Multi-energy virtual power plant energy distributed optimization management method considering network attack

Abstract

The invention discloses a multi-energy virtual power plant energy distributed optimization management method considering network attacks, which relates to the technical field of virtual power plant management and comprises the steps of establishing an electricity-heat-hydrogen multi-energy virtual power plant energy management model considering a heterogeneous energy network topology structure, generating a target function and determining constraint conditions. And establishing a subject credibility evaluation model based on an observation subnet to perform behavior evaluation and isolation on network nodes in the energy management model of the electricity-heat-hydrogen multi-energy virtual power plant, and performing iterative solution. According to the method, the main body credibility evaluation model is established, the network nodes which are converted into malicious false nodes due to network attacks are inquired, the credibility of the network nodes corresponding to each main body in the multi-energy virtual power plant management model is evaluated, and node isolation is carried out; the phenomenon that false nodes transmit false information to adjacent nodes is avoided, the possibility that distributed energy main bodies corresponding to other network nodes make wrong decisions is reduced, and solving correctness can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual power plant management, and in particular to a method for energy distributed optimization management of a multi-energy virtual power plant taking network attacks into account. Background Art

[0002] With the rapid expansion of distributed energy, virtual power plant technology has gradually been regarded as an effective means to aggregate scattered resources in a region and achieve efficient energy management. With the rise of energy Internet technology, the demand for multi-energy heterogeneous resources and their complementarity is increasing, which also promotes the evolution of traditional virtual power plants to multi-energy virtual power plants. In order to realize the management and scheduling of heterogeneous resources, a multi-energy virtual power plant energy management method for multi-energy virtual power plants is designed. However, as a typical information-physical system, the current multi-layer virtual power plant is at high risk of being attacked by information network. When the identity information of the node in the system is stolen by an external attacker for some reason, the attacker can forge its identity and generate a malicious node. The malicious node can destroy the decision-making behavior of other real entities by sending false information to adjacent entities, thereby causing damage to the system, reducing the correctness of the decisions made by the multi-energy virtual power plant management method for energy management and scheduling.

[0003] The defects of the existing power plant management methods are:

[0004] 1. The patent document CN111682526A mainly considers how to allocate and manage the energy of the virtual power plant, thereby improving the overall benefits of the virtual power plant, but does not consider how to prevent the main body of the multi-energy virtual power plant from being attacked by the network and reducing the correctness of management decisions;

[0005] 2. Patent document CN118195338B mainly considers how to improve the overall operating efficiency of building virtual power plants and reduce the peak-to-valley difference of distribution network tie lines, but does not consider how to improve the data privacy in the energy management model of multi-energy virtual power plants;

[0006] 3. Patent document CN102509167B mainly considers how to improve the reliability and economy of photovoltaic power generation systems, but does not consider how to reduce the differences between the physical characteristics of different energy systems and reduce information barriers;

[0007] 4. Patent document CN110991919A mainly considers how to improve the reliability and stability of energy management and scheduling in virtual power plants, but does not consider how to reduce the computing burden. Summary of the invention

[0008] The purpose of the present invention is to provide a method for distributed optimization management of energy of a multi-energy virtual power plant taking into account network attacks, so as to solve the problems raised in the above-mentioned background technology.

[0009] To achieve the above-mentioned purpose, the present invention provides the following technical solution: a multi-energy virtual power plant energy distributed optimization management method taking into account network attacks, comprising the following steps:

[0010] S1. Based on multi-energy heterogeneous energy, an energy management model of an electric-heat-hydrogen multi-energy virtual power plant is established taking into account the topological structure of heterogeneous energy networks;

[0011] S2. With the goal of minimizing the operating cost of distributed resources, construct an objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant, and determine the constraints of the objective function;

[0012] S3. Establishing a subject reputation evaluation model based on the observation subnet, wherein the subject reputation evaluation model is used to evaluate the behavior of network nodes in the electric-heat-hydrogen multi-energy virtual power plant energy management model, and isolate malicious false nodes according to the evaluation results;

[0013] S4. For non-malicious false nodes, based on the objective function and the constraints, the energy management model of the electric-thermal-hydrogen multi-energy virtual power plant is iteratively solved using a collaborative neural dynamics optimization method combined with an improved Push-sum protocol;

[0014] S5. Repeat the iterative solution process to obtain the optimization result.

