Power distribution system uncertainty online scheduling method considering privacy protection demand

Abstract

The invention specifically relates to a power distribution system uncertainty on-line scheduling method considering privacy protection demands, and the method comprises the following steps: S1), building a power distribution system operation cost model considering uncertainty new energy input, and enabling a to-be-optimized cost function to be the sum of power generation cost functions of all power distribution region subsystems in a period of time, the constraint comprises a power generation constraint of a power distribution area subsystem and a time-varying power generation and utilization balance coupling constraint; s2) establishing a Lagrange dual optimization model corresponding to the operation optimization problem of the original power distribution system, and solving the power generation and power utilization balance coupling constraint by decomposing the global time-varying power utilization demand and the uncertain new energy total input into a local power utilization demand and an uncertain new energy region input; and S3) aiming at the obtained Lagrange dual optimization model, constructing a power distribution system uncertainty economic dispatching collaborative optimization solution method based on a differential privacy technology, and giving out local power generation optimal solutions of all power distribution area subsystems.

Description

Technical Field

[0001] The present invention relates to the technical field of economic dispatch optimization of distribution networks, and in particular to an uncertainty online dispatch method for distribution systems taking into account privacy protection requirements. Background Art

[0002] In recent years, how to improve the overall efficiency of the distribution network system and reduce costs has become one of the key issues for the power industry to achieve its goals. With the development and popularization of smart grid technology, the efficient operation of the distribution system is increasingly dependent on accurate data transmission and analysis capabilities. However, in this process, data security and privacy protection have become issues that cannot be ignored. Traditional power system dispatching methods often ignore the protection of key system parameters, which may lead to user privacy leakage, provide a design basis for malicious attackers to conduct physical information attacks, and bring huge challenges to energy security and information security.

[0003] In the distribution system, uncertain factors such as time-varying renewable energy generation and variable loads pose new challenges to the collaborative solution of economic dispatch. The increase of these uncertainties makes it difficult for traditional deterministic optimization models to adapt to the complexity and dynamics of modern power systems. Therefore, studying how to effectively deal with these uncertainties while ensuring the efficiency of economic dispatch and ensuring the security of key privacy parameters has become a key issue that needs to be solved urgently.

[0004] Literature 1 "Differential Dispatch Optimization Method for Power Communication Information Based on Homomorphic Encryption" (Microcomputer Applications, 2023, Vol. 39, No. 6, pp. 156-158) proposes a differential dispatch optimization method based on homomorphic encryption to improve the security of power communication information. By constructing homomorphic data encryption keys, arithmetic coding and other modules, the privacy protection function of power system communication information is realized. However, the method mentioned in this document only focuses on the protection of the transmission data level, and does not consider the protection requirements for specific key privacy parameters in the entire dispatch optimization process. In addition, the distribution network optimization dispatch model constructed in Literature 2 "Research on Optimal Dispatch of Distribution Network Based on Uncertainty of New Energy Generation" (China New Technologies and New Products, 2024, No. 7, pp. 25-27) takes into account the uncertain effects of wind power and photovoltaic power generation, and designs a multi-agent distribution management system composed of regional agents and unit agents, realizing real-time optimal dispatch of the distribution network. However, the hierarchical structure and non-encrypted information transmission method of the multi-agent system make it have certain limitations in privacy protection function and scalability.

[0005] Therefore, developing a method that can not only handle the inherent uncertainty of renewable energy input in the distribution system, but also perform collaborative economic dispatch while protecting key privacy parameters is necessary to achieve the efficiency and privacy of the distribution network system. Summary of the Invention

[0006] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide an online scheduling method for the uncertainty of a distribution system considering privacy protection requirements. By introducing online optimization technology and adding a differential privacy protection link in the interaction and communication of subsystems, on the premise of ensuring that key privacy parameters will not be leaked, the renewable energy input with uncertainty characteristics is fully considered, and the local generation optimization solutions of all distribution area subsystems are given, reducing the overall operation cost of the distribution network.

