Active distribution network control method and equipment considering wind and solar randomness and dynamic pricing

Through conditional generation adversarial network and two-layer optimization scheduling model, combined with the comprehensive dynamic pricing mechanism of load renewable energy matching, the problems of wind and light randomness and renewable energy uncertainty in the existing technology are solved, and the microgrid and activeness are realized Coordination of the distribution network's interest distribution and optimization of the distribution network voltage distribution.

CN118659471BActive Publication Date: 2024-12-03STATE GRID TIANJIN ELECTRIC POWER CO CHENGXI POWER SUPPLY BRANCH +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411144138.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-12-03
Estimated Expiration
2044-08-20

AI Technical Summary

Technical Problem

When considering the randomness of the scenery and dynamic pricing, the existing technology failed to fully stimulate the microgrid's ability to optimize the distribution network trend, and rarely consider the uncertainty of renewable energy, which affects the operating reliability and benefits of the distribution network. distribute.

Method used

The conditional generation adversarial network generation scenario is adopted, and the comprehensive dynamic pricing mechanism of load renewable energy matching through a two-layer optimization scheduling model is combined with the comprehensive dynamic pricing mechanism of load renewable energy matching, coordinate the power interaction between the microgrid and the active distribution network, and scene reduction is carried out through Kmeans++ to optimize the distribution network voltage distribution.

Benefits of technology

The benefit distribution coordination between the microgrid and the active distribution network is achieved, the voltage distribution and operating reliability of the distribution network is improved, and the joint operation cost is reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118659471B_ABST
    Figure CN118659471B_ABST
Patent Text Reader

Abstract

The present invention discloses an active distribution network control method and device considering the randomness of wind and solar power and dynamic pricing, including: first, in view of the uncertainty of wind and solar power output, a conditional generative adversarial network is used to generate scenarios, and then the scenarios are reduced based on Kmeans++. Secondly, based on the distribution of upper and lower energy and load conditions, a comprehensive dynamic pricing mechanism considering the matching degree of load and renewable energy is proposed to guide the reasonable interaction between the power of microgrids and active distribution networks. Finally, verification and analysis are carried out on the improved IEEE33 node system. The results show that compared with time-of-use electricity prices, the proposed dynamic pricing mechanism and scheduling strategy can effectively coordinate the distribution of benefits between the distribution network and multiple microgrids, and improve the voltage distribution of the distribution network.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power grid control, and in particular to an active distribution network control method and device that takes into account the randomness of wind and solar power and dynamic pricing. Background Art

[0002] Microgrids are composed of multiple distributed generation equipment (DG), energy storage equipment (EES) and local loads, which can provide power support for active distribution networks and improve the overall power quality of the power grid. In the operation of active distribution networks containing multiple microgrids, the generation and consumption of power by microgrid operators are not subject to the unified dispatch of the distribution network, forming different stakeholders. In response to the problem of interest distribution among multiple stakeholders, there are currently studies on the optimization dispatching means that focus on the economic dispatch of multiple MGs connected to the ADN system. However, in reality, most microgrids are distributed at different nodes of the distribution network. If the impact of microgrids on ADN power flow is not considered, it will be difficult to ensure that the distribution network meets the power flow constraints in actual operation, threatening the reliability and safety of the coordinated operation between the ADN system and multiple microgrids. There are also studies on the interactive two-stage dynamic pricing mechanism between the distribution network and multiple microgrids. At the upper level of the pricing mechanism, the distribution energy market operator guarantees the quality of operation by trading energy with the coupled microgrid, and maximizes the profit of each MG at the lower level. However, the transaction process of the above dynamic pricing mechanism is relatively cumbersome, does not fully stimulate the optimization ability of the microgrid on the distribution network power flow, and rarely considers the uncertainty factors of renewable energy. Summary of the invention

[0003] The purpose of the present invention is to provide an active distribution network control method and device taking into account the randomness of wind and solar power and dynamic pricing, so as to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] Active distribution network control methods considering wind and solar randomness and dynamic pricing include:

[0006] Step 1: In view of the uncertainty of wind and solar power output, a conditional generative adversarial network is used to generate scenarios, and then the scenarios are reduced;

[0007] Step 2: Through the two-layer optimization dispatching model with the optimal joint objective function of multiple microgrids and active distribution networks, a comprehensive dynamic pricing mechanism considering the matching degree of load and renewable energy is established to guide the power interaction between microgrids and active distribution networks on the basis of the upper and lower energy and load distribution conditions, effectively coordinate the distribution of benefits between distribution networks and multiple microgrids, and improve the voltage distribution of distribution networks;

[0008] After step 2, you can also:

[0009] Step 3: Perform verification analysis on the improved IEEE33 node system.

[0010] Furthermore, step 1 specifically includes: generating scenes based on CWGAN-GPCT, generating networks based on GAN using CGAN, and retaining the generator of GAN G and the discriminator D The game structure is consistent with The distributed random noise z and conditional label y are input to the generator G ;

[0011] Generator G The goal is to learn the parameters , generate distribution scenarios that match given labels , the generated distribution is recorded as ;

[0012] Discriminator D The input is historical data x, random noise z and conditional label y, the discriminator D Target differentiation and , and determine whether the random noise z meets the conditional label y. In CWGAN-GP based on Wasserstein distance and adding GP terms, the objective function And the loss function They are described as follows:

[0013] (1)

[0014] (2)

[0015] (3)

[0016] In the formula, is the Wasserstein distance, It is a discriminator D The expected upper bound, is a 1-Lipschitz constraint, and the observed data points , It's expectation. is the expectation of generating data, is inputting x to the discriminator under label y, is the differential, yes Enter the discriminator.