[0015] Preferably, S3 specifically includes the following steps:

[0016] S31, establishing an observation subnet based on the communication network foundation, when the network node i transmits information to the network node j, the transmitted information can be sent from the network node i to the two-hop neighbor of the network node j through the observation subnet;

[0017] Based on the observation subnet, a subject reputation evaluation model is established. According to the information transmission behavior between network nodes, the subject reputation evaluation model is used to evaluate the behavior of network nodes in the energy management model of the electric-heat-hydrogen multi-energy virtual power plant. The specific steps are as follows:

[0018] The energy mismatch of the local estimate of the network node i is calculated as follows:

[0019]

[0020] Wherein, z(k, v) represents the detection threshold function determined by the number of iterations k and the maximum ramp rate v of the energy device, and the detection threshold function satisfies the following relationship:

[0021]

[0022] Function Y + (a) and Y - (a) are defined as follows:

[0023]

[0024] in, is the sum of the weights of the edges connecting network node j and all its outgoing neighbors in the initial communication network graph;

[0025] S32. During the kth iteration, network node i calculates y according to equations (1) and (2). v (k+1) and compare the reasonable boundary value with the actual value received by the network node i. The network node behavior is evaluated according to the comparison result, and the evaluation method is as follows:

[0026] when When it is established, network node i considers network node v to be of good behavior, otherwise it considers network node v to be of malicious behavior and records network node v as a malicious false node;

[0027] According to the node behavior, update the detection result variable d of the network node v v (k), the detection result variable d v (k) is transmitted from network node i to network node j through the observation subnet, and network node j is an incoming neighbor of network node v, d v (k) satisfies the following relationship:

[0028]

[0029] Preferably, S3 also includes:

[0030] S33. Based on the network node behavior evaluation results, node reputation evaluation and malicious false nodes isolation are performed, specifically including the following steps:

[0031] In each iteration, network node i counts the counter T according to the detection result of network node v. iv (k) is updated, and the initial value of the counter is T iv (0) = 0, T iv (k) satisfies the following relationship:

[0032]

[0033] A subject reputation evaluation model is established. Network node i uses the subject reputation evaluation model to evaluate the node reputation of network node v and obtain the node reputation evaluation value. The node reputation evaluation value satisfies the following relationship:

[0034]

[0035] Rep iv (k)∈[0,1];

[0036]

[0037] Set a node reputation threshold, compare the node reputation evaluation value with the node reputation threshold, and when the node reputation evaluation value is less than the node reputation threshold, disconnect the communication link between the malicious false node and its neighbor node to isolate the malicious false node;

[0038] The initial communication network diagram is updated based on the isolation results and the communication link disconnection results to obtain the actual communication network diagram, which is used to perform the solution step of S4 in combination with the objective function and constraint conditions in S2.

[0039] Preferably, in S2, the objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant is constructed with the goal of minimizing the operating cost of distributed resources to satisfy the following relationship:

[0040]

[0041]

[0042] Ax = b;

[0043] G(x)≤0 N .

[0044] Preferably, S2 also includes constraints for determining the objective function, and the constraints include output power constraints in the decision variable vector, equipment climbing power constraints, feasible domain constraints of cogeneration units, energy storage equipment operation constraints, ideal charging range constraints for electric vehicles, power network line flow constraints, heat network node pipeline temperature and flow constraints, and hydrogen network node pipeline pressure and flow constraints.

[0045] Preferably, the iterative solution of the electric-heat-hydrogen multi-energy virtual power plant energy management model in S4 includes solving the electric-heat-hydrogen multi-energy virtual power plant energy management model based on the objective function and constraints according to the actual network communication diagram, and the solution process includes the following steps:

[0046] Decompose the objective function and constraints into multiple groups of sub-problems;

[0047] Each group of sub-problems is converted into the form of Lagrangian function, and each node body solves the sub-problem converted into Lagrangian function, and the sub-problem is solved by a single-layer projection neural network;

[0048] Network node i and network node j exchange information, and network node i locally backs up the information of network node j obtained through the information exchange. The information exchange adopts an improved Push-sum protocol for privacy protection.

[0049] Preferably, the network node i and the network node j perform information exchange, the network node i performs a local backup of the information of the network node j obtained through the information exchange, and the information exchange adopts an improved Push-sum protocol for privacy protection, including the following steps:

[0050] For the initial variable information x to be exchanged i (0), network node i randomly generates an initial sub-state value from U(-M, M) And initialize the remaining sub-state variables, and initialize the remaining sub-state variables to satisfy the following relationship:

[0051]

[0052] Where M is a positive constant;

[0053] Generate sub-state weights. When k = 0, network node i randomly selects an initial weight set from the distribution Ψ(0, M) The weight normalization process is performed, and the weight normalization process satisfies the following relationship:

[0054]

[0055] Among them When ji (0) = 0;

[0056] When k ≥ 1, network node i generates a random number α that follows uniform distribution U(0, 1) i (K), and generate weights ω from the distribution U(0, 1) ji (k), weight normalization is performed so that and When ji (0) = 0;

[0057] Perform sub-state status updates. For all k ≥ 0, after network node i receives information from neighbor node j, it updates two sub-states. and The updates of the two sub-states satisfy the following relationship:

[0058]

[0059] Network node i calculates the average value based on the updated sub-state, and the calculation expression is as follows:

[0060]

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. The present invention establishes a subject reputation evaluation model for the observation subnet to promptly detect network nodes that have been transformed into malicious false nodes due to network attacks, and uses the subject reputation evaluation model to evaluate the reputation of the network nodes corresponding to each physical subject in the multi-energy virtual power plant management model, obtains the node reputation evaluation value, compares it with the node reputation threshold, and chooses to disconnect the communication link between the malicious false node and its neighboring node based on the comparison result, isolates the malicious false node, avoids the false node from transmitting false information to its adjacent network nodes, and reduces the possibility of distributed energy subjects corresponding to other network nodes making wrong decisions, which is beneficial to improving the correctness of solving the objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant, and thus improving the correctness of the decision.

