Comprehensive energy system distributed optimization scheduling method and system considering privacy protection

Abstract

The invention discloses an integrated energy system distributed optimization scheduling method and system considering privacy protection, and relates to the technical field of integrated energy systems.The method comprises the steps that a communication network topological graph of an integrated energy system is constructed with all devices in the system as nodes and energy interaction between the nodes as edges; with the minimum total cost as an objective function, configuring constraint conditions of system operation, and constructing an optimal scheduling model of the integrated energy system; assuming that the communication network topological graph is an undirected connected graph, adopting a preset time distributed optimization algorithm introducing time-varying gain to solve the optimization scheduling model, and enabling the integrated energy system to converge to an optimal solution from any initial state within preset time to obtain an optimal optimization strategy; wherein in the solving process, through an encryption and decryption module based on a deep learning adversarial network, encryption and decryption are performed on interaction variables between neighbor nodes, and privacy protection is realized. According to the method, the rapid convergence and privacy protection requirements of scheduling optimization solution of the integrated energy system can be met.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of integrated energy systems, and particularly relates to a privacy-considered distributed optimization scheduling method and system for integrated energy systems. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] An integrated energy system is a typical multi-agent information-physical system, which internally couples multiple heterogeneous energies such as electricity, heat, gas, and cold to achieve multiple energy complementation and mutual aid and meet user energy demand. At the physical layer, the system contains multiple energy forms and supplies energy to the load through different energy conversion devices; at the information layer, each node in the system interacts with information to ensure the coordinated and efficient operation of the system. In the context of large energy consumption and high consumption of fossil energy, it is of great strategic significance to establish an integrated energy system with low carbon, environmental protection, and high energy utilization rate.

[0004] Currently, the optimization scheduling problem of an integrated energy system can be solved by centralized optimization and distributed optimization. The centralized optimization method requires global information and has large communication data volume and poor communication network robustness. Although the existing distributed optimization method can solve these problems to some extent, as the system scale expands, the optimization scheduling of the system requires higher convergence speed of the optimization algorithm and privacy protection, and the existing distributed optimization method cannot effectively cope with it. For the low-carbon economic optimization scheduling problem of a large-scale integrated energy system with higher convergence speed and privacy protection requirements, although the preset time algorithm for solving the high convergence speed problem has been proposed, the existing method does not consider the privacy protection problem in the solving process, which cannot complete the encryption task within the minimum time interval required by the algorithm, and the accuracy and efficiency of the existing encryption method are low, which cannot be directly applied to the algorithm solving process. SUMMARY

[0005] To solve the above problems of the prior art, the present application provides a privacy-considered distributed optimization scheduling method and system for integrated energy systems, which uses a preset time distributed optimization algorithm independent of the initial state of the system to solve the optimization scheduling strategy of the integrated energy system, and uses Lyapunov's second method to prove that the algorithm can guarantee the preset time convergence of the system from any initial state, and designs an encryption and decryption module to be applied to the solving process to improve the privacy of the interactive information between nodes in the optimization solving process and meet the privacy protection requirements of the scheduling optimization solution.

[0006] In a first aspect, the present application provides a privacy-considered distributed optimization scheduling method for integrated energy systems.

[0007] A privacy-considered integrated energy system distributed optimization scheduling method, comprising: A communication network topology graph of the integrated energy system is constructed by taking each device in the system as a node and energy interaction between nodes as an edge; An optimization scheduling model of the integrated energy system is constructed by taking the minimum total cost as an objective function and configuring constraint conditions of system operation; The communication network topology graph is assumed to be an undirected connected graph, a preset time distributed optimization algorithm with time-varying gain is adopted to solve the optimization scheduling model, so that the integrated energy system converges to an optimal solution from an arbitrary initial state within a preset time, and an optimal optimization strategy is obtained; wherein, in the solving process, the interaction variables between neighbor nodes are encrypted and decrypted through the encryption and decryption module based on the deep learning adversarial network, and privacy protection is realized.

[0008] In a further technical solution, the encryption and decryption module based on the deep learning adversarial network is implemented based on a multi-agent game framework, and includes an encrypter Alice, a decrypter Bob and an attacker Eve constructed based on a neural network; The encrypter Alice and the decrypter Bob form a cooperative alliance, and a joint key is deployed in the neighbor nodes; wherein, the interaction variables between the neighbor nodes are taken as plaintext vectors, the encrypter Alice learns to generate ciphertext vectors by taking the plaintext vectors and the key as inputs, and the decrypter Bob learns to reconstruct the original plaintext vectors by taking the ciphertext vectors and the same key as inputs as an authorized receiver; The attacker Eve forms an adversarial game with the cooperative alliance, and guides the encrypter Alice and the decrypter Bob to learn complex nonlinear encryption transformation through adversarial training.

[0009] In a further technical solution, the encrypter Alice, the decrypter Bob and the attacker Eve all adopt a multi-layer perceptron with deep residual as a basic framework to build, and the multi-layer perceptron with deep residual includes an input linear layer, a core network stacked by a plurality of residual blocks and an output linear layer; Wherein, each residual block is defined as H(X) = X + F(X), F(X) is a residual mapping learned by the network layer, and X is the original input transmitted through a shortcut connection; the network layer includes linear layers, batch normalization layers and activation layers connected in turn.

[0010] In a further technical solution, the encryption and decryption module adopts a two-stage update strategy for training in each training batch, including: In phase 1, the network parameters of Alice the encryptor and Bob the decryptor are frozen. The ciphertext vector generated by Alice the encryptor is passed to Eve the attacker, and Alice the encryptor weight is not updated. Based on minimizing Eve's reconstruction loss, backpropagation is performed to update Eve's network weight. In the second phase, the attacker Eve's network parameters are frozen, and the encryptor Alice and the decryptor Bob perform the complete encryption and decryption process, and perform backpropagation based on the minimization of the composite loss function, while updating the network weights of the encryptor Alice and the decryptor Bob.