[0017] Furthermore, the consistency term is introduced in CWGAN-GP CT , by using two , Close to the observed data point The perturbed data points improve the continuity condition, which can be expressed as follows:

[0018] (4)

[0019] In the formula, It's the border. d is the two-norm input space, is the perturbed data point Input Discriminator D , is the perturbed data point Input Discriminator D , is the perturbed data point Input Discriminator D The second to last layer, is the perturbed data point Input Discriminator D The second to last floor;

[0020] Discriminator D The loss function It is expressed as:

[0021] (5)

[0022] In the formula, the proportional parameter , scale parameter .

[0023] Furthermore, step 2 specifically includes: coordinating the power interaction between the upper model of the active distribution network and the lower model of the multiple microgrids by jointly optimizing the objective function.

[0024] Furthermore, the equivalent load is used to describe the load demand of the lower layer in each period. The lower equivalent load includes the electric and thermal load of the microgrid and the electric and thermal energy storage power. The lower equivalent load is expressed as:

[0025] (6)

[0026] Is Microgrid 1 in t Time period electrical equivalent load, Is Microgrid 2 in t Time period electrical equivalent load, Is Microgrid 3 in t Time period electrical equivalent load, Is Microgrid 1 in t Thermal equivalent load of the time period, Is Microgrid 1 in t Initial electrical load of the period, Is Microgrid 1 in t The power load can be shifted in different time periods. Is Microgrid 1 in t Time period energy storage output, Is Microgrid 1 in t The power load can be interrupted during the period. Is Microgrid 1 in t Electric boiler power consumption during the period, Is Microgrid 2 in t Initial electrical load of the period, Is Microgrid 2 in t Time period energy storage output, Is Microgrid 3 in t Initial electrical load of the period, Is Microgrid 3 in t The power load can be shifted in different time periods. Is Microgrid 3 in t The power load can be interrupted during the period. Is Microgrid 1 in t Initial heat load of the period, Is Microgrid 1 in t Heat load can be shifted by time period. Is Microgrid 1 in t The heat load can be interrupted during the period. Is Microgrid 1 in t Output of period thermal energy storage.

[0027] Furthermore, the heat load required by the microgrid in the lower layer is generated by the electric boiler, which converts the heat load into the electrical load demand. Electrical equivalent load (power) during the iteration It is expressed as:

[0028] (7)

[0029] The dynamic pricing mechanism considering load distribution is expressed as:

[0030] (8)

[0031] (9)

[0032] Where: is the energy conversion efficiency of electric boiler, is the heat-to-electricity efficiency of the microgrid 1 electric boiler, Time-of-use electricity price; is the electricity price coefficient, For the The dynamic electricity price of the iteration, is the wind-solar combined output value of the active distribution network (or distributed power source power), It is the load / wind / solar ratio coefficient (the quotient of load power and distributed generation power).

[0033] Furthermore, the objective function of the upper model of the active distribution network includes the network loss cost and the operating cost of the renewable energy power generation unit, minimizing the weighted average of the cumulative output of each scenario as shown in formula (10):

[0034] (10)

[0035] Where T is the scheduling period, For the generated active distribution network scenario, is the probability of active distribution network scenario, is the upper objective function, , They are respectively the wind power operation and maintenance cost coefficient and the photovoltaic power generation operation and maintenance cost coefficient. For branch ij exist t The current during the period, For branch ij The resistance, , for t During the period i The actual active power output of the node wind and photovoltaic power generation units, is the set of all nodes in the distribution network, is the set of all branches of the distribution network, is the loss coefficient.

[0036] Furthermore, the branch flow constraint is adopted for the branch, specifically, formula (11):

[0037] (11)

[0038] In the formula, For Node i exist t The voltage during the time period, , For branch ij Node i For the first end t Active power and reactive power during the time period, , For Node j exist t The injected power during the time period, For branch ij The square of the current allowed to flow.

[0039] Furthermore, the wind turbine and photovoltaic constraints are expressed as formulas (13) and (14):

[0040] (13)

[0041] (14)

[0042] In the formula, , They are i The active power limit of the node wind and photovoltaic power generation units.

[0043] Furthermore, the SVC constraint in the active distribution network is expressed as formula (15):

[0044] (15)

[0045] In the formula, for t SVC in the active distribution network at the node during the period i The actual reactive power output, i The node's SVC is t The reactive power compensated during the period, , They are the upper and lower limits of SVC reactive power compensation respectively.

[0046] Furthermore, the node power balance constraints are given by equations (16) and (17):

[0047] (16)

[0048] (17)

[0049] Where: , They are i Active and reactive power consumed by node loads, For active distribution networks i Active power consumed by nodes connected to the microgrid.

[0050] Furthermore, the objective function of the multi-microgrid lower-level model includes the power interaction cost between the microgrid and the distribution network, the natural gas purchase cost, and the demand response compensation fee, which is expressed by formula (18):

[0051] (18)

[0052] In the formula, Scenario generated for multi-microgrid, The probability of generating scenarios for multi-microgrids, is the lower layer objective function, is the gas purchase price for the microgrid, is the collection of multiple microgrids at the lower layer, It is a microgrid i exist t The compensation price for power load interruption during the period, It is a microgridi exist t The compensation price for heat load interruption during the period, It is a microgrid i exist t The power load can be interrupted during the period. It is a microgrid i exist t The heat load can be interrupted during the period. It is a dynamic electricity price. is the purchased power, is the power consumption of the electric boiler, It is a microgrid i exist t The power generated by the gas turbine during the period.

[0053] Furthermore, the gas turbine constraint is expressed by formula (19):

[0054] (19)

[0055] In the formula, Microgrid i exist t Gas turbine power generation during the period, It is a microgrid i exist t The natural gas power input to the gas turbine during the period, It is a microgrid i Minimum power generation of gas turbine, It is a microgrid i The maximum power of gas turbine generator, It is a microgrid i Gas turbine energy conversion efficiency.

[0056] Furthermore, the electric boiler constraint is expressed by formula (20):

[0057] (20)

[0058] In the formula, It is a microgrid i exist t Thermal power of electric boiler during the period, It is a microgrid i exist t Electric power of electric boiler during the period, It is a microgrid i The electricity-to-heat efficiency of electric boilers, It is a microgrid i Minimum electric power of electric boiler, It is a microgrid i Maximum electric power of electric boiler.