[0063] 2. The present invention uses an improved Push-sum protocol to protect the privacy of the information exchange process during information exchange among network nodes. That is, node i weights the random sub-states of the state variables and sends them to neighboring nodes instead of directly exposing the actual data. This effectively ensures data privacy and reduces the possibility of privacy leakage during the information exchange process. The collaborative neural dynamics optimization method combined with the improved Push-sum protocol is used to iteratively solve the energy model of the electric-thermal-hydrogen multi-energy virtual power plant, thereby improving the computational security in the iterative solution process of the objective function.

[0064] 3. The present invention establishes an energy management model of an electric-heat-hydrogen multi-energy virtual power plant taking into account heterogeneous energy sources dispersed in a region, and the management model includes an actual network topology. Through the existing communication infrastructure, each distributed multi-energy heterogeneous energy point is used as a communication node to form an actual communication network diagram, thereby realizing mutual communication between the operating and managing parties of the distributed heterogeneous energy points in the energy management model of the electric-heat-hydrogen multi-energy virtual power plant, reducing the differences between the physical characteristics of different energy systems, and facilitating energy management and optimized scheduling within the multi-energy virtual power plant.

[0065] 4. The present invention constructs an objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant with the goal of minimizing the operating cost of distributed resources, which is used to solve the management and energy optimization scheduling plan of the multi-energy virtual power plant. The obtained plan minimizes the daily operating cost of the multi-energy virtual power plant, which is conducive to maximizing the economic benefits of the multi-energy virtual power plant. In the solution process, the objective function is converted into multiple groups of sub-problems with equality and inequality constraints, and each sub-problem is converted into the form of Lagrangian function, which facilitates the solution of inequality constraints and objective functions and reduces the computational burden. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 is a flow chart of the steps of the present invention;

[0067] Figure 2 It is a graph of the objective function and constraint conditions of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] See also Figure 1 and Figure 2 The present invention provides an embodiment: a multi-energy virtual power plant energy distribution optimization management method taking into account network attacks, comprising:

[0070] S1. Based on multi-energy heterogeneous energy, an energy management model of an electric-heat-hydrogen multi-energy virtual power plant is established taking into account the topological structure of heterogeneous energy networks;

[0071] S2. With the goal of minimizing the operating cost of distributed resources, construct an objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant, and determine the constraints of the objective function;

[0072] S3. Establishing a subject reputation evaluation model based on the observation subnet, wherein the subject reputation evaluation model is used to evaluate the behavior of network nodes in the electric-heat-hydrogen multi-energy virtual power plant energy management model, and isolate malicious false nodes according to the evaluation results;

[0073] S4. For non-malicious false nodes, based on the objective function and the constraint conditions, use the improved Push-sum protocol

[0074] The collaborative neural dynamics optimization method is used to iteratively solve the energy management model of the electric-thermal-hydrogen multi-energy virtual power plant;

[0075] S5. Repeat the iterative solution process to obtain the optimization result.

[0076] Furthermore, in view of the distributed heterogeneous energy sources scattered in the region, an electric-thermal-hydrogen multi-energy virtual power plant energy management model taking into account the heterogeneous energy is established, and the management model includes an actual network topology structure. Through the existing communication infrastructure, each distributed multi-energy heterogeneous energy point is used as a communication node to form an actual communication network diagram, so as to realize mutual communication between the operating and managing parties of the distributed heterogeneous energy points in the electric-thermal-hydrogen multi-energy virtual power plant energy management model, reduce the differences between the physical characteristics of different energy systems, and facilitate energy management and optimized scheduling within the multi-energy virtual power plant.

[0077] See also Figure 1 and Figure 2 The present invention provides an embodiment: a method for distributed optimization management of energy of a multi-energy virtual power plant taking into account network attacks, comprising constructing an objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant with the goal of minimizing the operating cost of distributed resources, and determining the constraints of the objective function, wherein the specific steps are as follows:

[0078] With the goal of minimizing the operating cost of distributed resources, the objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant is constructed to satisfy the following relationship:

[0079]

[0080] Ax = b;

[0081] G(x)≤0 N ;

