time t;

[0095] Step 3.2: Coordinated power control strategy implementation, including:

[0096] Update the state λ of the virtual power plant control agent i as follows: i :

[0097]

[0098] Where k is the number of iterations, For collection The base number, in, and is the state information range of virtual power plant control agent i, is the state sent by the neighbor agent j of the virtual power plant control agent i for the kth time, N i is the neighbor set of virtual power plant control agent i, β k is the step length, f i ′(k) is the objective function f of the virtual power plant control agent i i (k) is the gradient, μ is the deviation coefficient, ξ i (k) is the deviation of the virtual power plant control agent i;

[0099] Update the power output P of the virtual power plant control agent i as follows: i v (k+1):

[0100] Step 3.3: Deployment of information exchange mechanism, including:

[0101] Initialize the interactive state information λ of the virtual power plant control agent i(j) i(j) , randomly generate weight parameter d i(j)→j(i) ;

[0102] Virtual power plant control agent i sends its encrypted negative state E to virtual power plant control agent j i (-λ i ) and the public key E i ;

[0103] Then the virtual power plant control agent j calculates the weight difference E i (d j→i (λ j -λ i )),include:

[0104] Using the virtual power plant to control the public key E of agent i i Encryption Lambda j , that is, j →E i (λ j );

[0105] Compute the difference in the ciphertext:

[0106] E i (λ j -λ i )=E i (λ j +(-λ i ))=E i (λ j )*E i (-λ i ) (7);

[0107] Compute the weighted difference in the ciphertext:

[0108]

[0109] Then the virtual power plant control agent j sends E i (d j→i (λ j -λ i )), the virtual power plant control agent i accepts and uses the private key D i Decrypt it and compare it with the weight d i→j Multiply them together to get the weighted difference d i→j d j→i (λ j -λ i );

[0110] After each VPP control agent collects the weighted differences from all its neighbors, it updates its state accordingly using (5).

[0111] In this embodiment, the software MATLAB is used to complete the writing and running of the proposed collaborative power control strategy code under encryption operation. In order to test the convergence performance of the proposed strategy, Figure 2 and Figure 3 The comparison results of power deviation and total operating cost under different methods are given respectively, among which comparison method 1 adopts the consistency-based control method under network attack, and comparison method 2 is the consistency-based control method without considering network attack. Figure 2 It can be seen that the proposed method can converge to the optimal value, while the comparison method 1 cannot converge, and the comparison method 2 can converge but the power deviation is large. Figure 3 It can be seen that compared with the above comparison methods, the total operating cost of the proposed method is very close to the optimal value. Therefore, according to the simulation results, the proposed method can effectively improve the economic efficiency of its operation while using virtual power plants to assist in supporting grid frequency regulation and ensure the security of its information interaction.

[0112] Step 4: Based on the optimal reference power regulation of the virtual power plant, the inverter is controlled by the dynamic power control method to adjust the output of renewable energy inside the virtual power plant to meet the power regulation requirements of the upper-level power grid, thereby achieving the power grid frequency regulation target.

[0113] Step 4.1: Construct a renewable energy state space model, including:

[0114]

[0115] Δy c =C c Δx c (10);

[0116] Where Δx c , Δy c is the incremental state and output vector, Δu c is the incremental input vector, A c , B c , C c They are the state transfer matrix, input control matrix and output matrix respectively;

[0117] Using the sampling time T s Discretizing the model, we get:

[0118] Δx m (k+1)=A m Δx m (k)+B m Δu(k) (11);

[0119] Δy m (k) = C m Δx m (k) (12);

[0120] Where Δx m , Δy m is the incremental state and output vector after discretization, Δu is the incremental input vector after discretization, is the discretized state transfer matrix, e is the base of the natural logarithm, is the input control matrix, I is the identity matrix, C m =C c is the discretized output matrix;

[0121] The model is further rewritten as:

[0122] l(k+1)=Al(k)+BΔu(k) (13);

[0123] Δy(k)=Cl(k) (14);

[0124] Among them, l is the new state vector, Δy is the new incremental output vector, C=[C m 0] are the new state transfer matrix, input control matrix and output matrix respectively;

[0125] Step 4.2: The dynamic power control target J is:

[0126]

[0127] Among them, T is the matrix transpose, n is the time, N p is the prediction period, l(k+n|k) is the state quantity predicted at time k+n, Δu(k+n|k) is the incremental input quantity predicted at time k+n, and W′ and V′ represent the corresponding weight matrices;

[0128] The constraints are:

[0129]

[0130] Among them, Δu max With y max are the maximum ranges of incremental input and control output respectively;

[0131] Step 4.3: Calculate the renewable energy reference power control law Δu by the following inequality:

[0132]

[0133] Where K = ηH(k) -1 is the auxiliary matrix, H(k) is a positive definite matrix, and η is the maximum range of the objective function J;

[0134]

[0135] Wherein, L=G(k)K is the auxiliary matrix, and G(k) is the feedback gain matrix;

[0136]

[0137] Among them, M and N are symmetric matrices, and finally G(k)=LK is calculated by the obtained K and L -1 , thus obtaining the power control law Δu(k+n|k)=G(k)l(k+n|k).

[0138] In this embodiment, by Figure 4 The dynamic response comparison results in Figure 2 verify the effectiveness of the proposed dynamic power control method, where the comparison method 3 uses the traditional power tracking method based on model predictive control. Figure 4As shown in the figure, it can be seen that although both methods can track the reference power well, the comparison method will produce a larger overshoot during the control process. Therefore, the simulation results show that the proposed method can well ensure the stability of inverter control, thereby improving the reliability of frequency regulation of virtual power plants.