[0059] Furthermore, the microgrid also includes battery energy storage, and the battery energy storage constraint is expressed by formula (21):

[0060] (twenty one)

[0061] In the formula, CN For energy storage collection, for CN equipment t The power of the moment, for CN equipment t The capacity of the moment, , They are CN equipment t The charging and discharging power at each moment, , They are CN The charging and discharging efficiency of the equipment, , They are CN The maximum and minimum values ​​of the equipment capacity, is a binary variable for charge and discharge, , They are CN The maximum value of the device's charge and discharge power. It's the first moment CN The capacity of the device, It's the 24th moment CN The capacity of the device, It is t- 1 moment CN The capacity of the device.

[0062] Furthermore, the objective function of the multi-microgrid lower-level model also includes a comprehensive demand response constraint, which is expressed by formula (22):

[0063] (twenty two)

[0064] In the formula, en For energy categories, E , H They are electricity demand response and heat demand response, is the total power of demand response load, , They are interruptible demand response and time-shiftable demand response. is the initial load, is the maximum power that the load can use to respond to demand, is the interruptible load ratio coefficient, is the maximum interruptible load value, is the time-shiftable load proportionality factor, yes t Time period energy demand response load, Meaning is t Time period demand response load.

[0065] Furthermore, the microgrid power balance constraint is expressed by formula (23):

[0066] (twenty three)

[0067] It is a microgrid i exist t The actual power of photovoltaic power generation during the period, It is a microgrid i exist t Time period electrical equivalent load, It is a microgrid i exist t Thermal power of electric boiler during the period.

[0068] Furthermore, the joint optimization objective function is expressed as :

[0069] (twenty four)

[0070] In the formula, , are the weight coefficients of the upper and lower layers respectively, is the upper objective function, is the lower objective function.

[0071] To achieve the above object, the present invention also provides the following technical solutions:

[0072] Active distribution network control equipment considering the randomness of wind and solar power and dynamic pricing, including generation module and output module,

[0073] The generation module is used to generate scenarios based on the uncertainty of wind and solar power output using a conditional generative adversarial network, and then perform scenario reduction.

[0074] The output module is used to guide the power interaction between microgrids and active distribution networks based on the distribution of upper and lower energy and loads through a two-layer optimization scheduling model with the optimal joint objective function of multiple microgrids and active distribution networks. It also considers the comprehensive dynamic pricing mechanism of load-renewable energy matching, effectively coordinates the distribution of benefits between distribution networks and multiple microgrids, and improves the voltage distribution of distribution networks.

[0075] Verification module, used to perform verification analysis on the improved IEEE33 node system.

[0076] To achieve the above object, the present invention also provides the following technical solutions:

[0077] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0078] To achieve the above object, the present invention also provides the following technical solutions:

[0079] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of any one of the methods described above.

[0080] Compared with the prior art, the present invention has the following beneficial effects:

[0081] The present invention is a dual optimization scheduling strategy that takes into account the uncertainty of wind and solar power and a dynamic pricing mechanism. First, a conditional generative adversarial network is used to generate a scene set, and Kmeans++ is used to reduce the scenes. Secondly, a two-layer optimization model of the upper-layer ADN and the lower-layer multiple MGs is constructed, and multiple factors such as the upper and lower power sources and load distribution are fully considered. Based on the load-renewable energy matching degree, a comprehensive dynamic pricing mechanism is established to guide the reasonable interaction of microgrids and active distribution network power. Finally, through example analysis, it is verified that the dynamic pricing mechanism and scheduling strategy proposed in the present invention can reasonably guide the joint optimization of the upper and lower layers, coordinate the distribution of interests of multiple stakeholders, and effectively improve the voltage distribution of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 This is a schematic diagram of the double-layer optimization scheduling model of the present invention.

[0083] Figure 2 The present invention provides a flow chart for solving the model.

[0084] Figure 3 This is the improved IEEE33 node topology diagram of the present invention.

[0085] Figure 4 This is a load data diagram of each microgrid in the lower layer of the present invention.

[0086] Figure 5 This is a day-ahead scenario diagram of new energy in the distribution network of the present invention.

[0087] Figure 6 This is a scene diagram of the microgrid new energy day before the present invention.

[0088] Figure 7 It is the discriminator gradient change diagram of the present invention.

[0089] Figure 8 It is the discriminator Wasserstein distance map of the present invention.

[0090] Fig. 9It is an iterative process diagram of the double-layer model of the present invention.

[0091] Fig.10 This is the electricity price map provided by the upper distribution network of the present invention to the three lower microgrids.

[0092] Fig.11 It is the extreme value diagram of node voltage in the scheduling period of strategy 1 and strategy 4 of the present invention.

[0093] Fig.12 It is the optimal node voltage distribution diagram at different times in the distribution network under the double-layer scheduling of the present invention.

[0094] Fig.13 This is a system model diagram of the method of the present invention.

[0095] Fig.14 It is a diagram of the internal structure of the computer device of the present invention. DETAILED DESCRIPTION

[0096] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0097] See also Figures 1 to 14 , the present invention provides a technical solution:

[0098] Aiming at the day-ahead dispatching problem of active distribution network with multiple microgrids under multi-subject interest scenario, this paper proposes a two-layer optimal dispatching strategy considering the uncertainty of wind and solar power and dynamic pricing mechanism. Firstly, considering the uncertainty of wind and solar power output, conditional generative adversarial network is used to generate scenarios, and then Kmeans++ is used to reduce scenarios. Secondly, based on the distribution of upper and lower energy and load, in order to guide the reasonable interaction of microgrid and active distribution network power, a comprehensive dynamic pricing mechanism considering the matching degree of load and renewable energy is proposed. Finally, the improved IEEE33 node system is used for verification and analysis. The results show that compared with time-of-use electricity price, the proposed dynamic pricing mechanism and dispatching strategy can effectively coordinate the distribution of benefits between distribution network and multiple microgrids, and improve the voltage distribution of distribution network.

