A method for outsourcing calculation and parameter desensitization of distribution network dispatch decision model

Abstract

The present invention discloses a method for outsourcing calculation of a distribution network dispatching decision model and its parameter desensitization. First, based on the structural characteristics of the distribution network system, a distribution network dispatching decision model is constructed and characterized in the standard form of the dispatching decision problem. Then, based on the scale-free network model, a companion virtual distribution network with the same scale as the real distribution network is constructed to protect the physical information of the real distribution network. Furthermore, according to the deceptive privacy protection method, a conversion key between the real distribution network and the companion virtual distribution network is designed, and a virtual distribution network dispatching decision model with privacy protection properties is generated based on the key. Finally, a double verification mechanism for cloud service providers is designed to ensure the optimality of the calculation results. On the one hand, the present invention can deal with the potential information security risks brought about by data migration to the cloud; on the other hand, by reasonably designing the key, the construction of the virtual distribution network dispatching decision model can ensure the sparsity of the encryption problem and improve the efficiency of cloud computing.

Description

Technical Field

[0001] The present invention relates to the field of electrical engineering technology, and in particular to a method for outsourcing calculation of a distribution network scheduling decision model and a parameter desensitization method thereof. Background Art

[0002] With the vigorous advancement of incremental distribution business reforms, incremental distribution network companies have emerged as independent stakeholders, independently investing in and operating local distribution networks (such as campus distribution networks and community microgrids). However, these stakeholders are typically small and medium-sized enterprises (SMEs) that lack the computing power and professional management teams comparable to grid companies, making it difficult to effectively address the increasingly complex challenges of distribution network dispatching and decision-making. At the same time, computing power is becoming a flexibly purchasable public resource. The integration of next-generation information technologies such as cloud computing is a growing trend in the distribution network. Currently, public cloud service providers offer commercial cloud computing solutions for scenarios such as smart microgrids and distributed renewable energy grid integration management. In this context, incremental distribution network companies can outsource their distribution network dispatching and decision-making analysis to cloud service providers, effectively addressing their own technical expertise and saving operating costs in hardware and software procurement, system maintenance, and on-site manpower.

[0003] However, incremental distribution network companies and public cloud service providers have different interests. Cloud computing requires users to upload data related to computing tasks to the cloud. This makes it possible for cloud service providers to spy on sensitive information of incremental distribution network companies, such as transaction prices, load data, and grid structure. This creates a potential risk of information leakage and is a major challenge to the widespread application of cloud computing technology. If information leaks from the power system can be used for malicious attacks by other entities, it could cause power safety incidents or direct economic losses. Therefore, privacy information encryption is a prerequisite for the widespread application of cloud computing technology.

[0004] Currently, research has proposed encrypting the model parameters of the scheduling decision problem using random nonsingular matrices to achieve privacy protection. However, the distribution network scheduling decision problem is a sparse problem. Using random nonsingular matrix encryption destroys the sparsity of the original problem, significantly increasing the complexity of the encryption problem. This reduces the efficiency and increases the cost of cloud computing, making it difficult to apply in practice. Therefore, it is urgent to find an encryption method that can protect data privacy while maintaining the computational efficiency of the distribution network scheduling decision model. Summary of the Invention

[0005] In response to the problems existing in the prior art, the present invention proposes a distribution network dispatching decision model outsourcing calculation and parameter desensitization method based on the affine transformation principle and the mathematical properties of the distribution network topology matrix to achieve the unity of safety and efficiency.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for outsourcing calculation of a distribution network dispatching decision model and its parameter desensitization, comprising the following steps: A first aspect of an embodiment of the present invention provides a method for outsourcing calculation of a distribution network dispatching decision model and its parameter desensitization, the method comprising the following steps:

[0007] (1) Based on the actual operating characteristics of the distribution network system, a standard representation distribution network decision model is constructed;

[0008] (2) According to the number of nodes in the distribution network system in step (1), a virtual distribution network topology is generated based on a scale-free network model, and then the admittance parameters of each branch are randomly scaled by random positive real numbers to construct an associated virtual distribution network with the same number of nodes as the distribution network system;