[0139] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A virtual power plant collaborative frequency control method based on a multi-agent system, characterized in that: include: Build a collaborative control architecture based on a multi-agent system; Obtain the power regulation requirements of the upper-level power grid automatic power generation control system; Based on the power regulation requirements, the optimal reference regulation power of the virtual power plant is calculated through the joint design of the virtual power plant collaborative power control and information interaction mechanism; Based on the optimal reference regulation power of the virtual power plant, the inverter is controlled by the dynamic power control method to adjust the output of renewable energy inside the virtual power plant to meet the power regulation needs of the upper power grid, thereby achieving the grid frequency regulation target.

2. The virtual power plant collaborative frequency control method based on multi-agent system according to claim 1 is characterized in that: The collaborative control architecture based on the multi-agent system includes a physical grid and an information grid, wherein the physical grid includes multiple virtual power plants composed of renewable energy sources including wind power and photovoltaics, and the information grid includes a virtual power plant control agent for implementing collaborative power control and a renewable energy unit agent for performing inverter dynamic power control.

3. The virtual power plant collaborative frequency control method based on multi-agent system according to claim 2 is characterized in that: The virtual power plant collaborative power control and information interaction mechanism joint design scheme includes: Power control optimization modeling, including: objective function and constraints; Objective function: Among them, V is the set of virtual power plants, a i ,b i ,c i is the cost coefficient of virtual power plant i, P i v (t) is the power output of virtual power plant i at time t; Constraints: in, is the power loss of virtual power plant i at time t, P a (t) is the power regulation requirement of the upper grid automatic power generation control system, and are the upper and lower limits of the power output of virtual power plant i at time t, is the change in power output of virtual power plant i at time t, and are the upper and lower limits of the power output ramp rate of virtual power plant i at time t; Coordinated power control strategies, including: Update the state λ of the virtual power plant control agent i as follows: i : Where k is the number of iterations, For collection The base number, in, and is the state information range of virtual power plant control agent i, is the state sent by the neighbor agent j of the virtual power plant control agent i for the kth time, N i is the neighbor set of virtual power plant control agent i, β k is the step length, f i ′(k) is the objective function f of the virtual power plant control agent i i (k) is the gradient, μ is the deviation coefficient, ξ i (k) is the deviation of the virtual power plant control agent i; Update the power output of virtual power plant control agent i as follows: Information exchange mechanisms include: Initialize the interactive state information λ of the virtual power plant control agent i(j) i(j) , randomly generate weight parameter d i(j)→j(i) ; Virtual power plant control agent i sends its encrypted negative state E to virtual power plant control agent j i (-λ i ) and the public key E i ; Then the virtual power plant control agent j calculates the weight difference E i (d j→i (λ j -λ i )),include: Using the virtual power plant to control the public key E of agent i i Encryption Lambda j , that is, j →E i (λ j ); Compute the difference in the ciphertext: E i (l j -l i )=E i (l j +(-λ i ))=E i (l j )*E i (-l i ) (7); Compute the weighted difference in the ciphertext: Then the virtual power plant control agent j sends E i (d j→i (λ j -λ i )), the virtual power plant control agent i accepts and uses the private key D i Decrypt it and compare it with the weight d i→j Multiply them together to get the weighted difference d i→j d j→i (λ j -λ i ); After each VPP control agent collects the weighted differences from all its neighbors, it updates its state accordingly using (5).

4. The virtual power plant collaborative frequency control method based on multi-agent system according to claim 3 is characterized in that: The dynamic power control method comprises: Constructing a renewable energy state space model: Δy c =C c Δx c (10); Where Δx c , Δy c is the incremental state and output vector, Δu c is the incremental input vector, A c , B c , C c They are the state transfer matrix, input control matrix and output matrix respectively; Using the sampling time T s Discretizing the model, we get: Δx m (k+1)=A m Δx m (k)+B m Δu(k) (11); Δy m (k)=C m Δx m (k) (12); Where Δx m , Δy m is the incremental state and output vector after discretization, Δu is the incremental input vector after discretization, is the discretized state transfer matrix, e is the base of the natural logarithm, is the input control matrix, I is the identity matrix, C m =C c is the discretized output matrix; Rewrite the model as: l(k+1)=Al(k)+BΔu(k) (13); Δy(k)=Cl(k) (14); Among them, l is the new state vector, Δy is the new incremental output vector, C=[C m 0] are the new state transfer matrix, input control matrix and output matrix respectively; The dynamic power control target J is: Among them, T is the matrix transpose, n is the time, N p is the prediction period, l(k+n|k) is the state quantity predicted at time k+n, Δu(k+n|k) is the incremental input quantity predicted at time k+n, and W′ and V′ represent the corresponding weight matrices; The constraints are: Among them, Δu max With y max are the maximum ranges of incremental input and control output respectively; Then the renewable energy reference power control law Δu is calculated by the following inequality: Where K = ηH(k) -1 is the auxiliary matrix, H(k) is a positive definite matrix, and η is the maximum range of the objective function J; Wherein, L=G(k)K is the auxiliary matrix, and G(k) is the feedback gain matrix; Among them, M and N are symmetric matrices, and finally G(k)=LK is calculated by the obtained K and L -1 , thus obtaining the power control law Δu(k+n|k)=G(k)l(k+n|k).

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements the virtual power plant collaborative frequency control method based on the multi-agent system as described in any one of claims 1 to 4 above.

6. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instruction is executed by the processor, the collaborative frequency control method of a virtual power plant based on a multi-agent system as described in any one of claims 1 to 4 is implemented.

Images