[0007] The present invention adopts the following technical solutions to achieve the above-mentioned invention purpose: An economic dispatch collaborative solution method for the uncertainty of a distribution system considering privacy protection requirements, including the following steps:

[0008] S1: Establish an operation cost model of a distribution system considering uncertain new energy input. The cost function to be optimized is the sum of the generation cost functions of all distribution area subsystems within a period of time, including the generation constraints of the distribution area subsystems and the time-varying generation and consumption balance coupling constraints;

[0009] S2: Establish a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem, and solve the generation and consumption balance coupling constraints by decomposing the global time-varying power consumption demand and the total uncertain new energy input into local power consumption demands and regional uncertain new energy inputs;

[0010] S3: For the obtained Lagrangian dual optimization model, a collaborative optimization solution method for the uncertainty economic dispatch of a distribution system based on differential privacy technology is constructed, and the local generation optimization solutions of all distribution area subsystems are given on the premise of ensuring that key privacy parameters cannot be inferred.

[0011] As a preferred technical solution of the present invention: The present invention proposes an operation cost model of a distribution system considering uncertain new energy input, which is specifically shown in formula (1):

[0012]

[0013] In formula (1), t is the optimization period index, (t ∈ [1, T]); T is the total number of artificially set optimization periods; n is the number of distribution area subsystems; g i (t) is the power generation to be optimized of the i-th distribution area subsystem in the optimization period t; r i (t) is the uncertain new energy input of the i-th distribution area subsystem in the optimization period t, which is a time-varying constant that cannot be optimized; is the generation cost of the i-th distribution area subsystem in the optimization period t; is the uncertain new energy usage cost of the i-th distribution area subsystem in the optimization period t;

[0014] In formula (1), the power generation cost of the $i$-th distribution area subsystem during the optimization period $t$ The specific expression is as shown in formula (2):

[0015]

[0016] In formula (2), and are the quadratic term and the linear term coefficient of the power generation cost of the $i$-th distribution area subsystem;

[0017] In formula (1), the cost of using uncertain new energy of the $i$-th distribution area subsystem during the optimization period $t$ Specifically, it is as shown in formula (3):

[0018]

[0019] In formula (3), $k$ r is the coefficient of the cost of using uncertain new energy;

[0020] The constraints in the distribution system operation cost model considering the input of uncertain new energy are as follows:

[0021] The power generation capacity constraint of the $i$-th distribution area subsystem:

[0022]

[0023] In formula (4), is the minimum power generation of the $i$-th distribution area subsystem; is the maximum power generation of the $i$-th distribution area subsystem;

[0024] The global power generation and power consumption balance coupling constraint of the distribution system:

[0025]

[0026] In formula (5), $D$ is the global power consumption demand of the distribution system during the optimization period $t$;

[0027] As a preferred technical solution of the present invention: The present invention establishes a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem, and solves the power generation and power consumption balance coupling constraint by decomposing the global time-varying power consumption demand and the total input of uncertain new energy into local power consumption demand and regional input of uncertain new energy, specifically as follows:

[0028] During the optimization period $t$, define the following Lagrangian function:

[0029]

[0030] In formula (6), $\mu$i (t) is the Lagrangian dual variable related to the global power generation and consumption balance coupling constraint of the distribution system; D i is the local power consumption demand of the i-th distribution area subsystem, satisfying Based on formula (6), at the optimization time period t, a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem can be established:

[0031]

[0032] As a preferred technical solution of the present invention: for the obtained Lagrangian dual optimization model, the present invention constructs a collaborative optimization solution method for the uncertain economic dispatch of the distribution system based on differential privacy technology, and gives the local generation optimization solutions of all distribution area subsystems, specifically as follows:

[0033] First, determine the out-degree and the communication weight matrix A = [a ij according to the fixed directed network topology composed of each sub-region in the distribution system. The specific formula is:

[0034]

[0035] In formula (8), is the set composed of all distribution area subsystems connected to the j-th distribution area subsystem as the signal output party;

[0036] Then, taking the i-th distribution area subsystem as an example, the collaborative optimization solution steps for the uncertain economic dispatch of the distribution system based on differential privacy technology are specifically as follows:

[0037] S3.1: Set the optimization time index to 0, that is, t = 0; initialize the subsystem local optimization variables and auxiliary variables:

[0038]

[0039]

[0040] Among them, p i (t), q i (t), z i (t) are auxiliary variables; q i (t) is used to estimate and compensate for the optimization imbalance problem caused by the imbalance of the directed communication topology; is the Laplace distribution noise used to implement the differential privacy protection function; α i (t) is the optimization step size to be selected.

[0041] S3.2: Obtain the out-degree neighbor distribution network subsystem and in-degree neighbor distribution network subsystem sets Determine the communication out-degree and the communication in-degree

[0042] S3.3: Read the

[0043] stored after the last update The specific generation rule is as follows:

[0044]

[0045] S3.5: Execute the differential privacy protection mechanism, that is, add the Laplace distribution noise generated in step 3.4 to the local optimization auxiliary variable p i (t) respectively to obtain the transmission variable to be transmitted in the communication network The specific formula is as follows:

[0046]

[0047] S3.6: Send the local optimization information calculated in step S3.5 to the nodes in the set of out-degree neighbor distribution network subsystems q i (t) read in step S3.3 and the communication out-degree obtained in step S3.2

[0048] S3.6: Receive the local optimization information sent by the nodes in the set of in-degree neighbor distribution network subsystems

[0049] S3.7: Perform the operation of incrementing the optimization time index, that is, t = t + 1

[0050] S3.8: Based on what is received in step S3.6 Update the auxiliary variable z i (t + 1), and the update basis is as follows:

[0051]

[0052] S3.9: Based on what is received in step S3.6 Update the auxiliary variable q i (t + 1), and the update basis is as follows:

[0053]

[0054] S3.10: Based on z calculated in step S3.8 i ​(t + 1) and q obtained in step S3.9 i (t + 1), perform an update operation on the local optimization variable μ i (t + 1), and the update basis is as follows:

[0055]

[0056] S3.11: Based on the local optimization variable μ obtained in step S3.10 i (t + 1), the local optimized power generation g of the distribution network subsystem can be calculated i (t + 1), and the calculation basis is as follows:

[0057]

[0058] In formula (14), is the optimal power generation corresponding to μ i (t + 1) obtained from the Lagrangian function (6) without considering the power generation capacity constraint of the distribution area subsystem. The min-max operation in formula (13) can ensure that the local optimized power generation g of the distribution network subsystem i (t + 1) must satisfy the power generation capacity constraint.

[0059] S3.12: Obtain the local power demand D of the distribution network subsystem at the current moment i and the uncertain new energy input r i (t + 1).

[0060] S3.13: Generate an updated optimization step size α i (t + 1) = 1 / (t + 1); α i (t + 1) can also be generated according to other rules, but it must satisfy the following constraints:

[0061]

[0062] S3.14: Based on the local optimized power generation g of the distribution network subsystem obtained in step S3.11 i (t + 1) and D obtained in S3.12 i and r i (t + 1) calculate the optimization gradient The calculation basis is as follows:

[0063]

[0064] S3.15: Calculate the local optimized information p to be transmitted at the next moment i (t + 1), according to the following:

[0065]

[0066] α in formula (17) i (t + 1) is the optimized step size of the distribution network subsystem at the current moment generated in step S.13; is the optimized gradient at the current moment obtained in step S3.14.

[0067] S3.16: Store the

[0068] S3.17: Determine the condition t < T. If it is satisfied, repeat step S3.3; if not, end all steps.