[0011] In a further technical solution, the encryption and decryption modules are end-to-end optimized using an asymmetric loss function. The composite loss function of the cooperative alliance formed by the encryptor Alice and the decryptor Bob is: ; The L1 reconstruction loss function of the decryptor Bob is for: ; The L1 reconstruction loss function of the attacker Eve for: ; Among them, U is the original plaintext vector, U' is the decryption result of the decryptor Bob, and U'' is the cracking result of the attacker Eve. is a hyperparameter.

[0012] A further technical solution is to use a preset time distributed optimization algorithm with time-varying gain to solve the optimization scheduling model, which is: ; in, , is the gradient, Indicates the i The cost function of each node, Indicates the i The power of each node; is the projection operator, is a closed convex set, For about Inequality constraint function of ; definition is the Laplacian operator of the graph, represents the adjacency matrix, Indicates the adjacency matrix i Rank j elements, nodes i The neighbor node set of N i}, 、 、 is an auxiliary variable; is an introduced time-varying gain function, is a constant gain.

[0013] Further technical solutions, the comprehensive energy system is a heat and electricity coupled energy system; The target function is to minimize the sum of the operation costs of all power supply devices, heat supply devices, heat and electricity coupled devices and energy storage devices; The constraint conditions include heat and electricity ratio constraints, energy supply and demand balance constraints and safe operation constraints.

[0014] In a second aspect, the present application provides a comprehensive energy system distributed optimization scheduling system considering privacy protection.

[0015] A comprehensive energy system distributed optimization scheduling system considering privacy protection comprises: A system topology graph construction module is configured to construct a communication network topology graph of the comprehensive energy system by taking each device in the system as a node and energy interaction between nodes as an edge; A scheduling model construction module is configured to construct a comprehensive energy system optimization scheduling model by taking the minimum total cost as a target function and configuring constraint conditions for system operation; An optimal scheduling solution module is configured to assume that the communication network topology graph is a undirected connected graph, adopt a preset time distributed optimization algorithm with an introduced time-varying gain, solve the optimization scheduling model, make the comprehensive energy system converge to an optimal solution from an arbitrary initial state within a preset time, and obtain an optimal optimization strategy; wherein, in the solving process, an encryption and decryption module based on a deep learning adversarial network is used to encrypt and decrypt interaction variables between neighbor nodes, thereby realizing privacy protection.

[0016] In a third aspect, the present application further provides an electronic device comprising a memory for storing executable instructions and a processor for executing the executable instructions stored in the memory to implement the comprehensive energy system distributed optimization scheduling method considering privacy protection.

[0017] In a fourth aspect, the present application further provides a computer readable storage medium storing executable instructions for causing a processor to execute the executable instructions to implement the comprehensive energy system distributed optimization scheduling method considering privacy protection.

[0018] The above one or more technical solutions have the following beneficial effects: The application provides a privacy-considered integrated energy system distributed optimization scheduling method and system, aiming at the typical constraint convex optimization problem of integrated energy system economic scheduling, a preset time distributed optimization algorithm is designed by introducing time-varying gain, the preset time distributed optimization algorithm which does not depend on the initial state of the system is used to solve the integrated energy system optimization scheduling strategy, and the second Lyapunov method is used to prove that the algorithm can guarantee the preset time convergence of the system from any initial state; at the same time, the encryption and decryption module based on the deep learning adversarial network is designed and applied to the solving process, the interactive information in the process is encrypted, so as to improve the privacy of the interactive information between the nodes in the optimization solving process, and meet the privacy protection requirements of scheduling optimization solving.

[0019] The advantages of the additional aspects of the application will be partially given in the following description, partially become obvious from the following description, or be known by the practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0020] The drawings accompanying the specification of this application form a part thereof, serve to provide further understanding of the application, and together with the description of the exemplary embodiments of the application and their description serve to explain the application, and do not constitute an improper limitation of the application.

[0021] Figure 1 The logic block diagram of the encryption and decryption module proposed in the embodiment of the application; Figure 2 The network structure diagram of the deep residual multi-layer perception introduced in the embodiment of the application; Figure 3 The structure and communication network topology diagram of the thermoelectric coupled integrated energy system in the embodiment of the application; Figure 4 The distribution diagram of the decryption result of the data by the decryptor Bob network in the embodiment of the application; wherein, (a) is the distribution diagram of the decryption result of data 1 by the Bob network, and (b) is the distribution diagram of the decryption result of data 2 by the Bob network; Figure 5 The distribution diagram of the decryption result of the data by the attacker Eve network in the embodiment of the application; wherein, (a) is the distribution diagram of the decryption result of data 1 by the Eve network, and (b) is the distribution diagram of the decryption result of data 2 by the Eve network; Figure 6 The decryption error schematic diagram of the decryptor Bob and the attacker Eve network in the embodiment of the application; Figure 7 The comparison diagram of the decryption error results under different network structures in the embodiment of the application; Figure 8 The power convergence curve schematic diagram in the embodiment of the application; wherein, (a) is a thermal power convergence curve, and (b) is an electric power convergence curve; Figure 9 Fig. 12 is a schematic diagram of a power balance curve in an embodiment of the present application; wherein (a) is a thermal power balance curve, and (b) is an electrical power balance curve; Figure 10 Fig. 13 is a schematic diagram of a thermal energy 0-125s power convergence curve in an embodiment of the present application; Figure 11 Fig. 14 is a comparison diagram of power convergence curves with or without topology transformation in an embodiment of the present application; wherein (a) is a comparison diagram of power convergence curves without topology transformation, and (b) is a comparison diagram of power convergence curves with topology transformation; Figure 12 Fig. 15 is a comparison diagram of original information and encrypted information of interaction variables in an embodiment of the present application; wherein (a) is a comparison diagram of original information and encrypted information of interaction variable y, and (b) is a comparison diagram of original information and encrypted information of interaction variable z; Figure 13 Fig. 16 is a comparison diagram of original information and decrypted information of decryptor Bob in an embodiment of the present application. DETAILED DESCRIPTION