[0082] Among them, f total is the overall operating cost of the electricity-heat-hydrogen multi-energy virtual power plant (MEVPP), Ω CG is the collection of thermal power units, is the operating cost function of the thermal power unit, Ω DRG is a collection of distributed renewable power generation units, is the output penalty cost function of distributed renewable generation units, Ω ESS is a collection of electrical energy storage units, is the operating cost function of the electric energy storage unit, Ω EV For the collection of electric vehicles, is the charging utility loss cost function of electric vehicles, Ω EL An electrolyzer assembly for hydrogen production, is the operating cost function of the electrolyzer, Ω CHP is a collection of combined heat and power units, is the operating cost of the combined heat and power unit, Ω EB For the collection of electric boiler units, is the operating cost of the electric boiler unit, x is a vector consisting of decision variables, Among them, the output of thermal power units is Powering distributed renewable generation units, is the charging and discharging power of the energy storage unit, The charging power of electric vehicles, is the power consumption of the electrolytic cell, is the electrical output of the combined heat and power unit, is the power consumption of the electric boiler unit, is the hydrogen production power of the electrolyzer, is the amount of hydrogen flowing into / out of the hydrogen storage tank per unit time, is the thermal power of the cogeneration unit, is the output of the electric boiler unit, A and b are the coefficient matrices of the equality constraints, G(x) is the inequality constraint, 0 N is a column vector whose elements are all 0, and N is the number of inequality constraints.

[0083] Furthermore, by constructing an objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant with the goal of minimizing the operating cost of distributed resources, it is used to solve the management and energy optimization scheduling plan of the multi-energy virtual power plant. The obtained plan minimizes the daily operating cost of the multi-energy virtual power plant, which is conducive to maximizing the economic benefits of the multi-energy virtual power plant.

[0084] See also Figure 1 and Figure 2 , an embodiment provided by the present invention: a multi-energy virtual power plant energy distributed optimization management method taking into account network attacks, comprising determining the constraint conditions of the objective function, wherein the constraint conditions include the output power constraint in the decision variable constituting vector, the equipment climbing power constraint, the feasible domain constraint of the cogeneration unit, the energy storage equipment operation constraint, the ideal charging range constraint of the electric vehicle, the power network line flow constraint, the heat network node pipeline temperature and flow constraint, and the hydrogen network node pipeline pressure and flow constraint;

[0085] The output power constraints in the decision variable vector are as follows:

[0086]

[0087]

[0088] in and are the lower and upper limits of thermal power unit output, is the output limit of distributed renewable power generation units, and are the lower and upper limits of the power output of the cogeneration unit, and are the lower and upper limits of the thermal power of the cogeneration unit, and are the lower and upper limits of the power consumption of the electric boiler unit, and are the lower and upper limits of the power consumption of the electrolytic cell, and are the lower and upper limits of the hydrogen production power of the electrolyzer, respectively;

[0089] The feasible region constraints of the cogeneration unit are as follows:

[0090]

[0091] Among them, C m，i is the static heat-to-electricity conversion efficiency coefficient of the cogeneration unit, η i is the efficiency of the combined heat and power unit, c v，i is the dynamic heat-to-electricity conversion efficiency coefficient of the cogeneration unit

[0092] The operating constraints of energy storage equipment are as follows:

[0093]

[0094] in, and are the lower and upper limits of the charging and discharging of the energy storage unit, respectively. and are the lower and upper limits of the capacity of the electric energy storage unit, respectively. and are the lower and upper limits of the hydrogen inflow / outflow of the hydrogen storage tank per unit time, is the upper limit of the capacity of the hydrogen storage tank, θ ESS is the attenuation coefficient of the electric energy storage unit;

[0095] The equipment climbing power constraints are as follows:

[0096]

[0097] in and are the lower and upper limits of the ramp rate of thermal power units, and are the lower and upper limits of the ramp rate of the cogeneration unit, and are the power output of the cogeneration unit at time t-1 and time t, respectively. and are the power outputs of thermal power units at time t-1 and at time t respectively;

[0098] The ideal charging range constraints of electric vehicles are as follows:

[0099]

[0100] in, The ideal charging power for electric vehicles, is the battery capacity of the electric vehicle, and are the lower and upper limits of the battery state required for electric vehicles, is the initial state of the electric vehicle battery, For charging efficiency, The maximum acceptable charging time for electric vehicles;

[0101] The power grid flow route constraints are as follows:

[0102] P line =P -T P in , Q line =P -T Q in ;

[0103]

[0104]

[0105] Where P in and Q in are the active and reactive power injected into the node, P line and Q line are the active and reactive power flows on the line, respectively; B is a submatrix of the node-line graph adjacency matrix; R and X are diagonal matrices whose diagonal elements are composed of line resistance and reactance, respectively. and are the active and reactive power flows on line mn, is the voltage amplitude of bus i;

[0106] The heat network node pipeline temperature and flow balance constraints are as follows:

[0107]

[0108] in, is the heat generation power of the cogeneration unit at node g, is the heating power of the electric boiler at node g, is the heat load at node g, c is the specific heat capacity of water, m g is the mass flow rate of the heat source or heat load, and are the supply and return water temperatures at the heat source, and are the water temperatures injected into or out of node g from pipe i, τ i，g，in and τ i，g，out are the mass flow rate injected into node g and the mass flow rate out of node g, respectively. and are the temperatures at the start and end of pipe p, respectively, l p is the length of the pipe p, T AM is the ambient temperature, δ is the thermal conversion efficiency, τ p and are the lower and upper limits of the pipe temperature respectively;

[0109] The gas pressure and flow balance constraints of hydrogen network node pipelines are as follows:

[0110]

[0111] Among them, h ij，t is the hydrogen flow rate in the pipeline between nodes i and j, φ ij is the hydrogen transmission parameter of the pipeline, π i is the hydrogen pressure at node i, D gp , D ge and D gi are the association matrices of hydrogen pipelines, electrolyzers, hydrogen storage tanks and hydrogen network nodes, respectively. p，t is the amount of hydrogen in the pipeline at time t, is the amount of hydrogen produced by the electrolyzer at time t, is the amount of hydrogen flowing in / out of the hydrogen storage tank at time t.