[0069] The collaborative solution method for the uncertainty economic dispatch of a distribution system considering privacy protection requirements described in the present invention, compared with the prior art by adopting the above technical solutions, has the following technical effects:

[0070] 1. By introducing differential privacy protection technology, key privacy parameters are protected from being deduced by malicious attackers;

[0071] 2. It can optimize and adjust the power generation of the distribution network subsystem based on the distributed uncertain new energy input, and overall reduce the operating cost of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 is a schematic diagram of the method flow proposed by the present invention;

[0073] Figure 2 is an iterative flowchart of the collaborative solution for the uncertainty economic dispatch of a distribution system considering privacy protection requirements;

[0074] Figure 3 is the specific structure diagram of the IEEE14 bus system;

[0075] Figure 4 is the communication network structure diagram between buses in the IEEE14 bus system;

[0076] Figure 5 is the consistency convergence result diagram of the local optimization variables of 5 power generation units in the IEEE14 bus system;

[0077] Figure 6 is the evolutionary calculation result diagram of the local power generation of 5 power generation units in the IEEE14 bus system;

[0078] Figure 7 is the true value of the key privacy parameter and the speculated value of the malicious attacker comparison result diagram;

[0079] Figure 8 is the true value of the key privacy parameter and the speculation value of malicious attackers The comparison result between them Specific implementation manners

[0080] The present invention will be further described below in conjunction with the accompanying drawings.

[0081] The present invention proposes a collaborative solution method for the uncertain economic dispatch of a distribution system considering privacy protection requirements, which is divided into the following Figure 1 3 steps as shown, and are specifically described as follows:[[]]END]]

[0082] S1: Establish an operation cost model of the distribution system considering the input of uncertain new energy. The cost function to be optimized is the sum of the power generation cost functions of all distribution area subsystems within a period of time, including the power generation constraints of the distribution area subsystems and the time-varying power generation and power consumption balance coupling constraints;

[0083] S2: Establish a Lagrangian dual optimization model corresponding to the original operation optimization problem of the distribution system. By decomposing the global time-varying power consumption demand and the total input of uncertain new energy into local power consumption demands and regional inputs of uncertain new energy, the power generation and power consumption balance coupling constraints are solved;

[0084] S3: For the obtained Lagrangian dual optimization model, a collaborative optimization solution method for the uncertain economic dispatch of the distribution system based on differential privacy technology is constructed. On the premise of ensuring that the key privacy parameters cannot be speculated, the local power generation optimization solutions of all distribution area subsystems are given.

[0085] The present invention first constructs an operation cost model of the distribution system considering the input of uncertain new energy, as specifically shown in formula (1):

[0086]

[0087] In formula (1), t is the optimization period index, (t ∈ [1, T]); T is the total number of artificially set optimization periods; n is the number of distribution area subsystems; g i (t) is the power generation to be optimized of the i-th distribution area subsystem in the optimization period t; r i (t) is the input of uncertain new energy of the i-th distribution area subsystem in the optimization period t, which is a time-varying constant that cannot be optimized; is the power generation cost of the i-th distribution area subsystem in the optimization period t; is the cost of using uncertain new energy of the i-th distribution area subsystem in the optimization period t;

[0088] In formula (1), the power generation cost of the i-th distribution area subsystem in the optimization period t The specific expression is as shown in formula (2):

[0089]

[0090] In formula (2), and are the quadratic term and the linear term coefficient of the power generation cost of the \(i\)-th distribution area subsystem;

[0091] In formula (1), the cost of using uncertain new energy of the \(i\)-th distribution area subsystem during the optimization period \(t\) Specifically, it is shown in formula (3) as follows:

[0092]

[0093] In formula (3), \(k\) r is the coefficient of the cost of using uncertain new energy;

[0094] The constraints in the operation cost model of the distribution system considering the input of uncertain new energy are as follows:

[0095] Power generation capacity constraint of the \(i\)-th distribution area subsystem:

[0096]