[0009] (3) Based on the associated virtual distribution network generated in step (2), a conversion key between the associated virtual distribution network and the real distribution network is designed according to the deceptive privacy protection method, and a virtual scheduling decision model is generated based on the key;

[0010] (4) According to steps (2) and (3), two different virtual distribution network topologies and virtual branch parameters are generated, two corresponding different associated distribution networks are constructed, and two different virtual dispatching decision models are designed. Then, based on the double verification mechanism, the different virtual dispatching decision models are outsourced to two different cloud service providers for calculation, and the calculation results of different cloud service providers are compared and verified to verify the optimality of the cloud computing results.

[0011] A second aspect of an embodiment of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the above-mentioned distribution network dispatching decision model outsourcing calculation and parameter desensitization method.

[0012] A third aspect of an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned distribution network dispatching decision model outsourcing calculation and parameter desensitization method.

[0013] Compared with the prior art, the present invention has the following beneficial effects:

[0014] 1. In the process of outsourcing the calculation of the distribution network dispatch decision model in the method of the present invention, the uploaded cloud data, such as grid information and generator parameters, can be effectively encrypted, thereby reducing the information security risks brought by cloud computing;

[0015] 2. In the process of desensitizing the parameters of the distribution network dispatch decision model in the method of the present invention, the conversion principle between grid structures is deduced to realize the conversion of the real grid into the associated virtual grid, thereby ensuring the sparsity of the encryption model and doubling the efficiency of solving the encryption model;

[0016] 3. In the process of outsourcing calculation and verification of the distribution network in the method of the present invention, a double verification mechanism is constructed, and the virtual scheduling decision model is outsourced to two cloud service providers with competing interests. The calculation results of different cloud service providers are compared and verified to further ensure the correctness of the cloud computing results. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a framework diagram of a distribution network dispatch decision model outsourcing calculation and parameter desensitization method according to the present invention;

[0018] Figure 2 Schematic diagram of distribution network topology transformation;

[0019] Figure 3 A schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation examples.

[0021] Figure 1 This is a framework diagram of a distribution network dispatch decision model outsourcing calculation and parameter desensitization method of the present invention, including the following steps:

[0022] (1) According to the actual operating characteristics of the distribution network system, a standard representation distribution network dispatch decision model is constructed.

[0023] The dispatch decision constraints include line static thermal stability constraints, unit output upper and lower limits, unit ramping constraints, node voltage safety constraints, and node phase angle constraints. Based on the dispatch decision constraints and the objective function of the distribution network dispatch decision model, a distribution network dispatch decision model is constructed. To facilitate analysis and modeling, the distribution network dispatch decision model must be represented in a standard form.

[0024] Specifically, the objective function and constraints of the distribution network dispatch decision model are as follows:

[0025] The objective function of the scheduling decision model is:

[0026]

[0027] Among them, p t represents the vector composed of the active output of each generator at time t; C a represents the diagonal matrix composed of the quadratic coefficients of each generator's power generation cost; cb 、c c Represents the vector consisting of the primary coefficient and constant term of each generator’s power generation cost, Represents the vector c b 、c c Transpose; 1 represents a vector with 1 elements and the same dimension as c c same.

[0028] The node power balance constraint of the scheduling decision model is:

[0029] -Gv t +Bθ t +N G p t =p d,t

[0030] Bv t +Gθ t +N G q t =q d,t

[0031] Where G / B represents the real part / imaginary part of the grid node admittance matrix; v t / θ t Represents the vector composed of voltage / phase angle of each node at time t; N G Represents the node-unit association matrix. If unit j is on node i, the matrix element N G,i,j is 1, otherwise it is 0; t represents the vector composed of reactive output of each generator at time t; p d,t / q d,t Represents the vector composed of active / reactive loads at each node at time t.

[0032] The line thermal stability constraint of the scheduling decision model is:

[0033] -f max ≤G b A T v t -B b A T θ t ≤f max

[0034] Among them, G b / B b represents the real / imaginary part of the branch admittance matrix; A represents the grid node-branch association matrix; f max Represents the branch maximum transmission power upper limit vector.