[0022] It should be noted that the following detailed description is exemplary only and is intended to provide further description of the present application in order to describe specific embodiments and is not intended to limit the exemplary embodiments according to the present application. Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application pertains. In addition, it should be understood that the terms "comprise" and / or "include" as used in this specification mean that there are features, steps, operations, devices, components, and / or combinations thereof.

[0023] Embodiment One The embodiment provides a comprehensive energy system distributed optimization scheduling method considering privacy protection, and specifically comprises the following steps: Step S1, constructing a communication network topology graph of the comprehensive energy system by taking each device in the system as a node and taking energy interaction between nodes as an edge; configuring constraint conditions of system operation as an objective function to minimize total cost, and constructing an optimization scheduling model of the comprehensive energy system.

[0024] Step S2, assuming that the communication network topology graph is a connected graph, using a preset time distributed optimization algorithm introducing time-varying gain to solve the optimization scheduling model, so that the comprehensive energy system converges to an optimal solution from an arbitrary initial state within a preset time, and an optimal optimization strategy is obtained; wherein, in the solving process, the interaction variables between neighbor nodes are encrypted and decrypted by the encryption and decryption module based on the deep learning adversarial network, so as to realize privacy protection.

[0025] The privacy-protected integrated energy system distributed optimization scheduling method proposed in this embodiment is described in more details by the following.

[0026] Firstly, in step S1, a communication network topology graph of an integrated energy system (IES) is constructed, and an integrated energy system optimization scheduling model is established. Specifically, the integrated energy system is a typical multi-agent cyber-physical system, and this embodiment takes a typical heat and electricity coupled integrated energy system as an example for illustration. In the physical layer, the system contains multiple energy forms, and the loads are supplied with energy through different energy conversion devices, so that the system contains a supply side and a demand side. The supply side can be divided into: a single power supply device, a single heat supply device, a heat and electricity coupled device, and an energy storage device. The demand side includes: a heat load and an electrical load. In the information layer, each node in the system exchanges information to ensure the coordinated and efficient operation of the system, so that the communication network of the integrated energy system can be represented by an undirected connected graph, in which each device in the system is taken as a node, the energy exchange between nodes is taken as an edge, and the nodes are represented by A (i), the edges are represented by E (i, j), the adjacency matrix is represented by A, the element in the i-th row and the j-th column of the adjacency matrix is represented by A (i, j), the neighbor node set of the node is represented by N (i), and the neighbor node set of the node is represented by N (i). i j i N i n m l q r s

[0027] Secondly, by solving the economic scheduling problem of the IES, the optimal economic operation of the system can be realized under the premise of safe and stable operation. The optimization objective of the IES is to minimize the sum of the operation costs of all power supply devices, heat supply devices, heat and electricity coupled devices, and energy storage devices, which can be represented as: (1) (2) (3) (4) (5) (6) ​​​​​​​​​​​​​​​​​​ (7) in, C is the overall operating cost of the integrated energy system, Represents the cost function of each subject, each subject PH, P, H, S, DR They represent thermoelectric coupling equipment, separate heating equipment, separate power supply equipment, energy storage equipment, and demand response equipment respectively; , Respectively represent i The electrical power and thermal power of each subject, Indicates the i The cost coefficient of each entity (i.e., the quadratic coefficient).

[0028] In addition, configure the constraints for system operation. For thermoelectric coupling devices, the thermoelectric ratio constraint should be met, which is: , and introduce penalty terms to improve the model, so C PH It can be expressed as: (8) in, represents the thermoelectric ratio, Represents the penalty coefficient.

[0029] In the economic dispatch problem, the energy supply and demand balance constraint and the safe operation constraint should be met, which can be expressed as: (9) (10) in, Indicates the j The load demand of each entity.

[0030] (11) (12) in, Indicates the minimum and maximum values ​​of the output electric power. Indicates the minimum and maximum values ​​of the output thermal power. Indicates the i The output power function and the climbing rate of the output thermal power of each subject, Represents the electrical power and thermal power output at time t.

[0031] To facilitate the description of subsequent questions, the IES economic dispatch model can be expressed as follows: (13) in, , , is a closed convex set, For the inequality constraint function of .

[0032] Secondly, in step S2, the embodiment proposes a preset time distribution optimization algorithm independent of the initial state of the system, which is used to solve the optimization scheduling model. By introducing a specific time-varying gain in the system, it can be ensured that the economic dispatching problem of the comprehensive energy system with constraints converges to the optimal solution from any initial state within a preset time, and the optimal optimization strategy is obtained. Considering the security of information interaction between neighbor nodes in the solving process, the embodiment also proposes an encryption and decryption module based on a deep learning adversarial network design. Based on the adversarial game theory, a customized three-body adversarial framework is designed, in which the core cryptography mapping is generated by a deep residual multi-layer perceptron (ResMLP) self-learning. The entire module is driven and optimized end-to-end through an asymmetric loss function designed to prioritize decryption fidelity, ensuring that neighbor nodes with key authorization can achieve nearly lossless reconstruction accuracy, while ensuring that encrypted data cannot be cracked by external nodes without the key in terms of calculation.