[0097] In formula (4), is the minimum power generation of the \(i\)-th distribution area subsystem; is the maximum power generation of the \(i\)-th distribution area subsystem;

[0098] Global power generation and power consumption balance coupling constraint of the distribution system:

[0099]

[0100] In formula (5), \(D\) is the global power consumption demand of the distribution system during the optimization period \(t\);

[0101] The present invention constructs a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem, and solves the power generation and power consumption balance coupling constraint by decomposing the global time-varying power consumption demand and the total input of uncertain new energy into local power consumption demand and regional input of uncertain new energy, specifically as follows:

[0102] During the optimization period \(t\), the following Lagrangian function is defined:

[0103]

[0104] In formula (6), \(\mu\) i (t) is the Lagrangian dual variable related to the global power generation and power consumption balance coupling constraint of the distribution system; \(D\) i is the local power consumption demand of the \(i\)-th distribution area subsystem, satisfying Based on formula (6), during the optimization period t, a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem can be established:

[0105]

[0106] For the obtained Lagrangian dual optimization model, the present invention constructs a collaborative optimization solution method for the uncertainty economic dispatch of the distribution system based on differential privacy technology, and gives the local generation optimization solutions of all distribution area subsystems, as follows:

[0107] First, determine the out-degree according to the fixed directed network topology composed of each sub-region in the distribution system and the communication weight matrix A = [a ij , and the specific formula is:

[0108]

[0109] In formula (8), is the set composed of all distribution area subsystems connected to the j-th distribution area subsystem as the signal output party;

[0110] Such as Figure 2 shown, taking the i-th distribution area subsystem as an example, the specific steps of the collaborative optimization solution for the uncertainty economic dispatch of the distribution system based on differential privacy technology are as follows:

[0111] S3.1: Set the optimization time index to 0, that is, t = 0; initialize the subsystem local optimization variables and auxiliary variables:

[0112]

[0113] Among them, p i (t), q i (t), z i (t) are auxiliary variables; q i (t) is used to estimate and compensate for the optimization imbalance problem caused by the imbalance of the directed communication topology; is the Laplace distribution noise used to implement the differential privacy protection function; α i (t) is the optimization step size to be selected.

[0114] S3.2: Obtain the out-degree neighbor distribution network subsystem and in-degree neighbor distribution network subsystem sets Determine the communication out-degree and the communication in-degree

[0115] S3.3: Read the

[0116] S3.4: Generate the Laplace noise required for this round of update The specific generation rules are as follows:

[0117]

[0118] S3.5: Execute the differential privacy protection mechanism, that is, add the Laplace distribution noise generated in step 3.4 to the local optimization auxiliary variable p i (t) respectively to obtain the transmission variable to be transmitted in the communication network The specific formula is as follows:

[0119]

[0120] S3.6: Send the local optimization information calculated in step S3.5 to the nodes in the set of in-degree neighbor distribution network subsystems q read in step S3.3 i (t) and the communication out-degree obtained in step S3.2

[0121] S3.6: Receive the local optimization information sent by the nodes in the set of in-degree neighbor distribution network subsystems

[0122] S3.7: Perform the operation of incrementing the optimization time index, that is, t = t + 1.

[0123] S3.8: Based on what is received in step S3.6 Update the auxiliary variable z i (t + 1), and the update basis is as follows:

[0124]

[0125] S3.9: Based on what is received in step S3.6 Update the auxiliary variable q i (t + 1), and the update basis is as follows:

[0126]

[0127] S3.10: Based on z calculated in step S3.8 i (t + 1) and q calculated in step S3.9 i (t + 1), update the local optimization variable μ i (t + 1), and the update basis is as follows:

[0128] μ i (t + 1) = zi (t + 1) / q i (t + 1)(13)

[0129] S3.11: Based on the locally optimized variable μ obtained in step S3.10 i (t + 1), the locally optimized power generation g of the distribution network subsystem can be calculated i (t + 1), and the calculation basis is as follows:

[0130]

[0131] In formula (14) is the optimal power generation corresponding to μ i (t + 1) obtained from the Lagrangian function (6) without considering the power generation capacity constraint of the distribution area subsystem. The min-max operation in formula (13) can ensure that the locally optimized power generation g i (t + 1) of the distribution network subsystem will surely satisfy the power generation capacity constraint.