[0035] The node status safety constraints of the scheduling decision model are:

[0036] v min ≤v t ≤v max

[0037] θ min ≤θ t ≤θ max

[0038] Among them, v max / v min Represents the vector consisting of the upper and lower limits of the voltage of each node; θ max / θ min A vector representing the upper / lower limits of the phase angle at each node.

[0039] The generator output constraint of the scheduling decision model is:

[0040] p min ≤p t ≤p max

[0041] q min ≤q t ≤q max

[0042] Among them, p max / p min The vector representing the upper and lower limits of each unit’s active output; q max / q min A vector representing the upper / lower limits of reactive output of each unit.

[0043] The distribution network dispatch decision model is relaxed and standardized to form the following specific form:

[0044] P QP :min x T Cx+c T x

[0045] stMx=b

[0046] I s x≥0

[0047] Where C∈R n×n The quadratic coefficient matrix representing the cost of the decision variables is a semi-positive matrix, denoted by C∈S + ;c∈R n×1 A first-order coefficient vector representing the cost; M∈R m×n , b∈R m×1 Represents the equality constraint coefficient matrix / vector; I s ∈R h×n The coefficient matrix representing the inequality constraints; 0∈R h×1is a zero vector with the same dimension as the inequality constraint;

[0048] Furthermore, in the standard scheduling decision model, the structure of the decision variables is:

[0049]

[0050] Among them, x pri represents the actual decision variables, including p t ,q t 、v t ,θ t ;x sl represents the corresponding slack variable after the inequality constraint is relaxed. Specifically, x pri The structure is as follows:

[0051]

[0052] Furthermore, the specific structure of each parameter of the standard model is:

[0053]

[0054]

[0055] Among them, C pri 、c pri Represents x pri The cost coefficient of the objective function is that only the active output p of the generator is a non-zero coefficient, and the other decision variables are 0; sl Represents x sl The cost coefficient of M is 0; eq 、b eq Represents the coefficient matrix / vector of the equality constraint; M ine 、b ine Represents the coefficient matrix / vector of the inequality constraint respectively; P sl Represents x sl The corresponding random monomial matrix.

[0056]

[0057] Among them, I v , I θ , I g , I q They represent the unit matrix coefficients corresponding to voltage, phase angle, generator active output, and generator reactive output respectively.

[0058] (2) According to the number of nodes in the real distribution network system in step (1), a virtual distribution network topology is generated based on the scale-free network model, and then the admittance parameters of each branch are randomly scaled by random positive real numbers to construct an accompanying virtual distribution network with the same number of nodes as the distribution network system. Figure 2 shown.

[0059] The method for generating the associated virtual distribution network in step (2) is specifically as follows:

[0060] In the first stage, a virtual distribution network topology A is generated. enc For a radial distribution network topology A with n nodes and n-1 branches, the algorithm for constructing its associated virtual distribution network topology using the scale-free network model is as follows: set an initial node n1, then gradually add a node and connect it to an existing node. i When n i Connect to existing node n j The probability Π(k j ) and n j degree k j Related to:

[0061]

[0062] Among them, k l Represents node n l degree, that is, the degree of node n l Number of connected branches; Represents the sum of the degrees of existing nodes.

[0063] Through the above algorithm, after gradually adding n-1 new nodes and completing the connection, a radial topology of n nodes and n-1 branches can be generated, which will be used as the virtual distribution network topology A later. enc .

[0064] In the second stage, the virtual branch parameters are generated. The real branch parameters G are calculated by a diagonal matrix Λ with random positive real coefficients. b and B b Scaling is performed to obtain the virtual branch parameter G as shown below b,enc and B b,enc .