[0033] Specifically, in step S2, first, the communication network topology graph is a connected undirected graph, and a preset time distribution optimization algorithm with time-varying gain is used to solve the optimization scheduling model, so that the comprehensive energy system converges to the optimal solution from any initial state within a preset time, and the optimal optimization strategy is obtained.

[0034] For the convenience of subsequent description and proof, the following assumptions are given: Assumption 1: For each agent i , its cost function C i is strongly convex, and there exists a constant such that .

[0035] Assumption 2: The communication topology graph G is a connected undirected graph.

[0036] Assumption 3: Define the function must be continuously differentiable on the interval [0, T c ), and satisfy , .

[0037] To handle the inequality constraints in the model, the following lemma is given: Lemma 1: Let be a closed convex set, and take the variable x, yFor the projection dynamic system , whose projection operator satisfies the following variational inequality, for any point in the feasible region , .

[0038] The above problem (13) is a typical distributed optimization problem with constraints. To solve this problem, the embodiment proposes a preset time distributed optimization algorithm, which can be represented as: (14) wherein, , is the gradient, denotes the cost function of the i th node (i.e., the principal part), denotes the power of the i th node; is the projection operator, is a closed convex set, is an inequality constraint function about ; define as the Laplacian operator of the graph, denotes the adjacency matrix, denotes the i th element in the j th row of the adjacency matrix, and the neighbor node set of the node i is denoted as N i}, , , is an auxiliary variable; is a time-varying gain function introduced, is a constant gain.

[0039] By introducing the projection operator to deal with the inequality constraint, a more accurate solution can be achieved, that is, compared with other algorithms that obtain an approximate solution, the above method can theoretically obtain an accurate solution, thereby ensuring that the final scheduling effect is better.

[0040] Theorem 1: Under the conditions of assumptions 1 to 3, the algorithm (14) can guarantee that the system converges to the optimal equilibrium point T c from any initial state within the preset time .

[0041] Proof: Define the error variable , and consider the following Lyapunov candidate function: (15) Since , the derivative of (15) along the system trajectory is obtained: (16) For the term containing , we use to represent for the convenience of subsequent description. According to Lemma 1, let , and the optimal point , thus we can get: (17) For the term containing , we can get: (18) Substituting the error variable and the KKT condition , we can get: (19) The KKT condition (Karush-Kuhn-Tucker condition) is a set of necessary conditions in optimization theory for solving nonlinear programming problems with equality and inequality constraints.

[0042] For the term containing , substituting the KKT condition , we can get: (20) For the term containing , we can get: (21) Substituting (19)-(21) into (16) and extracting the common factor , we can get: (22) According to the commutative law of vector inner product and the symmetry of Laplacian matrix (Assumption 2), all cross terms completely cancel out, and equation (22) can be simplified to: (23) Inequality (23) shows that , which means that the system state is bounded. Let S be the largest invariant set of , and within S , we have and . According to graph theory, means that for some constant vector c , there is . Substituting into the dynamics equation of , it shows that must be equal to ,Right now ,therefore . Subsequently, according to and The dynamic equation (in Under the conditions of must also be zero. Therefore, S The only point inside is According to the LaSalle invariance principle, the system is asymptotically stable.

[0043] For the algorithm proposed in this embodiment, the Lyapunov function is quadratic, and its derivative is bounded by a negative semidefinite quadratic form; according to Lyapunov's second method, there exists a positive constant , making Satisfies the following differential inequality: (twenty four) in, Is a system parameter Related constants.

[0044] Introducing time-varying gain , we can get: (25) make , solving inequality (25) yields: (26) According to hypothesis 3, when hour, Divergence, which ensures that when hour, ,therefore This shows that the algorithm can ensure that the system Converges to the optimal value, and Theorem 1 is proved.

[0045] Secondly, considering that during the calculation process of algorithm (14), neighbor nodes need information about the interaction variables y and z, and that information is easily intercepted during the interaction process, resulting in poor privacy, the privacy of the algorithm can be greatly improved by encrypting the interaction information. To this end, this embodiment designs a deep learning encryption method based on the multi-agent game theory, which learns to master a secure, key-dependent data confidentiality transformation. That is, during the solution process, the interaction variables between neighbor nodes are encrypted and decrypted through an encryption and decryption module based on a deep learning adversarial network, thereby achieving privacy protection.

[0046] Specifically, inspired by the generative adversarial network (GAN) and game theory, the embodiment constructs an encryption and decryption module based on a deep learning adversarial network, which is implemented based on a multi-agent game framework composed of three functionally explicit neural networks: encryptor (Alice), decryptor (Bob), and attacker (Eve). In this game framework, Alice and Bob form a cooperative alliance and are jointly deployed in neighbor nodes, and jointly pursue the accuracy of information restoration, while the alliance and Eve form an adversarial game, and through competitive pressure (i.e., adversarial training), the network learns a secure encryption strategy, and the logical relationship is as shown in Figure 1 .

[0047] The encryptor Alice, the decryptor Bob, and the attacker Eve are all based on neural networks, and the functions of each neural network are as follows: Alice (encryptor): taking the interaction variables between neighbor nodes as the plaintext vector, and taking the plaintext vector U and the key K as inputs, learning to generate the ciphertext vector W . Its mapping definition is: W = Alice( U , K ).

[0048] Bob (decryptor): as an authorized receiver, taking the ciphertext C and the same key K as inputs, learning to accurately reconstruct the original plaintext. Its mapping definition is: U' = Bob( W , K ).

[0049] Eve (attacker): representing an interceptor C but without the key K , whose goal is to reconstruct the plaintext from C alone. Its mapping definition is: U'' = Eve( W ).