[0112] Furthermore, by setting multiple sets of constraints for the objective function, the objective function variables of the multi-energy virtual power plant management model are constrained, which helps to reduce the difficulty of solving the problem.

[0113] See also Figure 1 An embodiment of the present invention provides a method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks, including S3, establishing a subject reputation evaluation model based on an observation subnet, the subject reputation evaluation model is used to evaluate the behavior of network nodes in the energy management model of the electric-heat-hydrogen multi-energy virtual power plant, and isolate malicious false nodes according to the evaluation results, S3 specifically includes the following steps:

[0114] S31, establishing an observation subnet based on the communication network foundation, when the network node i transmits information to the network node j, the transmitted information can be sent from the network node i to the two-hop neighbor of the network node j through the observation subnet, and the establishment of the observation subnet enables each network node in the actual communication network to receive the information of its two-hop neighbor through the observation subnet;

[0115] Based on the observation subnet, a subject reputation evaluation model is established. According to the information transmission behavior between network nodes, the subject reputation evaluation model is used to evaluate the behavior of network nodes in the energy management model of the electric-heat-hydrogen multi-energy virtual power plant. The specific steps are as follows:

[0116] The energy mismatch of the local estimate of the network node i is calculated as follows:

[0117]

[0118] Wherein, z(k, v) represents the detection threshold function determined by the number of iterations k and the maximum ramp rate v of the energy device, and the detection threshold function satisfies the following relationship:

[0119]

[0120] Function Y + (a) and Y - (a) are defined as follows:

[0121]

[0122] ρ min ≤1;

[0123] in, is the sum of the weights of the edges connecting network node j and all its outgoing neighbors in the initial communication network graph;

[0124] S32. During the kth iteration, network node i calculates y according to equations (1) and (2). v (k+1) a reasonable boundary value, and compare the reasonable boundary value with the actual value received by the network node i, and evaluate the network node behavior according to the comparison result, and the evaluation method is as follows:

[0125] when When it is established, network node i considers network node v to be of good behavior, otherwise it considers network node v to be of malicious behavior and records network node v as a malicious false node;

[0126] According to the node behavior, update the detection result variable d of the network node v v (k), the detection result variable d v(k) is transmitted from network node i to network node j through the observation subnet, and network node j is an incoming neighbor of network node v, d v (k) satisfies the following relationship:

[0127]

[0128] S33. Based on the network node behavior evaluation results, node reputation evaluation and malicious false nodes isolation are performed, specifically including the following steps:

[0129] In each iteration, network node i counts the counter T according to the detection result of network node v. iv (k) is updated, and the initial value of the counter is T iv (0) = 0, T iv (k) satisfies the following relationship:

[0130]

[0131] A subject reputation evaluation model is established. Network node i uses the subject reputation evaluation model to evaluate the node reputation of network node v and obtain the node reputation evaluation value. The node reputation evaluation value satisfies the following relationship:

[0132]

[0133] Rep iv (k)∈[0,1];

[0134]

[0135] Set a node reputation threshold, compare the node reputation evaluation value with the node reputation threshold, and when the node reputation evaluation value is less than the node reputation threshold, disconnect the communication link between the malicious false node and its neighbor node to isolate the malicious false node;

[0136] The initial communication network diagram is updated based on the isolation results and the communication link disconnection results to obtain the actual communication network diagram, which is used to perform the solution step of S4 in combination with the objective function and constraint conditions in S2.

[0137] Furthermore, by establishing a subject reputation evaluation model for the observation subnet, supervision of each network node in the actual communication network is achieved, and network nodes that have been transformed into malicious false nodes due to network attacks are discovered in a timely manner. The subject reputation evaluation model is used to evaluate and calculate the reputation of the network nodes corresponding to each physical subject, and the node reputation evaluation value is obtained. By comparing it with the node reputation threshold, and according to the comparison result, the communication link between the malicious false node and its neighboring nodes is disconnected, the malicious false node is isolated, and the false node is prevented from transmitting false information to its adjacent network nodes. The possibility of distributed energy subjects corresponding to other network nodes making wrong decisions is reduced, which is conducive to improving the correctness of the solution of the objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant.