[0065] G b,enc =G b ΛB b,enc =B b Λ

[0066] Through this two-stage operation, for the grid topology, due to the virtual topology A encIt can be any radial network structure, so it can be considered that the cloud service provider cannot be A enc Inversely deduce the real distribution network topology; for parameters, since G b,enc and B b,enc By G b and B b Random scaling is obtained, so the information of the real distribution network branch parameters can also be hidden. The distribution network parameter desensitization is achieved. For example, Figure 2 As shown in the figure, the virtual distribution network protects the privacy of the network topology and branch parameters of the real distribution network. Taking the connection relationship and branch parameters between node 1 and node 6 as an example, the real direct connection and g2+jb2=347-j177 are encrypted to no direct connection and g2+jb2=173-j88.

[0067] (3) Based on the associated virtual distribution network generated in step (2), a conversion key between the associated virtual distribution network and the real distribution network is designed according to the deceptive privacy protection method, and a virtual scheduling decision model is generated based on the key.

[0068] The key design method is:

[0069]

[0070]

[0071] Among them, P A 、P A,enc 、 All are components of the key; is a matrix The transpose of the row simplest transformation matrix of , the specific form of the remaining key components is:

[0072] P A =[A l] -1

[0073] P A,enc =[A enc l] -1

[0074]

[0075]

[0076]

[0077] Among them, I f Represents the identity matrix with dimensions corresponding to the number of branches.

[0078] Then, a virtual scheduling decision model is generated based on the key. The formula of the virtual scheduling decision model is as follows:

[0079]

[0080] sT enc x=b enc

[0081] I s x≥0

[0082] Among them, C enc 、c enc 、M enc 、b enc are the parameters of the virtual scheduling decision model.

[0083] The conversion relationship between the coefficients of the virtual scheduling decision model and the real scheduling decision model is as follows:

[0084] C enc =βQ T CQ

[0085] c enc =β(Q T c-2Q T Cr)

[0086] M enc =PMQ

[0087] b enc =P(b+Mr)

[0088] Where β is a random positive real number and r is a random vector.

[0089] (4) According to the method proposed in steps (2) and (3), considering the probability Π(k j ) and the randomness of the diagonal matrix Λ, resulting in two different virtual distribution network topologies A enc , virtual branch parameter G b,enc and B b,enc , constructing two corresponding associated distribution networks and designing a virtual dispatch decision model for each associated distribution network. Then, based on a double-verification mechanism, the different virtual dispatch decision models were outsourced to two different cloud service providers for computation. The computational results from the different cloud service providers were decrypted, compared, and verified to verify the optimality of the cloud computing results.

[0090] The decryption method of the calculation result returned by the cloud service provider is:

[0091]

[0092]

[0093] Among them, x * represents the solution of the real scheduling decision model; represents the solution of the virtual scheduling decision model.

[0094] The optimality of cloud computing results is judged through a double-check mechanism. The judgment principle is divided into two steps, specifically:

[0095] (a) Validity judgment. Substitute the result returned by the cloud service provider into the model constraints in step (1) to determine whether it meets the constraint requirements. If it does, it is a valid solution and further optimality judgment is performed; if it does not, it is an invalid solution and the judgment result is returned to the cloud service provider, which is punished to recalculate.

[0096] (b) Optimality is then determined. After both cloud providers return valid solutions, the cloud computing results are decrypted and compared based on the decrypted results. If the decrypted results of the two cloud providers match, the results returned by both are considered the optimal solution. If they do not match, the solutions of the real scheduling decision model are compared. The cloud provider with the larger real scheduling decision model solution can be judged as non-optimal, and the provider is penalized to recalculate.

[0097] Example 1: Verifying the computational efficiency of the distribution network dispatch decision model after parameter desensitization

[0098] The present invention is verified using the IEEE European 906 node low-voltage system, and the computational efficiency is compared with the conventional encryption method based on random matrix affine transformation. The computational efficiency of the four stages of key generation, model encryption, cloud computing, and result decryption are analyzed under three different time scales of 1 hour, 8 hours, and 24 hours.

[0099] In terms of solution correctness, the method proposed in the present invention can obtain the optimal solution of the real scheduling decision model after decrypting the virtual scheduling decision model, which proves that the privacy protection method proposed in the present invention can obtain the correct optimal calculation results.