[0132] S3.12: Obtain the local power demand D of the distribution network subsystem at the current moment i and the uncertain new energy input r i (t + 1).

[0133] S3.13: Generate the updated optimization step size α i (t + 1) = 1 / (t + 1); α i (t + 1) can also be generated according to other rules, but it must satisfy the following constraints:

[0134]

[0135] S3.14: Based on the locally optimized power generation g of the distribution network subsystem obtained in step S3.11 i (t + 1) and D obtained in S3.12 i and r i (t + 1) to calculate the optimization gradient The calculation basis is as follows:

[0136]

[0137] S3.15: Calculate the locally optimized information p to be transmitted at the next moment i (t + 1), based on the following:

[0138]

[0139] The α i (t + 1) in formula (17) is the optimization step size of the distribution network subsystem at the current moment generated in step S.13; is the optimized gradient at the current moment obtained in step S3.14.

[0140] S3.16: Store what is obtained from this update

[0141] S3.17: Determine the condition t < T. If it is satisfied, repeat step S3.3. If not, end all steps.

[0142] The following gives the privacy protection performance of the proposed algorithm for the key privacy parameter and k r :

[0143] According to the capabilities of malicious attackers, the attack scenarios can be divided into two scenarios:

[0144] C1: The attacker can eavesdrop on all transmitted information but does not know any information related to the local optimization rules.

[0145] C2: The attacker can not only eavesdrop on all transmitted information but also know any information related to the local optimization rules except for the randomly generated Laplace noise other than.

[0146] For scenario C1, although the attacker can obtain all the eavesdropped transmitted information and but does not know any formula information about variable updates, so it is impossible to deduce the optimization gradient information directly related to the key privacy parameter and Therefore, it is impossible to speculate on the key privacy parameter and k and k r for speculation.

[0147] For scenario C2, the attacker can not only obtain all the eavesdropped transmitted information and q i (t), but also know all local update rules, including formulas (11), (12), (13), (14) and (16). The only thing not known is the Laplace distribution noise introduced when implementing the differential privacy protection mechanism in step S3.5

[0148] The attacker can construct a system of equations for speculating on the key privacy parameter and k r according to formula (14), and the specific form is as follows:

[0149]

[0150] In formula (17), and kr is a parameter to be inferred, which is known to malicious attackers. The attacker can calculate g i (t) through formulas (11) and (16), and calculate μ i (t) through formulas (11), (12) and (13). Therefore, the malicious attacker can perform the inference process as follows:

[0151]

[0152] In formula (18), k 1 , k 2 , k 3 are three moments that do not perform the max or min operation. It should be noted that the update formula of the attacker for z i (t + 1) is actually

[0153]

[0154] The difference between formula (19) and formula (11) lies in the Laplace distribution noise From this, it can be seen that the attacker cannot obtain the accurate three equations for inferring k 1 , k 2 , k 3 and cannot achieve the theft of key privacy information.

[0155] The following gives an IEEE14 bus simulation example:

[0156] The IEEE14 bus contains 5 power generation units (buses 1, 2, 3, 6, 8) and 9 load units (buses 4, 5, 7, 9, 10, 11, 12, 13, 14), and the specific structure is as Figure 3 shown. The parameters and the upper and lower bounds of power generation are selected as shown in Table 1.

[0157] Table 1

[0158]

[0159] The g i (t) of the load units are all set to 0. The D i , and of all buses are selected as shown in Table 2, where is the upper bound of the random selection range of r i (t). The communication network structure between buses is as Figure 4 shown.