[0138] See also Figure 1 , an embodiment provided by the present invention: a multi-energy virtual power plant energy distributed optimization management method taking into account network attacks, in which the iterative solution of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant in S4 includes solving the energy management model of the electric-heat-hydrogen multi-energy virtual power plant based on the actual network communication diagram, the objective function and the constraint conditions, and the solution process includes the following steps:

[0139] Decompose the objective function and constraints into multiple groups of sub-problems;

[0140] Each group of sub-problems is converted into the form of Lagrangian function, and each node body solves the sub-problem converted into Lagrangian function. The sub-problem is solved by a single-layer projection neural network, and the iterative solution process is repeated until the attack behavior of all malicious false nodes is isolated.

[0141] The dynamics equation of a single-layer projection neural network is as follows:

[0142]

[0143] Among them, x represents the neuron state variable, that is, the decision variable in the optimization problem, u, θ, λ, γ are all Lagrange multipliers, σ represents a positive constant used to accelerate the convergence speed, P Ω Represents a projection operation, represents the gradient of f(x), represents the gradient of G(x), L p and L q The Kronecker product of the Laplacian matrix of the graph and the identity matrix is ​​expressed as follows:

[0144]

[0145]

[0146] L is the Laplacian matrix of the graph, I p and I q are both identity matrices, p and q represent the dimensions of the equality constraints and inequality constraints, respectively.

[0147] Furthermore, by converting the objective function into multiple groups of sub-problems with equality and inequality constraints, and converting each sub-problem into the form of Lagrangian function, it is easier to solve the inequality constraints and objective function, and the computational burden is reduced.

[0148] See also Figure 1 An embodiment of the present invention is: a method for distributed optimization management of energy of a multi-energy virtual power plant taking into account network attacks, including information exchange between network node i and network node j, network node i locally backs up the information of network node j obtained through information exchange, and the information exchange adopts an improved Push-sum protocol for privacy protection.

[0149] The network node i and the network node j exchange information, the network node i locally backs up the information of the network node j obtained through the information exchange, and the information exchange adopts the improved Push-sum protocol for privacy protection, including the following steps:

[0150] For the initial variable information x to be exchanged i (0), network node i randomly generates an initial sub-state value from U(-M, M) And initialize the remaining sub-state variables, and initialize the remaining sub-state variables to satisfy the following relationship:

[0151]

[0152] Where M is a positive constant;

[0153] Generate sub-state weights. When k = 0, network node i randomly selects an initial weight set from the distribution Ψ(0, M) The weight normalization process is performed, and the weight normalization process satisfies the following relationship:

[0154]

[0155] Among them When ji (0) = 0;

[0156] When k ≥ 1, network node i generates a random number α that follows uniform distribution U(0, 1) i (K), and generate weights ω from the distribution U(0, 1) ji (k), weight normalization is performed so that and When ji (0) = 0;

[0157] Perform sub-state status updates. For all k ≥ 0, after network node i receives information from neighbor node j, it updates two sub-states. and The updates of the two sub-states satisfy the following relationship:

[0158]

[0159] Network node i calculates the average value based on the updated sub-state, and the calculation expression is as follows:

[0160]

[0161] Furthermore, in the process of information exchange between network nodes, an improved Push-sum protocol is used to protect the privacy of the information exchange process, that is, node i weights the random sub-state of the state variable and sends it to the neighboring node instead of directly exposing the actual data, which effectively ensures data privacy and reduces the possibility of privacy leakage in the information exchange process. A collaborative neural dynamics optimization method combined with the improved Push-sum protocol is used to iteratively solve the energy model of the electric-thermal-hydrogen multi-energy virtual power plant, which improves the computational security in the iterative solution process of the objective function.

[0162] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A multi-energy virtual power plant energy distribution optimization management method taking into account network attacks, characterized in that: The following steps are involved: S1. Based on multi-energy heterogeneous energy, an energy management model of an electric-heat-hydrogen multi-energy virtual power plant is established taking into account the topological structure of heterogeneous energy networks; S2. With the goal of minimizing the operating cost of distributed resources, construct an objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant, and determine the constraints of the objective function; S3. Establishing a subject reputation evaluation model based on the observation subnet, wherein the subject reputation evaluation model is used to evaluate the behavior of network nodes in the electric-heat-hydrogen multi-energy virtual power plant energy management model, and isolate malicious false nodes according to the evaluation results; S4. For non-malicious false nodes, based on the objective function and the constraints, the energy management model of the electric-thermal-hydrogen multi-energy virtual power plant is iteratively solved using a collaborative neural dynamics optimization method combined with an improved Push-sum protocol; S5. Repeat the iterative solution process to obtain the optimization result.