[0099] This paper proposes a dual optimization scheduling strategy that considers the uncertainty of wind and solar power and a dynamic pricing mechanism. First, the conditional generative adversarial network is used to generate a scene set, and Kmeans++ is used to reduce the scenes. Secondly, a two-layer optimization model of the upper-layer ADN and the lower-layer multi-MG is constructed, and multiple factors such as the upper and lower power sources and load distribution are fully considered. Based on the load-renewable energy matching degree, a comprehensive dynamic pricing mechanism is established to guide the reasonable interaction of microgrids and active distribution network power. Finally, through example analysis, it is verified that the dynamic pricing mechanism and scheduling strategy proposed in this paper can reasonably guide the joint optimization of the upper and lower layers, coordinate the distribution of interests of multiple stakeholders, and effectively improve the voltage distribution of the distribution network.

[0100] Scene generation method based on CWGAN-GPCT

[0101] CGAN is a generative network based on GAN, which retains the game structure of the generator G and the discriminator D of GAN. The distributed random noise z and conditional label y are input to the generator G. The goal of the generator is to learn parameters , generate distribution scenarios that match given labels ( Distribution is the distribution law of the actual scene), and the generated distribution is recorded as The input of the discriminator D is historical data x, z and y, which is a two-class model. The discriminator aims to distinguish and , and determine whether z meets y. In the CGAN (CWGAN-GP) based on Wasserstein distance (this is a technical term, also called optimal transmission distance, earth moving distance or bulldozer distance) and adding GP terms, the objective function and loss function are described as follows:

[0102] (1)

[0103] (2)

[0104] (3)

[0105] In the formula, is the Wasserstein distance, It is a discriminator D The expected upper bound, is a 1-Lipschitz constraint, and the observed data points , It's expectation. is the expectation of generating data, is inputting x to the discriminator under label y, is the differential, yes Input Discriminator D , is the expectation of the original data, is the expectation of the generated noise, t Time period, U It is a 0-1 parameter set.

[0106] However, the GP penalty term of the CWGAN-GP model only enforces 1-Lipschitz continuity at specific sampling points, resulting in important parts of the support domain not being checked at all. This paper introduces a consistency term (CT) in CWGAN-GP to consider strengthening the 1-Lipschitz continuity of the manifold region around the sampling points. The model is called CWGAN-GPCT. Specifically, by using two , Close to the observed data point The perturbation data points further improve the continuity condition and are expressed as follows:

[0107] (4)

[0108] In the formula, It's the border. d is the two-norm input space, is the perturbed data point Input Discriminator D , is the perturbed data point Input Discriminator D , is the perturbed data point Input Discriminator D The second to last layer, is the perturbed data point Input Discriminator D The second to last floor;

[0109] The loss function of the discriminator is expressed as:

[0110] (5)

[0111] In the formula, , is the scale parameter.

[0112] This paper adopts the CWGAN-GPCT method to generate large-scale uncertain output scenarios. In order to select typical scenarios that meet the statistical characteristics of uncertain variables and have meaningful configurations, it is necessary to reduce the generated large-scale basic scenario set. To this end, the Kmeans++ clustering algorithm is applied, and the steps are as follows:

[0113] 1) From the input set Randomly select an object z i As the initial cluster center k i ;

[0114]

[0115] 2) Calculation The indicator, whose value determines the distance between the generated cluster and the historical data, the larger the indicator value, the smaller the distance between classes and the greater the dispersion of objects within the class, that is, the worse the clustering effect, so it is used as a reverse indicator, the smaller the indicator value, the better the clustering effect:

[0116]

[0117] In the formula, To calculate the (total) length of input data; is the input data matrix; represents the distance of the cluster;

[0118] 3) Calculate the value of each data Indicators, select The indicator with the highest probability is selected as the next cluster center;

[0119] 4) Repeat (2) and (3) until the cluster centers meet the requirements (i.e., meet the specified number, determined by the decision maker).

[0120] z i , k i where i=1,2,3…n, N is the total number of data, Z j is the input data length, j is the index.

[0121] A two-level optimization scheduling model considering multi-agent interest distribution

[0122] In order to coordinate the distribution of benefits between the distribution network and multi-microgrids, a two-layer optimization dispatch model with the optimal joint objective function of multi-microgrids and active distribution networks was established, and the load-renewable energy matching degree was considered as a bridge connecting the upper and lower layers. The schematic diagram is shown in Figure 1 .

[0123] Comprehensive dynamic pricing mechanism considering load-renewable energy matching

[0124] The specific formulation method of this dynamic pricing mechanism can be described as: 1. Determine the relationship between the equivalent load distribution of the microgrid group and the numerical value of new energy power generation in the distribution network, and formulate a unified microgrid electricity purchase price based on the time-of-use electricity price; 2. Consider the degree of matching between energy and load when the microgrid is connected to different nodes. At this time, the microgrid presents different electricity purchase prices; 3. The upper-level distribution network and the lower-level microgrid group iteratively update the dynamic electricity price.