[0050] Considering that the data to be encrypted is essentially a low-dimensional vector lacking spatial structure, the embodiment adopts a multi-layer perceptron (MLP) as the basic architecture, and on this basis, to overcome the performance degradation and gradient disappearance problems that may occur after stacking multiple layers in a traditional deep MLP, a residual learning mechanism is also innovatively introduced, and the final network architecture is a deep residual MLP (ResMLP) composed of an input linear layer, a core network stacked by multiple residual blocks, and an output linear layer, and its structure is as follows:Figure 2 As shown, each residual block is mathematically defined as H ( X )= X + F ( X ), F ( X ) is the residual mapping learned by the network layer, and X It is the original input passed through the skip-connection. Inside each residual block, the network layer includes a linear layer, a batch normalization layer, and an activation layer connected in sequence. The input matrix X undergoes a linear transformation in the linear layer. (The linear transformation formula is ) Output matrix , and then output the final result through the normalization layer and activation layer. The application of batch normalization technology can stabilize the data distribution of each layer input, alleviate the internal covariate shift, and thus accelerate the convergence of the model. Through the above-mentioned deep residual design, the network is given a powerful ability to learn highly complex and nonlinear cryptographic transformations. Among them, is the weight matrix, is the bias matrix; for The matrix form of Representation matrix The mean and standard deviation of is a smaller constant, usually 0.00005; Indicates the parameters to be optimized.

[0051] Furthermore, to achieve the dual goals of high precision and high security, this embodiment adopts a systematic training optimization strategy, including dynamic data generation and cosine annealing learning rate scheduling, as follows: (1) To ensure that the model learns a general function mapping rather than overfitting to a specific dataset, a dynamic data generation strategy is adopted. In each training cycle, the system generates new batches of random data covering the entire numerical range in real time, providing the model with a theoretically infinite and diverse training set. This strategy not only fundamentally avoids the overfitting problem caused by a limited dataset, but also ensures that the cryptographic transformation ultimately learned has strong generalization and robustness, and its security does not rely on any specific statistical characteristics of the plaintext data.

[0052] (2) A cosine annealing scheduler is used to dynamically adjust the learning rate. Starting with a high learning rate promotes rapid initial convergence, the learning rate then decays smoothly along a cosine curve. This allows the model to explore the minimum point of the loss function with finer step sizes in the later stages of training, which is crucial for improving the final decryption accuracy.

[0053] In a multi-agent game framework, by introducing the attacker Eve, the system is forced to autonomously explore and learn complex nonlinear transformations to hide any exploitable patterns in the data. This process is similar to an automated, security-oriented algorithm discovery process. This approach significantly improves encryption security and reduces the risk of data cracking. Furthermore, continuous adversarial training between the Alice / Bob combination and Eve ensures that the resulting encryption method is robust against its specific adversary, thereby creating a security boundary that continuously evolves and strengthens during training.

[0054] Considering that the encryption system must not only ensure that the decryptor can accurately restore the ciphertext when the key is available, but also ensure that the attacker cannot accurately restore the ciphertext without the key, this embodiment adopts different optimization objectives for the Alice / Bob alliance and Eve. By introducing the hyperparameter λ to decouple the asymmetric loss function, adjusting λ can accurately control the trade-off between the system's assurance and security.

[0055] Specifically, the goal of the alliance between Alice and Bob is to minimize the following composite loss function ,Right now: (27) in, and are the L1 reconstruction losses of Bob and Eve respectively, U is the original plaintext, U' is the decryption result of Bob, and U'' is the cracking result of Eve. The hyperparameter λ is used to precisely trade-off the decryption accuracy and encryption security.

[0056] Meanwhile, the attacker Eve is independently trained to minimize its own reconstruction loss ,Right now: (28) The above asymmetric design ensures the implementation of the "accuracy first" principle and effectively guides the evolution direction of the three-body game.

[0057] In each training batch, a precise two-stage update strategy is followed to ensure effective cooperation and adversarial performance, including: Phase 1 is the update attacker (Eve-Step), which aims to maximize Eve's cracking ability. First, the network parameters of Alice and Bob are frozen. Alice generates ciphertext and passes it to Eve, but the gradient flow of the computation graph is cut off to ensure that Alice's weights are not updated. Subsequently, the entire module is based only on Eve's reconstruction loss. Perform backpropagation and update only Eve's network weights.

[0058] Stage 2 is to update the communication parties (Alice & Bob-Step), which aims to improve the accuracy and security of communication. At this time, the network parameters of Eve are frozen, and Alice and Bob perform the complete encryption and decryption process. The entire module is based on a composite loss function Backpropagation is performed, and the network weights of Alice and Bob are simultaneously updated.

[0059] By alternately performing the above two steps in each batch, the encryption and decryption module is forced to learn to generate more difficult-to-crack ciphertexts under the increasing pressure of Eve, while maintaining the decryption ability on the Bob side, ultimately achieving a dynamic balance with high fidelity and high security.

[0060] The effectiveness and advancement of the method of the embodiment are verified through the following simulation examples. The simulation system is an improved thermal-electric coupling comprehensive energy system of IEEE30 power bus-14 thermal bus, and its structure and communication topology are as shown in Figure 3 Fig. 1. Among them, G1, G2, G3, and G4 are generator sets, L1, L2, L3, L4, L5, L6, L7, L8, L9, L10, L11, L12, L13, and L14 are electric load devices, S1 and S2 are electric energy storage devices, CHP1 and CHP2 are CHP units (i.e. combined heat and power units), P1 and P2 are heat pump devices, B1 and B2 are boiler devices, H1, H2, H3, H4, H5, and H6 are thermal load devices, PV represents photovoltaic devices, WT represents wind power devices, EnergyHub (abbreviated as EH) represents an energy hub, and the numerical designations 1-30 represent 30 system nodes. In addition, its simulation platform is a PC host: Intel(R) Xeon(R) Platinum 8358P@ 2.60GHz with 32 cores, 64GB RAM, NVIDIA A40 GPU with 48GB RAM; the simulation software is MATLAB R2023b and Python 3.8.