[0100] In terms of computational efficiency, during the key generation phase, the proposed method achieved key generation times of 1.6s, 1.8s, and 2.0s at three different time scales: 1 hour, 8 hours, and 24 hours. This time does not significantly increase with increasing problem size. Conventional methods achieved key generation times of 1.7s and 119.2s at 1 hour and 8 hours, respectively. This time cost increases rapidly with increasing problem size, and a memory overflow occurs at the 24-hour scale.

[0101] During the model encryption phase, the proposed method achieved high encryption efficiency, taking 0.7s, 0.8s, and 1.3s at three time scales, respectively. However, conventional encryption methods took 8.7s at a one-hour scale and exceeded computing capacity at 8-hour and 24-hour scales.

[0102] During the cloud computing phase, the computational cost of the proposed method increases linearly with the problem size, with computation times of 8.4s, 64.2s, and 237.1s for the three timescales, respectively. However, conventional encryption methods cannot solve the problem at the one-hour timescale due to its excessive complexity.

[0103] Results: In the decryption phase, the method of the present invention takes 0.3s, 0.5s and 0.9s respectively, which is a low decryption time cost. However, the conventional method cannot obtain a decryption result because the pre-stage is not completed.

[0104] In summary, the proposed method for outsourcing the computation of a distribution network dispatch decision model and its parameter desensitization effectively protects the physical information of the actual distribution network and, using a decryption formula, restores the true optimal dispatch decision result. Furthermore, compared to existing methods, the proposed method offers significant advantages in computational efficiency, effectively reducing the complexity of solving the encryption problem and enabling rapid encryption and decryption operations.

[0105] like Figure 3 As shown, it will be understood by those skilled in the art that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0106] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0107] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for outsourcing calculation of a distribution network dispatch decision model and its parameter desensitization, characterized in that: The method comprises the following steps: (1) Based on the actual operating characteristics of the distribution network system, a standard representation distribution network decision model is constructed; (2) According to the number of nodes in the distribution network system in step (1), a virtual distribution network topology is generated based on a scale-free network model, and then the admittance parameters of each branch are randomly scaled by random positive real numbers to construct an associated virtual distribution network with the same number of nodes as the distribution network system; (3) Based on the associated virtual distribution network generated in step (2), a conversion key between the associated virtual distribution network and the real distribution network is designed according to the deceptive privacy protection method, and a virtual scheduling decision model is generated based on the key; (4) According to steps (2) and (3), two different virtual distribution network topologies and virtual branch parameters are generated, two corresponding associated distribution networks are constructed, and two different virtual dispatching decision models are designed. Then, based on the double verification mechanism, the different virtual dispatching decision models are outsourced to two different cloud service providers for calculation, and the calculation results of different cloud service providers are compared and verified to verify the optimality of the cloud computing results; The standard representation distribution network decision model constructed in step (1) is as follows: P QP :min x T Cx+c T x stMx=b I s x≥0 Where C∈R n×n The quadratic coefficient matrix representing the cost of the decision variables is a semi-positive matrix, denoted by C∈S + ;c∈R n×1 A first-order coefficient vector representing the cost; M∈R m×n , b∈R m×1 Represents the equality constraint coefficient matrix / vector; I s ∈R h×n The coefficient matrix representing the inequality constraints; 0∈R h×1 is a zero vector with the same dimension as the inequality constraint; In the standard scheduling decision model, the structure of the decision variables is: Among them, x pri represents the actual decision variables, including p t ,q t 、v t ,θ t ;x sl represents the slack variable corresponding to the relaxation of the inequality constraint; x pri The structure is as follows: The specific structure of each parameter of the standard model is: Among them, C pri 、c pri Represents x pri The cost coefficient of the objective function is that only the active output p of the generator is a non-zero coefficient, and the other decision variables are 0; sl Represents x sl The cost coefficient of M is 0; eq 、b eq Represents the coefficient matrix / vector of the equality constraint; M ine 、b ine Represents the coefficient matrix / vector of the inequality constraint respectively; P sl Represents x sl The corresponding random single-entry matrix; Among them, I v , I θ , I g , I q Represent the unit matrix coefficients of the variables corresponding to voltage, phase angle, generator active output, and generator reactive output respectively.