[0125] (1) Equivalent load calculation

[0126] Since the formulation of upper-layer energy prices is affected by the lower-layer load demand, this paper introduces the concept of equivalent load (EL) to describe the load demand of the lower layer in each period. The lower-layer equivalent load includes the microgrid electric and thermal loads as well as the electric and thermal energy storage power. The lower-layer equivalent load can be expressed as:

[0127] (6)

[0128] In the formula, Meaning is that microgrid 1 is t Time period electrical equivalent load, Meaning is that Microgrid 2 is t Time period electrical equivalent load, Meaning is that Microgrid 3 is t Time period electrical equivalent load, Meaning is that microgrid 1 is t Thermal equivalent load of the time period, Meaning is that microgrid 1 is t Initial electrical load of the period, Meaning is that microgrid 1 is t The power load can be shifted in different time periods. Meaning is that microgrid 1 is t Time period energy storage output, Meaning is that microgrid 1 is t The power load can be interrupted during the period. Meaning is that microgrid 1 is t Electric boiler power consumption during the period, Meaning is that Microgrid 2 is t Initial electrical load of the period, Meaning is that Microgrid 2 is t Time period energy storage output, Meaning is that Microgrid 3 is t Initial electrical load of the period, Meaning is that Microgrid 3 is t The power load can be shifted in different time periods. Meaning is that Microgrid 3 is t The power load can be interrupted during the period. Meaning is that microgrid 1 is t Initial heat load of the period, Meaning is that microgrid 1 is t Heat load can be shifted by time period. Meaning is that microgrid 1 is t The heat load can be interrupted during the period. Meaning is that microgrid 1 is t Output of period thermal energy storage.

[0129] The heat load required by the lower microgrid is generated by the electric boiler. For the upper distribution network, the heat load can be converted into electrical load demand. The electrical equivalent load in the iteration process It can be expressed as:

[0130] (7)

[0131] Where: is the heat-to-electricity efficiency of the microgrid 1 electric boiler, It means heat to electricity.

[0132] (2) Dynamic pricing mechanism considering load-renewable energy matching

[0133] Taking the MG1 system as an example, the dynamic pricing mechanism considering load distribution can be expressed as:

[0134] (8)

[0135] (9)

[0136] Where: Time-of-use electricity price; is the electricity price coefficient, For the The dynamic electricity price of the iteration, is the wind-solar combined output value of the active distribution network, It is the ratio coefficient of load / wind and solar.

[0137] Active distribution network upper model

[0138] Objective Function

[0139] In order to cope with the challenges of wind power and photovoltaic uncertainty, we use CWGAN-GPCT to generate a large number of renewable energy operation scenarios to improve the reliability of scheduling results. The upper objective function includes network loss costs and renewable energy generation unit operation costs. Equation (10) aims to minimize the weighted average of the cumulative output of each scenario.

[0140] (10)

[0141] Where T is the scheduling period (24h); For the generated active distribution network scenario, is the probability of active distribution network scenario, is the upper objective function, , They are respectively the wind power operation and maintenance cost coefficient and the photovoltaic power generation operation and maintenance cost coefficient. For branch ij exist t The current during the period, For branch ij The resistance, , for t During the period i The actual active power output of the node wind and photovoltaic power generation units, is the set of all nodes in the distribution network, is the set of all branches of the distribution network, is the loss coefficient.

[0142] Constraints

[0143] Branch flow constraints:

[0144] This paper adopts the AC power flow method and uses the DistFlow model to implement the node power balance constraint. The DistFlow model is relaxed and converted into a second-order cone model for solution. The specific model is formula (11).

[0145] (11)

[0146] In the formula, For Node i exist t The voltage during the time period, , For branch ij Node i For the first end t Active power and reactive power during the time period, , For Node j exist t The injected power during the time period, For branch ij The square of the current allowed to flow.

[0147] 2) Distributed Generation (DG) Constraints

[0148] The distributed photovoltaic units and distributed wind power units are connected to corresponding nodes through inverters.

[0149] (13)

[0150] (14)

[0151] In the formula, , They are t During the period i Actual active power output of wind and photovoltaic power generation units at the node; , They are i The active power limit of the node wind and photovoltaic power generation units.

[0152] 3) SVC constraints:

[0153] (15)

[0154] In the formula, for t SVC in the active distribution network at the node during the period i The actual reactive power output, i The node's SVC is t The reactive power compensated during the period, , They are the upper and lower limits of SVC reactive power compensation respectively.

[0155] Node power balance constraints

[0156] (16)

[0157] (17)

[0158] Where: , They are i Active and reactive power consumed by node loads, For active distribution networks i Active power consumed by nodes connected to the microgrid.

[0159] Multi-microgrid underlying model

[0160] Objective Function

[0161] The objective function of the lower multi-microgrid includes the electricity interaction cost between the microgrid and the distribution network, the natural gas purchase cost, and the demand response compensation fee. Similarly, in order to deal with the uncertainty of wind power and photovoltaic power in multiple microgrids, the same processing method as the upper objective function (10) is adopted.

[0162] (18)

[0163] In the formula, Scenario generated for multi-microgrid, The probability of generating scenarios for multi-microgrids, is the lower layer objective function, is the gas purchase price for the microgrid, , are the compensation prices for electric load interruption and thermal load interruption respectively; MG is the collection of multiple microgrids in the lower layer. Microgrid i Time t Electricity load interruption compensation price, Microgrid i Time t Heat load interruption compensation price, It means dynamic electricity price. It means the purchased power. It means the power consumed by the electric boiler.

[0164] Constraints

[0165] Gas Turbine Constraints

[0166] (19)

[0167] In the formula, Microgrid i exist t Gas turbine power generation during the period, Microgrid i exist t The natural gas power input to the gas turbine during the period, Microgrid i Gas turbine energy conversion efficiency, Microgrid i Minimum power generation of gas turbine, Microgrid i The maximum power generation capacity of the gas turbine.

[0168] Electric boiler constraints

[0169] (20)

[0170] In the formula, Microgrid i exist t Thermal power of electric boiler during the period, Microgrid i exist t Electric power of electric boiler during the period, Microgrid i The electricity-to-heat efficiency of electric boilers, Microgrid i Minimum electric power of electric boiler, Microgrid i Maximum electric power of electric boiler.

[0171] Battery storage constraints

[0172] (twenty one)

[0173] In the formula, CN For energy storage collection; ESS , HSD For electrical and thermal energy storage; for CN The power of the device at time t; for CN The capacity of the equipment; , for CN Charging and discharging power of the equipment; , for CN The charging and discharging efficiency of the equipment; , for CN Upper and lower limits of equipment capacity; is a binary variable for charge and discharge; , for CN The maximum value of the device's charge and discharge power. Meaning is the first moment CN The capacity of the device, Meaning is the 24th moment CN The capacity of the device, The meaning is t -1 moment CN The capacity of the device.