[0061] (1) Parameter setting Figure 3 The simulation system shown in the figure contains two energy forms of heat and electricity, and the scheduling time scales of the two energies are different, so it is necessary to decouple the problem and preset different convergence times. The CHP (Combined heat and power) is decoupled by heat conversion, and the EH (Energy Hub) contains multiple energy conversion devices, so its output state can be directly decoupled. The simulation system parameters are shown in Tables 1 and 2.

[0062] Table 1: Thermoelectric coupling unit parameters

[0063] Table 2: Electrical and thermal energy device parameters

[0064] In addition, the distributed algorithm related parameters are as follows, and the initial value of the auxiliary variable is: , , , thermal energy load , electrical energy load , adjustment coefficient , the time-varying gain function is: (29) Note 1: Although formula (29) does not strictly meet assumption 3, in order to avoid numerical singularity near t = T c , the gain function k(t) is processed, by selecting a small enough σ (in this embodiment, σ = 0.0001), the residual error of the system at t = T c can be ignored, and the numerical overflow phenomenon is avoided.

[0065] The parameter settings of the encryption system are as follows: the number of deep residual layers is 8, the hyperparameter , the key bit number is 16, and the initial value of the learning rate scheduler is 0.001.

[0066] (2) Effect verification of encryption and decryption modules The three deep learning networks of Alice, Bob and Eve are trained by using the method and parameters of the embodiment, and the three networks are tested.

[0067] First, 1000 groups of two-dimensional data are randomly generated, decrypted through the Alice network, and then decrypted through the Bob and Eve networks respectively, and the test effect is as Figure 4 Figure 5 shown in the figure. As can be seen from the figure, the neural encryption system proposed in the embodiment shows extremely high fidelity on the authorized decryption channel. Figure 4 The scatter plot distribution of the Bob decryption result is shown in the figure, which is closely and linearly distributed around y = x the standard line (true value), showing a strong positive correlation, and the two are almost indistinguishable visually. This shows that, with the correct key, the Bob network can restore the encrypted information almost losslessly, successfully achieving the primary goal of system design. Correspondingly, Figure 5 ​The scatter plot representing the Eve breaking result shows a disordered, diffuse cloud-like distribution, and there is no recognizable linear or nonlinear structure. This intuitively shows that the predicted value of Eve and the real plaintext are mutually independent, and the breaking performance is equivalent to random guessing. The security of the encryption system is proved, and information leakage is effectively prevented.

[0068] Secondly, in order to more intuitively analyze the decryption accuracy of the proposed encryption method under the condition of having a key, the embodiment uses the MAE index to compare and analyze the decryption error of Bob and Eve, and the comparison result is as shown in Figure 6 The bar height of Bob's decryption error is very low and always tends to the zero plane, while the breaking error of Eve shows a significant height and sharp fluctuation, and the two form a clear order of magnitude difference. This result again verifies the accuracy of the encryption module for authorized users and the robustness to attackers from the perspective of error quantization.

[0069] In summary, Figures 4-6 From the perspective of correlation and error quantization, the effectiveness of the proposed adversarial encryption framework is intuitively verified, which clearly shows the bivalence of the encryption mapping: for the legitimate user holding the key, the mapping is an approximate identity transformation, which ensures the lossless restoration of information; and for the attacker, the mapping degenerates into a random permutation, which cannot restore any effective information.

[0070] To verify that the proposed deep residual network structure (RESMLP) can improve the accuracy, ordinary MLP network / 3 layers / 5 layers / 8 layers / 12 layers RESMLP are used for model training, and 10000 groups of randomly generated data are used to test the decryption effect of Bob network, and the results are as shown in Figure 7 The RMSE, MAE, and MAPE are prediction evaluation indexes, which are root mean square error, mean absolute error, and mean absolute percentage error, respectively.

[0071] From the results, it can be seen that the introduction of residual network can improve the accuracy of the model, but the network depth must match the intrinsic complexity of the problem and the amount of available data. In the above experiment, the model trained by the 8-layer network has the highest decryption accuracy. This shows that the residual network is not the more the better, and beyond a certain optimal point, there may be a risk of overfitting and increase the training cost of the model.

[0072] Next, set , other parameters remain unchanged, train multiple deep learning networks, and test the effect by using 10000 groups of randomly generated data, and the results are shown in Table 3.

[0073] Table 3: Comparison of decryption errors under different hyperparameters

[0074] As can be seen from Table 3, as the increase of , the decryption error will become larger, the data fidelity will decrease, but the anti-cracking ability will be stronger. Therefore, by adjusting the hyperparameter , the fidelity and anti-cracking weight of the encryption and decryption module can be adjusted.

[0075] To verify the encryption efficiency of the proposed encryption system, the performance of the homomorphic encryption algorithm commonly used in distributed optimization is compared. In the comparative experiment, the key length is set to 16 bits, and the test environment is Intel(R) Xeon(R) Platinum8358P @ 2.60GHz with 1 cores, 256MB RAM virtual machine environment. To avoid the contingency of experimental results, two methods are used to test 200, 500, 1000, and 2000 groups of data. The experimental results are shown in Table 4. As can be seen from the table, the method proposed in this paper is far superior to the homomorphic encryption method in terms of encryption efficiency, and can meet the demand for fast encryption.