[0160] Table 2

[0161]

[0162]

[0163] Figure 5 Shows the consistency convergence result graph of the local optimization variables of 5 power generation units. Figure 6 Shows the evolutionary calculation result graph of the local power generation of 5 power generation units. Figure 7 Shows the true values of the key privacy parameters and the speculated values of malicious attackers The comparison results between them. Figure 8 Shows the true values of the key privacy parameters and the speculated values of malicious attackers The comparison results between them.

[0164] The present invention considers the privacy protection requirements in the distribution system, constructs a collaborative optimization solution method for the uncertainty economic dispatch of the distribution system based on differential privacy technology, and gives the local power generation optimization solutions of all distribution area subsystems on the premise of ensuring that the key privacy parameters cannot be speculated, reducing the overall operation cost of the distribution network.

[0165] The above are only specific embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. An uncertain online scheduling method for distribution system considering privacy protection requirements, characterized in that: The following steps are involved: S1: Establish a distribution system operation cost model considering the input of uncertain new energy. The cost function to be optimized is the sum of the power generation cost functions of all distribution area subsystems within a period of time, including the power generation constraints of the distribution area subsystems and the time-varying power generation and power consumption balance coupling constraints; S2: Establish a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem, and solve the power generation and power consumption balance coupling constraint by decomposing the global time-varying power demand and the total input of uncertain new energy into local power demand and regional input of uncertain new energy; S3: Based on the obtained Lagrangian dual optimization model, a collaborative optimization solution for uncertain economic dispatch of distribution systems based on differential privacy technology is constructed. Under the premise of ensuring that key privacy parameters are not speculated, the local power generation optimization solution of all distribution area subsystems is given.

2. According to the method of claim 1, the method is characterized in that: The distribution system operation cost model considering the uncertain new energy input described in step S1 is specifically shown in formula (1): In formula (1), t is the optimization period index, (t∈[1,T]); T is the total number of optimization periods set manually; n is the number of subsystems in the distribution area; g i (t) is the power generation to be optimized of the ith distribution area subsystem in the optimization period t; r i (t) is the uncertain new energy input of the ith distribution area subsystem in the optimization period t, which is a time-varying constant that cannot be optimized; is the power generation cost of the ith distribution area subsystem in the optimization period t; is the uncertain new energy use cost of the ith distribution area subsystem in the optimization period t; In formula (1), the power generation cost of the ith distribution area subsystem in the optimization period t is The specific expression is shown in formula (2): In formula (2), and are the coefficients of the quadratic and linear terms of the power generation cost of the ith distribution area subsystem; In formula (1), the uncertain new energy use cost of the ith distribution area subsystem in the optimization period t is The specific formula is as shown in formula (3): In formula (3), k r is the cost coefficient for the use of uncertain new energy; The constraints in the distribution system operation cost model considering uncertain new energy input are as follows: The generation capacity constraint of the ith distribution area subsystem is: In formula (4), is the minimum power generation of the ith distribution area subsystem; is the maximum power generation of the ith distribution area subsystem; The global power generation and consumption balance coupling constraints of the distribution system: In formula (5), D is the global power demand of the distribution system in the optimization period t.

3. According to claim 1, a distribution system operation cost model considering uncertain new energy input is characterized in that: The step of establishing the Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem in step S2 solves the power generation and power balance coupling constraint by decomposing the global time-varying power demand and the total input of uncertain new energy into local power demand and regional input of uncertain new energy, as follows: At the optimization time t, the following Lagrangian function is defined: In formula (6), μ i (t) is the Lagrangian dual variable related to the global generation and consumption balance coupling constraint of the distribution system; D i is the local power demand of the ith distribution area subsystem, satisfying Based on formula (6), in the optimization period t, a Lagrangian dual optimization model corresponding to the original distribution system operation optimization problem can be established:

4. The method for coordinated online optimization of power distribution system operation considering carbon emission costs according to claim 1, characterized in that: According to the Lagrangian dual optimization model obtained in step S3, a collaborative optimization solution method for uncertain economic dispatch of distribution system based on differential privacy technology is constructed to provide the local power generation optimization solution for all distribution area subsystems, as follows: First, the out-degree is determined according to the fixed directed network topology composed of each sub-area in the distribution system. and the communication weight matrix A = [a ij ], the specific formula is: In formula (8), is the set of all distribution area subsystems connected to the jth distribution area subsystem as the signal output party; Then, taking the i-th distribution area subsystem as an example, the steps for solving the collaborative optimization of uncertain economic dispatch of the distribution system based on differential privacy technology are as follows: S3.1: Set the optimization time index to 0, i.e. t = 0; initialize the subsystem local optimization variables and auxiliary variables: p i (0)=1,μ i (0)=1,q i (0)=1,z i (0)=0, a i (0)=1 where p i (t), q i (t), z i (t) is an auxiliary variable; q i (t) is used to estimate and compensate for the optimization imbalance caused by the imbalance of the directed communication topology; is the Laplace distributed noise used to implement differential privacy protection; α i (t) is the optimization step size to be selected; S3.2: Get the set of out-degree neighbor distribution network subsystems and in-degree neighbor distribution network subsystems Determine the communication out-degree and communication degree S3.3: Read the p stored after the last update i (t), q i (t), α i (t);; S3.4: Generate the Laplace noise required for this round of update The specific generation rules are as follows: S3.5: Execute the differential privacy protection mechanism, that is, to optimize the auxiliary variable p locally to interact with the neighboring electronic region i (t) are added with the Laplace distribution noise generated in step 3.4 Get the transmission variable to be transmitted in the communication network The specific formula is as follows: S3.6: Send the local optimization information calculated in step S3.5 to the nodes in the set of in-degree neighbor distribution network subsystems The q read in step S3.3 i (t) and the communication out-degree obtained in step S3.2 S3.6: Receive local optimization information sent from nodes in the in-degree neighbor distribution network subsystem set S3.7: Execute the optimization time index increment operation, that is, t=t+1; S3.8: Based on the received For the auxiliary variable z i (t+1) Perform update operation, the update basis is as follows: S3.9: Based on the received For the auxiliary variable q i (t+1) Perform update operation, the update basis is as follows: S3.10: Based on the z calculated in step S3.8 i (t+1) and q calculated in step S3.9 i (t+1), for the local optimization variable μ i (t+1) Perform update operation, the update basis is as follows: μ i (t+1)=z i (t+1) / q i (t+1) (13) S3.11: Based on the local optimization variable μ obtained in step S3.10 i (t+1), the local optimal power generation g of the distribution network subsystem can be calculated i (t+1), calculated as follows: In formula (14) is the Lagrangian function (6) obtained without considering the generation capacity constraint of the distribution area subsystem and μ i The operation of taking the smaller or larger value in formula (13) can ensure the local optimal power generation g of the distribution network subsystem. i (t+1) must satisfy the power generation capacity constraint; S3.12: Obtain the local power demand D of the distribution network subsystem at the current moment i With uncertainty new energy input i (t+1); S3.13: Generate updated optimization step size α i (t+1)=1 / (t+1); α i (t+1) can also be generated according to other rules, but must meet the following constraints: S3.14: Based on the local optimized power generation g of the distribution network subsystem obtained in step S3.11 i (t+1) and D obtained in S3.12 i With r i (t+1) Calculate the optimization gradient The calculation is based on the following: S3.15: Calculate the local optimization information p to be transmitted at the next moment i (t+1), based on the following: α in formula (17) i (t+1) is the optimization step size of the distribution network subsystem at the current moment generated in step S.13; is the current optimization gradient obtained in step S3.14; S3.16: Store the updated p i (t+1),q i (t+1), α i (t+1). S3.17: Determine the condition t<T. If it is satisfied, repeat step S3.

3. If not, end all steps.

Images