[0174] Comprehensive demand response constraints

[0175] (twenty two)

[0176] In the formula, E , H They are electricity demand response and heat demand response respectively; is the total power of demand response load; , They are interruptible demand response and time-shiftable demand response; is the initial load; is the maximum power that the load can use to respond to demand; is the interruptible load ratio coefficient; is the maximum value of interruptible load; is the time-shiftable load proportionality factor, Meaning is t Energy demand response load during the period, where en is the energy category, l is the load, Meaning is t Time period demand response load

[0177] MG power balance constraint

[0178] (twenty three)

[0179] In the formula, The meaning is that the microgrid is t The actual power of photovoltaic power generation during the period, The meaning is that the microgrid is t Time period electrical equivalent load.

[0180] Model Solution

[0181] Joint optimization objective function

[0182] In order to achieve the best economic benefit, the joint optimization objective function is expressed as:

[0183] (twenty four)

[0184] In the formula, , are the weight coefficients of the upper and lower layers respectively.

[0185] Model solving process

[0186] The model solution flow chart is as follows Figure 2 The specific steps include:

[0187] Calculate the expected value of wind and solar day-ahead scheduling based on improved generative adversarial networks;

[0188] Generate mathematical models of active distribution networks;

[0189] Using the second-order cone relaxation power flow equation, a scheduling model in linear programming form is obtained;

[0190] Enter the initial electricity price;

[0191] Use real-time pricing mechanism to obtain real-time electricity prices;

[0192] Input active distribution network parameters;

[0193] Call CPLEX to solve the active distribution network model;

[0194] Obtain the optimal dispatching plan for active distribution networks (real-time electricity prices);

[0195] Generate optimal dispatch model for multiple microgrids;

[0196] Input multiple microgrid parameters and real-time electricity prices;

[0197] Call the CPLEX algorithm to solve the multi-microgrid model;

[0198] Obtain the optimal scheduling solution for multiple microgrids;

[0199] Calculate the joint objective function;

[0200] Determine whether the termination condition is met;

[0201] If the equivalent load power is updated, the real-time pricing mechanism is adopted again to obtain the real-time electricity price;

[0202] If the optimal coordination scheduling solution is output.

[0203] Case verification and analysis

[0204] Case Study

[0205] The multi-stakeholder entity proposed in this paper consists of an upper-level IEEE33-node active distribution network operator and three lower-level microgrids. The three microgrids MG1, MG2, and MG3 are connected to the distribution network nodes 17, 20, and 30, respectively. The structure integrates power and information through power lines and communication lines. The ADN operates at a system voltage of 12.66 kV, with a 400 kW photovoltaic system at node 32 and a 300 kW wind turbine system at node 16. The SVC devices are located at nodes 5, 15, and 31, with a capacity range from [-100, 300] kVar. Table 1 shows the installed capacity of the equipment in the microgrid. Figure 4 For the load demand of multiple microgrids, the AND structure containing multiple microgrids is as follows Figure 3 shown.

[0206] Table 1 Equipment installation capacity in microgrid

[0207]

[0208] Scenery generation

[0209] This paper adopts the CWGAN-GPCT scenario generation model, and the dataset used covers the power generation data of wind turbines and photovoltaics in a certain area in northwest China from January 1, 2018 to July 1, 2022. The dataset contains the output data of three microgrids and one distributed renewable energy station in the distribution network, including the actual output and the forecast of the previous day. The daily sample consists of 24 continuous data sets. After data cleaning to remove significant outliers, 1000 days of data were selected for model training and verification, and another 200 days were used for testing. The scenario generation process is optimized using the Kmeans++ method. Figure 5 and Figure 6 The scenario of wind power and photovoltaic output on May 15, 2022 is shown.

[0210] To evaluate the accuracy of the CWGAN-GPCT model, we compared it to the CWGAN-GP model with the same network structure, focusing on the impact of adding the CT term to the training strategy for wind and solar scene generation. Wasserstein distance, a metric that measures the difference in data distribution between datasets, is used to evaluate the characteristics of the data distribution and directly evaluate the quality of the generated samples. A Wasserstein distance close to zero indicates higher scene quality as model training progresses. Applying this metric requires compliance with the 1-Lipschitz continuity constraint. The L2 norm of the gradient is used as an indicator that the model maintains the 1-Lipschitz condition, and its value closer to 1 indicates better compliance.

[0211] Figure 7 and Figure 8 The gradient norm and Wasserstein distance of the discriminator during the model training iteration are shown. Compared with the original CWGAN-GP model, the CWGAN-GPCT model introduced in this paper converges to the unit gradient norm significantly faster and with a correspondingly higher convergence level. The Wasserstein distance tends to 0 and the jitter amplitude is significantly reduced, which means better conformance to the probability distribution. In other words, adding GP-CT to the training strategy can better capture the distribution of historical observations and improve the quality of generated samples.

[0212] Iteration result analysis

[0213] This study targets a two-tier scheduling model of ADN and MGs, which is set to end after 10 iterations under specified hardware. The run takes 20.53 seconds to complete, which meets the computing time required for day-ahead scheduling.

[0214] Fig. 9 The iterative process of the two-layer model is shown. The analysis shows that the ninth iterative model has a joint optimal solution, with the upper layer operating income of 2243.37 yuan and the lower layer operating cost of 8096.67 yuan, and the joint optimal solution value is 5940.91 yuan.

[0215] Fig.10 The figure shows the electricity prices provided by the upper distribution network to the three microgrids at the lower level. It can be seen that MG3 at node 30 faces a higher electricity purchase price, while MG1 at node 17 enjoys a lower price. This is due to the fact that node 17 has more new energy access, which reduces its need to purchase electricity from the active distribution network; in contrast, MG3 lacks nearby power access and has a higher load demand, so it purchases more electricity from the ADN and therefore bears greater losses and higher prices. In addition, the electricity prices of all three microgrids peak between 9:00-11:00 and 15:00-16:00, which is related to the increase in AND load demand during these periods.