[0076] Table 4: Encryption speed comparison

[0077] (3) Performance of the preset time distributed optimization algorithm First, the convergence performance of the algorithm is discussed. According to the above analysis, the thermal and electrical energy forms have different scheduling time scales. Here, the preset convergence time of the thermal energy related subject is set to T c1 =40s, and the preset convergence time of the electrical energy related subject is set to T c2 =10s. Other parameters are as described above. The power curve of the system is shown in Figure 8 , and the supply-demand balance curve of the system is shown in Figure 9 .

[0078] As can be seen from Figure 8 , Figure 9 , using the algorithm proposed in this embodiment, the system can be guaranteed to converge to the optimal value within the preset time, and due to the introduction of the projection operator, it is guaranteed that each subject in the system realizes supply-demand balance under the premise of satisfying the inequality constraint. The above example shows that the proposed algorithm has good convergence.

[0079] Secondly, the plug-and-play feature of the algorithm is verified. This embodiment performs the following experiments: 1) from 0 to 25s, T c1 =10s, ; 2) in t= 25s, cut off the 200kW heat load; 3) At t = 50s, reconnect the 200kW heat load and set T c1 =15s; 4) at t=75s, disconnect device P2; 5) at t=100s, reconnect device P2. The simulation results are as follows Figure 10 As shown. Figure 10 It can be seen that the system can automatically adjust to the new optimal operating state in the event of load fluctuations and equipment switching, and the convergence time can be flexibly set according to demand. This shows that the proposed method has good plug-and-play characteristics.

[0080] Finally, it is verified that the proposed distributed algorithm can still meet the preset time convergence under dynamic topology conditions. Taking the thermal energy related subject as an example, set T c1 = 20s, the initial network topology is as follows Figure 3 As shown. t =10s, the network connection between B1 and B2 is cut off, but the communication topology is still guaranteed to be an undirected connected graph. The power convergence curve is as follows Figure 11 As shown. Figure 11 It can be seen that at t=10s, the network topology within the system changes, causing the power convergence curve to fluctuate, but the system can still achieve the preset time convergence under the new topology. The above results show that the algorithm proposed in this embodiment has good robustness under dynamic topology conditions.

[0081] (4) Privacy and Accuracy Analysis In order to verify that the designed encryption system can significantly improve the privacy of interactive information in the distributed optimization process, the privacy protection performance of the encryption system is analyzed by taking the interactive information between node B1 and node B2 as an example.

[0082] Auxiliary variables that need to interact in the algorithm y and z Encrypted by the encryptor, the variables used for interaction are obtained. The values ​​before and after encryption are as follows Figure 12 As shown. Figure 12 It can be seen that there are obvious differences between the encrypted ciphertext and the unencrypted plaintext in terms of both order of magnitude and change trend. This method can significantly improve the privacy of the algorithm.

[0083] In order to verify the accuracy of the decryptor and ensure that the algorithm can converge accurately, 15 points are taken continuously at t=1.25s, and the data decrypted by the decryptor is compared with the original plaintext. The results are as follows: Figure 13 As shown. Figure 13 It can be seen that the error between the decrypted data and the original plaintext is extremely small, which can ensure that the algorithm converges to the optimal value and has good accuracy.

[0084] Embodiment Two The embodiment provides a privacy-considered integrated energy system distributed optimization scheduling system, which comprises the following modules: a system topology graph construction module, configured to construct a communication network topology graph of the integrated energy system by taking each device in the system as a node and taking energy interaction between the nodes as an edge; a scheduling model construction module, configured to construct an integrated energy system optimization scheduling model by taking minimization of total cost as an objective function and configuring constraint conditions for system operation; an optimal scheduling solution module, configured to assume that the communication network topology graph is a connected graph, adopt a preset time distributed optimization algorithm with time-varying gain, solve the optimization scheduling model, make the integrated energy system converge to an optimal solution from an arbitrary initial state within a preset time, and obtain an optimal optimization strategy; wherein, in the solving process, an encryption and decryption module based on a deep learning adversarial network is used to encrypt and decrypt interaction variables between neighbor nodes, so as to realize privacy protection.

[0085] Embodiment Three The embodiment provides an electronic device, comprising: a memory configured to store executable instructions; and a processor configured to execute the executable instructions stored in the memory to implement the method provided in the embodiment.

[0086] Embodiment Four The embodiment also provides a computer-readable storage medium storing executable instructions, which, when executed by a processor, cause the processor to perform the method provided in the embodiment.

[0087] The steps and method of the above embodiments two to four correspond to the method of embodiment one, and the specific implementation can be referred to the related description part of embodiment one. The term "computer-readable storage medium" should be understood as including a single medium or multiple media of one or more instruction sets; and should also be understood as including any medium capable of storing, encoding or carrying instruction sets for execution by a processor and causing the processor to perform any method in the present application.

[0088] Those skilled in the art should understand that each module or step of the present application described above can be realized by a general computer device, and alternatively, they can be realized by program codes executable by a computing device, so that they can be stored in a storage device for execution by a computing device, or they can be respectively manufactured into each integrated circuit module, or a plurality of modules or steps among them can be manufactured into a single integrated circuit module. The present application is not limited to any specific combination of hardware and software.

[0089] The above merely describes the preferred embodiments of the present application, and the specific embodiments of the present application are described in conjunction with the drawings, but are not intended to limit the protection scope of the present application, and those skilled in the art should understand that various modifications or changes made on the basis of the technical solutions of the present application without creative labor are still within the protection scope of the present application.