2. According to claim 1, a multi-energy virtual power plant energy distribution optimization management method taking into account network attacks is characterized by: The S3 specifically includes the following steps: S31, establishing an observation subnet based on the communication network foundation, when the network node i transmits information to the network node j, the transmitted information can be sent from the network node i to the two-hop neighbor of the network node j through the observation subnet; Based on the observation subnet, a subject reputation evaluation model is established. According to the information transmission behavior between network nodes, the subject reputation evaluation model is used to evaluate the behavior of network nodes in the energy management model of the electric-heat-hydrogen multi-energy virtual power plant. The specific steps are as follows: The energy mismatch of the local estimate of the network node i is calculated as follows: Wherein, z(k, v) represents the detection threshold function determined by the number of iterations k and the maximum ramp rate v of the energy device, and the detection threshold function satisfies the following relationship: Function Y + (a) and Y - (a) are defined as follows: in, is the sum of the weights of the edges connecting network node j and all its outgoing neighbors in the initial communication network graph; S32. During the kth iteration, network node i calculates y according to equations (1) and (2). v (k+1) and compare the reasonable boundary value with the actual value received by the network node i. The network node behavior is evaluated according to the comparison result, and the evaluation method is as follows: when When it is established, network node i considers network node v to be of good behavior, otherwise it considers network node v to be of malicious behavior and records network node v as a malicious false node; According to the node behavior, update the detection result variable d of the network node v v (k), the detection result variable d v (k) is transmitted from network node i to network node j through the observation subnet, and network node j is an incoming neighbor of network node v, d v (k) satisfies the following relationship:

3. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 2 is characterized in that: The S3 also includes: S33. Based on the network node behavior evaluation results, node reputation evaluation and malicious false nodes isolation are performed, specifically including the following steps: In each iteration, network node i counts the counter T according to the detection result of network node v. iv (k) is updated, and the initial value of the counter is T iv (0) = 0, T iv (k) satisfies the following relationship: A subject reputation evaluation model is established. Network node i uses the subject reputation evaluation model to evaluate the node reputation of network node v and obtain the node reputation evaluation value. The node reputation evaluation value satisfies the following relationship: Rep iv (k)∈[0,1]; Set a node reputation threshold, compare the node reputation evaluation value with the node reputation threshold, and when the node reputation evaluation value is less than the node reputation threshold, disconnect the communication link between the malicious false node and its neighbor node to isolate the malicious false node; The initial communication network diagram is updated based on the isolation results and the communication link disconnection results to obtain the actual communication network diagram, which is used to perform the solution step of S4 in combination with the objective function and constraint conditions in S2.

4. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 3 is characterized in that: In S2, the objective function of the energy management model of the electric-heat-hydrogen multi-energy virtual power plant is constructed with the goal of minimizing the operating cost of distributed resources, and satisfies the following relationship: Ax = b; G(x)≤0 N ; Among them, f total is the overall operating cost of the electricity-heat-hydrogen multi-energy virtual power plant (MEVPP), Ω CG is the collection of thermal power units, is the operating cost function of the thermal power unit, Ω DRG is a collection of distributed renewable power generation units, is the output penalty cost function of distributed renewable generation units, Ω ESS is a collection of electrical energy storage units, is the operating cost function of the electric energy storage unit, Ω EV For the collection of electric vehicles, is the charging utility loss cost function of electric vehicles, Ω EL An electrolyzer assembly for hydrogen production, is the operating cost function of the electrolyzer, Ω CHP is a collection of combined heat and power units, is the operating cost of the combined heat and power unit, Ω EB For the collection of electric boiler units, is the operating cost of the electric boiler unit, x is a vector consisting of decision variables, Among them, the output of thermal power units is Powering distributed renewable generation units, is the charging and discharging power of the energy storage unit, The charging power of electric vehicles, is the power consumption of the electrolytic cell, is the electrical output of the combined heat and power unit, is the power consumption of the electric boiler unit, is the hydrogen production power of the electrolyzer, is the amount of hydrogen flowing into / out of the hydrogen storage tank per unit time, is the thermal power of the cogeneration unit, is the output of the electric boiler unit, A and b are the coefficient matrices of the equality constraints, G(x) is the inequality constraint, 0 N is a column vector whose elements are all 0, and N is the number of inequality constraints.

5. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 4 is characterized in that: The S2 also includes the constraints for determining the objective function, and the constraints include the output power constraint in the decision variable vector, the equipment climbing power constraint, the feasible domain constraint of the cogeneration unit, the energy storage equipment operation constraint, the ideal charging range constraint of the electric vehicle, the power network line flow constraint, the heat network node pipeline temperature and flow constraint, and the hydrogen network node pipeline pressure and flow constraint.

6. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 5 is characterized in that: The output power constraint in the decision variable vector in the S2 constraint condition is as follows: in and are the lower and upper limits of thermal power unit output, is the output limit of distributed renewable power generation units, and are the lower and upper limits of the power output of the cogeneration unit, and are the lower and upper limits of the thermal power of the cogeneration unit, and are the lower and upper limits of the power consumption of the electric boiler unit, and are the lower and upper limits of the power consumption of the electrolytic cell, and are the lower and upper limits of the hydrogen production power of the electrolyzer, respectively; The feasible domain constraints of the cogeneration unit are as follows: Among them, C m，i is the static heat-to-electricity conversion efficiency coefficient of the cogeneration unit, η i is the working efficiency of the cogeneration unit, c v，i is the dynamic heat-to-electricity conversion efficiency coefficient of the cogeneration unit The energy storage equipment operation constraints are as follows: in, and are the lower and upper limits of the charging and discharging of the energy storage unit, respectively. and are the lower and upper limits of the capacity of the electric energy storage unit, respectively. and are the lower and upper limits of the hydrogen inflow / outflow of the hydrogen storage tank per unit time, is the upper limit of the capacity of the hydrogen storage tank, θ ESS is the attenuation coefficient of the electric energy storage unit.

7. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 6 is characterized in that: The equipment climbing power constraint in the S2 constraint condition is as follows: in and are the lower and upper limits of the ramp rate of thermal power units, and are the lower and upper limits of the ramp rate of the cogeneration unit, and are the power output of the cogeneration unit at time t-1 and time t, respectively. and are the power outputs of thermal power units at time t-1 and time t respectively; The ideal charging range constraints of electric vehicles are as follows: in, The ideal charging power for electric vehicles, is the battery capacity of the electric vehicle, and are the lower and upper limits of the battery state required for electric vehicles, is the initial state of the electric vehicle battery, For charging efficiency, The maximum acceptable charging time for electric vehicles.

8. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 7 is characterized in that: The power grid flow route constraints in the S2 constraint condition are as follows: P line =P -T P in ,Q line =P -T Q in ; Where P in and Q in are the active and reactive power injected into the node, P line and Q line are the active and reactive power flows on the line, respectively; B is a submatrix of the node-line graph adjacency matrix; R and X are diagonal matrices whose diagonal elements are composed of line resistance and reactance, respectively. and are the active and reactive power flows on line mn, is the voltage amplitude of bus i; The heat network node pipeline temperature and flow balance constraints are as follows: in, is the heat generation power of the cogeneration unit at node g, is the heating power of the electric boiler at node g, is the heat load at node g, c is the specific heat capacity of water, m g is the mass flow rate of the heat source or heat load, and are the supply and return water temperatures at the heat source, and are the water temperatures injected into or out of node g from pipe i, τ i，g，in and τ i，g，out are the mass flow rate injected into node g and the mass flow rate out of node g, respectively. and are the temperatures at the start and end of pipe p, respectively, l p is the length of the pipe p, T AM is the ambient temperature, δ is the thermal conversion efficiency, τ p and are the lower and upper limits of the pipe temperature respectively; The hydrogen network node pipeline gas pressure and flow balance constraints are as follows: Among them, h ij，t is the hydrogen flow rate in the pipeline between nodes i and j, φ ij is the hydrogen transmission parameter of the pipeline, π i is the hydrogen pressure at node i, D gp , D ge and D gi are the association matrices of hydrogen pipelines, electrolyzers, hydrogen storage tanks and hydrogen network nodes, respectively. p，t is the amount of hydrogen in the pipeline at time t, is the amount of hydrogen produced by the electrolyzer at time t, is the amount of hydrogen flowing in / out of the hydrogen storage tank at time t.

9. The method for distributed optimization management of multi-energy virtual power plants taking into account network attacks according to claim 8 is characterized in that: The iterative solution of the electric-heat-hydrogen multi-energy virtual power plant energy management model in S4 includes solving the electric-heat-hydrogen multi-energy virtual power plant energy management model based on the objective function and constraint conditions according to the actual network communication diagram, and the solution process includes the following steps: Decompose the objective function and constraints into multiple groups of sub-problems; Each group of sub-problems is converted into the form of Lagrangian function, and each node body solves the sub-problem converted into Lagrangian function, and the sub-problem is solved by a single-layer projection neural network; Network node i and network node j exchange information, and network node i locally backs up the information of network node j obtained through the information exchange. The information exchange adopts an improved Push-sum protocol for privacy protection.

10. The method for distributed optimization management of energy in a multi-energy virtual power plant taking into account network attacks according to claim 9, characterized in that: The network node i and the network node j exchange information, the network node i locally backs up the information of the network node j obtained through the information exchange, and the information exchange adopts the improved Push-sum protocol for privacy protection, including the following steps: For the initial variable information x to be exchanged i (0), network node i randomly generates an initial sub-state value from U(-M, M) And initialize the remaining sub-state variables, and initialize the remaining sub-state variables to satisfy the following relationship: Where M is a positive constant; Generate sub-state weights. When k = 0, network node i randomly selects an initial weight set from the distribution Ψ(0, M) The weight normalization process is performed, and the weight normalization process satisfies the following relationship: Among them When ji (0) = 0; When k ≥ 1, network node i generates a random number α that follows uniform distribution U(0, 1) i (K), and generate weights ω from the distribution U(0, 1) ji (k), weight normalization is performed so that and When ji (0) = 0; Perform sub-state status updates. For all k ≥ 0, after network node i receives information from neighbor node j, it updates two sub-states. and The updates of the two sub-states satisfy the following relationship: Network node i calculates the average value based on the updated sub-state, and the calculation expression is as follows:

Images