[0216] Comparison of different pricing mechanism strategies

[0217] Economic Analysis

[0218] To verify the effectiveness of the dynamic pricing mechanism and optimized scheduling strategy proposed in this paper, this paper discusses four different scheduling scenarios, as follows:

[0219] Strategy 1: Joint optimization of the upper and lower layers, taking into account the minimum operating cost of the upper distribution network and the minimum operating cost of the lower microgrid, and adopting time-of-use electricity price as the pricing mechanism between the upper and lower layers.

[0220] Strategy 2: The upper layer is optimized independently, only the upper layer distribution network operation cost is considered to be minimum, and the lower layer microgrid operation cost is not taken into account. The dynamic pricing mechanism proposed in this paper is adopted between the upper and lower layers.

[0221] Strategy 3: The lower layer is optimized independently, only the operating cost of the lower layer microgrid is considered, and the operating cost of the upper layer distribution network is not taken into account. The dynamic pricing mechanism proposed in this paper is adopted between the upper and lower layers.

[0222] Strategy 4: Joint optimization of the upper and lower layers, taking into account the minimum operating cost of the upper distribution network and the minimum operating cost of the lower microgrid, and adopting the dynamic pricing mechanism proposed in this paper between the upper and lower layers.

[0223] Table 2 Analysis of the impact of different operation strategies on the two-level optimization model

[0224]

[0225] Table 2 shows the impact of different operation strategies on the distribution network-microgrid two-layer model. It can be seen that compared with the time-of-use electricity price mechanism of strategy 1, strategy 4 reduces the joint operation cost and gas turbine power by 13.84% and 78.37% respectively, and the total power purchased from the upper distribution network is increased by 32.74%. Compared with the independent optimization of the upper and lower layers in strategies 2 and 3, strategy 4 considers the dynamic pricing mechanism to achieve effective coordination of the upper and lower operation costs. The independent operation of the upper distribution network improves the operation benefits of the upper layer, but at the expense of the operation costs of the lower microgrid. Similarly, the independent operation of the lower layer reduces the operation benefits of the upper distribution network. In addition, compared with strategies 2, 3, and 4, the gas turbine output of the lower microgrid is the highest when the upper distribution network is independently optimized, and the gas turbine output of the microgrid is the lowest when the lower microgrid is independently optimized. Under the joint optimization of the proposed pricing mechanism, the gas turbine output of the microgrid is in the middle. The analysis shows that in terms of pricing mechanism, the dynamic pricing mechanism proposed in this paper can better coordinate the interests of the upper and lower layers compared with the time-of-use electricity price, and effectively stimulate the interaction between the microgrid and the active distribution network. By increasing the power purchase, the microgrid can participate more in the optimization of the upper distribution network on the basis of ensuring its own economic operation, reduce the output power of the controllable units of the lower microgrid, and effectively coordinate the economic operation of the upper and lower layers. In terms of the operation model, compared with independent optimization, joint optimization fully considers the interests of multiple subjects at the upper and lower levels, and coordinates through a dynamic pricing mechanism to achieve the joint optimal operation of multiple stakeholders at the upper and lower levels.

[0226] Security Analysis

[0227] When analyzing the safety of system operation, this study focuses on the enhancement effect of various pricing mechanisms on node voltage stability under different load conditions, taking into account the radial structure of the distribution network. Different pricing strategies play a key role in keeping node voltage within safe operating limits during peak and off-peak hours. Dynamic pricing has a particularly significant impact on energy consumption behavior and overall system stability.

[0228] Fig.11 The extreme values ​​of node voltages during the dispatch period for strategies 1 and 4 are compared. Due to the dynamic pricing incentive mechanism, strategy 4 significantly optimizes the minimum node voltage distribution, with a value close to 1 pu. This mechanism adjusts the incentive signal according to node power injection, enabling the microgrid to self-adjust, meet demand more efficiently, and enhance interaction with the distribution network, thereby achieving optimized voltage distribution.

[0229] Fig.12 The optimal node voltage distribution at different times in the distribution network under double-layer scheduling is shown. Fig.11, it is noted that the maximum node voltages at nodes 5, 15, and 31 are close to peak fluctuations due to the SVC equipment at these nodes. By providing sufficient reactive power, the network reduces operating losses by guiding the SVC to output reactive power and improve the node voltage level. However, despite the guidance of the dynamic mechanism, some node voltages are close to extreme values, and there is a risk of voltage violations. This emphasizes the need for network operators to maintain comprehensive voltage protection.

[0230] in conclusion

[0231] Aiming at the problem of coordinated optimization of multiple stakeholders in the distribution network-microgrid, this paper uses CWGAN-GPCT to generate renewable energy day-ahead scenarios, establishes a two-layer optimization model of the upper distribution network and the lower multi-microgrid, and proposes a dynamic pricing mechanism that considers the matching degree of load renewable energy. The established two-layer model is iteratively solved by CPLEX, and the following conclusions are obtained:

[0232] (1) This paper uses the CWGAN-GPCT model to generate high-quality wind turbine and photovoltaic day-ahead scheduling scenarios. Compared with the CWGAN-GP model, the improved model has better gradient norm convergence during the training process, and the Wasserstein distance can reach a value closer to zero.

[0233] (2) The two-layer optimization scheduling model established in this paper achieves a balance of interests between the distribution network and multiple microgrids by considering the energy-load distribution of the upper and lower layers and coordinating and flexible demand response.

[0234] (3) The dynamic pricing mechanism proposed in this paper can stimulate the demand response potential of MG, effectively reduce the joint operation cost of the two-layer model, and ensure that the interactive power of the microgrid and the distribution network is within a reasonable range. At the same time, the mechanism can optimize the flow of each branch of the distribution network and improve the overall voltage of the distribution network.