Claims

1. A distributed optimization scheduling method for an integrated energy system considering privacy protection, characterized in that: include: The communication network topology of the integrated energy system is constructed by taking each device in the system as a node and the energy interaction between nodes as an edge; Taking the minimization of total cost as the objective function, configuring the constraints of system operation, and building an optimal scheduling model for the integrated energy system; Assuming that the communication network topology is an undirected connected graph, a preset-time distributed optimization algorithm with time-varying gain is adopted to solve the optimization scheduling model, so that the integrated energy system converges to the optimal solution from any initial state within the preset time, thereby obtaining the optimal optimization strategy. During the solution process, the interaction variables between neighbor nodes are encrypted and decrypted through an encryption and decryption module based on a deep learning adversarial network to achieve privacy protection.

2. The distributed optimization scheduling method for an integrated energy system considering privacy protection according to claim 1, characterized in that: The encryption and decryption module based on deep learning adversarial network is implemented based on a multi-agent game framework, including an encryptor Alice, a decryptor Bob and an attacker Eve constructed based on a neural network; The encryptor Alice and the decryptor Bob form a cooperative alliance, and the joint key is deployed in the neighboring nodes. The interaction variable between the neighboring nodes is the plaintext vector. The encryptor Alice uses the plaintext vector and the key as input to learn to generate the ciphertext vector. The decryptor Bob, as the authorized receiver, uses the ciphertext vector and the same key as input to learn to reconstruct the original plaintext vector. The attacker Eve forms an adversarial game with the cooperative alliance, and guides the encryptor Alice and the decryptor Bob to learn complex nonlinear encryption transformations through adversarial training.

3. The distributed optimization scheduling method for an integrated energy system considering privacy protection according to claim 2, characterized in that: The encryptor Alice, decryptor Bob, and attacker Eve are all built using a multi-layer perceptron with deep residuals as their infrastructure. The multi-layer perceptron with deep residuals includes an input linear layer, a core network consisting of multiple residual blocks stacked together, and an output linear layer. Here, each residual block is defined as H(X)=X+F(X), where F(X) is the residual map learned by the network layer and X is the original input passed through the shortcut connection; the network layer includes a linear layer, a batch normalization layer, and an activation layer connected in sequence.

4. The distributed optimization scheduling method for an integrated energy system considering privacy protection according to claim 2, characterized in that: In each training batch, the encryption and decryption modules are trained using a two-stage update strategy, including: In phase 1, the network parameters of Alice the encryptor and Bob the decryptor are frozen. The ciphertext vector generated by Alice the encryptor is passed to Eve the attacker, and Alice the encryptor's weight is not updated. Based on minimizing Eve's reconstruction loss, backpropagation is performed to update Eve's network weight. In the second phase, the attacker Eve's network parameters are frozen, and the encryptor Alice and the decryptor Bob perform the complete encryption and decryption process, and perform backpropagation based on the minimization of the composite loss function, while updating the network weights of the encryptor Alice and the decryptor Bob.

5. The distributed optimization scheduling method for an integrated energy system considering privacy protection according to claim 4, characterized in that: The encryption and decryption modules are end-to-end optimized using an asymmetric loss function. The composite loss function of the cooperative alliance formed by the encryptor Alice and the decryptor Bob is: ； ； ； in, 、 are the L1 reconstruction losses of decryptor Bob and attacker Eve respectively, U is the original plaintext vector, U' is the decryption result of decryptor Bob, and U'' is the cracking result of attacker Eve. is a hyperparameter.

6. The distributed optimization scheduling method for an integrated energy system considering privacy protection according to claim 1, characterized in that: The integrated energy system is a thermoelectric coupling energy system, and its objective function is to minimize the sum of the operating costs of all power supply equipment, heating equipment, thermoelectric coupling equipment, and energy storage equipment; the constraints include: thermoelectric ratio constraint, energy supply and demand balance constraint, and safe operation constraint.

7. The distributed optimization scheduling method for an integrated energy system considering privacy protection according to claim 1, characterized in that: The preset time distributed optimization algorithm with time-varying gain is used to solve the optimization scheduling model, which is: ； in, , is the gradient, Indicates the i The cost function of each node, Indicates the i The power of each node; is the projection operator, is a closed convex set, For about Inequality constraint function of ; definition is the Laplacian operator of the graph, represents the adjacency matrix, Indicates the adjacency matrix i Rank j elements, nodes i The neighbor node set of N i }, 、 、 is an auxiliary variable; is the time-varying gain function introduced, is a constant gain.

8. A distributed optimization scheduling system for an integrated energy system considering privacy protection, characterized in that: include: The system topology construction module is used to construct the communication network topology of the integrated energy system, using each device in the system as a node and the energy interaction between nodes as an edge; The scheduling model construction module is used to configure the constraints of system operation and build an optimized scheduling model for the integrated energy system with the minimum total cost as the objective function; The optimal scheduling solution module is used to assume that the communication network topology is an undirected connected graph, and adopts a preset time distributed optimization algorithm that introduces time-varying gain to solve the optimization scheduling model, so that the integrated energy system converges from any initial state to the optimal solution within the preset time and obtains the optimal optimization strategy; in the solution process, the interaction variables between neighbor nodes are encrypted and decrypted through the encryption and decryption module based on the deep learning adversarial network to achieve privacy protection.

9. An electronic device, characterized in that: include: a memory for storing executable instructions; The processor is configured to implement the distributed optimization scheduling method for an integrated energy system considering privacy protection as described in any one of claims 1 to 7 when executing the executable instructions stored in the memory.

10. A computer-readable storage medium, characterized in that Executable instructions are stored, which are used to cause the processor to execute the executable instructions to implement the distributed optimization scheduling method of the integrated energy system considering privacy protection as described in any one of claims 1-7.

Images