2. A distribution network dispatch decision model outsourcing calculation and parameter desensitization method according to claim 1, characterized in that: The method for generating the associated virtual distribution network in step (2) is specifically as follows: In the first stage, for a radial distribution network topology A with n nodes and n-1 branches, a scale-free network model is used to construct its associated virtual distribution network topology A. enc :Set an initial node n1, then gradually add a node and connect it to an existing node; when connecting the i-th node n i When n i Connect to existing node n j The probability ∏(k j ) and n j degree k j Related to: Among them, k l Represents node n l degree, that is, the degree of node n l Number of connected branches; Represents the sum of the degrees of existing nodes; By using the above method, after gradually adding n-1 new nodes and completing the connection, a radial topology with n nodes and n-1 branches can be generated as the virtual distribution network topology A. enc ; In the second stage, the virtual branch parameters are generated by using a diagonal matrix Λ with random positive real coefficients to generate the real branch parameters G b 、B b Scaling is performed to obtain the virtual branch parameter G as shown below b,enc 、B b,enc : G b,enc =G b ΛB b,enc =B b Λ Using virtual branch parameter G b,enc 、B b,enc Hiding the real virtual distribution network topology A enc The branch parameter information is used to complete the desensitization of distribution network parameters.

3. A distribution network dispatch decision model outsourcing calculation and parameter desensitization method according to claim 1, characterized in that: The formula for the key design method in step (3) is as follows: Among them, P A 、P A,enc 、 Q NG All are components of the key; is a matrix The transpose of the row simplest transformation matrix of , the specific form of the remaining key components is: P A =[A l] -1 P A,enc =[A enc l] -1 Among them, I f Represents the identity matrix with dimensions corresponding to the number of branches.

4. A distribution network dispatch decision model outsourcing calculation and parameter desensitization method according to claim 3, characterized in that: The formula of the virtual scheduling decision model in step (3) is as follows: s.t.M enc x=b enc I s x≥0 Among them, C enc 、c enc 、M enc 、b enc are the parameters of the virtual scheduling decision model; The conversion relationship between the coefficients of the virtual scheduling decision model and the real scheduling decision model is as follows: C enc =βQ T CQ c enc =β(Q T c-2Q T Cr) M enc =PMQ b enc =P(b+Mr) Where β is a random positive real number and r is a random vector.

5. A distribution network dispatch decision model outsourcing calculation and parameter desensitization method according to claim 1, characterized in that: The decryption method of the calculation result returned by the cloud service provider in step (4) is: Among them, x * represents the solution of the real scheduling decision model; represents the solution of the virtual scheduling decision model.

6. A distribution network dispatch decision model outsourcing calculation and parameter desensitization method according to claim 1, characterized in that: The step (4) compares and verifies the calculation results of different cloud service providers. The process of verifying the optimality of the cloud computing results includes: (a) Validity judgment: Substitute the results returned by the cloud service provider into the constraints of the distribution network decision model to determine whether they meet the constraints. If they do, the solution is valid and further optimality judgment is performed. If they do not meet the constraints, the solution is invalid and the judgment result is returned to the cloud service provider, which is punished to recalculate. (b) Optimality judgment: After both cloud service providers return valid solutions, the cloud computing results are decrypted and compared based on the decrypted results. If the decrypted results calculated by the two cloud service providers are consistent, the results returned by both parties can be considered the optimal solution. If they are inconsistent, the solutions of the real scheduling decision model are compared. It can be determined that the solution of the virtual scheduling decision model returned by the cloud service provider with the larger solution of the real scheduling decision model is not the optimal solution, and it is penalized to recalculate.

7. An electronic device comprising a memory and a processor, characterized in that: The memory is coupled to the processor; wherein, the memory is used to store program data, and the processor is used to execute the program data to implement the distribution network dispatching decision model outsourcing calculation and parameter desensitization method as described in any one of claims 1-6 above.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the distribution network dispatching decision model outsourcing calculation and parameter desensitization method as described in any one of claims 1 to 6 are implemented.

Images