[0235] like Fig.13 As shown in the figure, an active distribution network control device considering the randomness of wind and solar power and dynamic pricing is proposed, including a generation module and an output module.

[0236] The generation module is used to generate scenarios based on the uncertainty of wind and solar power output using a conditional generative adversarial network, and then perform scenario reduction.

[0237] The output module is used to guide the power interaction between microgrids and active distribution networks based on the distribution of upper and lower energy and loads through a two-layer optimization scheduling model with the optimal joint objective function of multiple microgrids and active distribution networks. It also considers the comprehensive dynamic pricing mechanism of load-renewable energy matching, effectively coordinates the distribution of benefits between distribution networks and multiple microgrids, and improves the voltage distribution of distribution networks.

[0238] Verification module, used to perform verification analysis on the improved IEEE33 node system.

[0239] like Fig.14 As shown, the active distribution network control device considering the randomness of wind and solar power and dynamic pricing can adopt various types of computing devices such as desktop computers, notebooks, PDAs and cloud servers. This device is composed of multiple components such as processors and memories, but is not limited to these and may include other components or different combinations. The device examples provided here are for illustrative purposes only and are not restrictive.

[0240] The processor can be various types of devices, such as CPU, DSP, ASIC, FPGA, etc., and can also include programmable logic devices, discrete gates, transistor logic devices and discrete hardware components, etc. The memory is used to store computer programs and modules. The processor runs or executes the computer program modules stored in the memory and calls the data stored in the memory, thereby realizing various functions of balancing the interests between the distribution network and multiple microgrids.

[0241] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0242] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of proposing an active distribution network control method that considers the randomness of wind and solar power and dynamic pricing.

[0243] If the active distribution network control method proposed in the present invention that takes into account the randomness of wind and solar power and dynamic pricing is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device.

[0244] Based on this understanding, the present invention implements all or part of the processes in the above-mentioned active distribution network control method considering the randomness of wind and solar power and dynamic pricing, and can also be completed by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned active distribution network control method considering the randomness of wind and solar power and dynamic pricing can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or preset intermediate form, etc.

[0245] Computer-readable storage media may refer to any entity or device that can carry computer program code, including recording media, USB flash drives, mobile hard drives, disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), software distribution media, etc.

[0246] It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electrical carrier signals and telecommunication signals.

[0247] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An active distribution network control method considering the randomness of wind and solar power and dynamic pricing, characterized in that: include: Step 1: In view of the uncertainty of wind and solar power output, a conditional generative adversarial network is used to generate scenarios, and then the scenarios are reduced; Step 2: Through the two-layer optimization dispatching model with the optimal joint objective function of multiple microgrids and active distribution networks, a comprehensive dynamic pricing mechanism considering the matching degree of load and renewable energy is established to guide the power interaction between microgrids and active distribution networks on the basis of the upper and lower energy and load distribution conditions, effectively coordinate the distribution of benefits between distribution networks and multiple microgrids, and improve the voltage distribution of distribution networks; Based on the CWGAN-GPCT generation scenario, CGAN generates a network based on GAN, retaining the game structure of the GAN generator G and the discriminator D, which conforms to The distributed random noise z and conditional label y are input to the generator G; The goal of the generator G is to generate a distribution scenario that conforms to the given label by learning the parameters G(z|y). The resulting distribution is denoted by The input of the discriminator D is historical data x, random noise z and conditional label y. The goal of the discriminator D is to distinguish and And determine whether the random noise z meets the conditional label y. In CWGAN-GP based on Wasserstein distance and adding GP term, the objective function L G And the loss function L D They are described as follows: In the formula, is the Wasserstein distance, is the upper bound of the discriminator D expectation, ‖D‖L≤1 is the 1-Lipschitz constraint, and the observed data point Exp is the expectation, is the expectation of generating data, [D(x|y)] is the input x to the discriminator under label y, is the differential, yes Input discriminator D, is the expectation of the original data, is the expectation of the generated noise, time period t, and U is a set of 0-1 parameters; The consistency term CT is introduced in CWGAN-GP by using two x′ and x″ close to the observed data point The perturbed data points improve the continuity condition, which can be expressed as follows: Where M′ is the boundary, d is the two-norm input space, D(x) is the perturbed data point x′ input to the discriminator D, D(x″) is the perturbed data point x″ input to the discriminator D, and D - (x′) is the perturbed data point x″ input into the penultimate layer of the discriminator D, D - (x″) is the perturbed data point x″ input into the penultimate layer of the discriminator D; The loss function L of the discriminator D D It is expressed as: Where, proportional parameter λ1, proportional parameter λ2.

2. The method according to claim 1, characterized in that Step 2 specifically includes: coordinating the power interaction between the upper model of the active distribution network and the lower model of multiple microgrids by jointly optimizing the objective function.

3. The method according to claim 2, characterized in that Power interaction includes using equivalent load to describe the load demand of the lower layer in each period. The lower equivalent load includes the microgrid electric and thermal load as well as the electric and thermal energy storage power. The lower equivalent load is expressed as: is the equivalent load of microgrid 1 in period t, is the equivalent load of microgrid 2 in period t, is the equivalent load of microgrid 3 in period t, is the thermal equivalent load of microgrid 1 in period t, is the initial load of microgrid 1 in period t, is the load that can be shifted in microgrid 1 during period t, P t ESS,MG1 is the energy storage output of microgrid 1 in period t, is the load that can be interrupted in microgrid 1 during period t, is the power consumption of the electric boiler in microgrid 1 during period t, is the initial load of microgrid 2 in period t, is the energy storage output of microgrid 2 during period t, is the initial load of microgrid 3 in period t, is the load that can be shifted in time in microgrid 3 during period t, is the load that can be interrupted in microgrid 3 during period t, is the initial heat load of microgrid 1 in period t, is the heat load that can be shifted in time in microgrid 1 during period t, The microgrid 1 can interrupt the heat load in the period t. is the thermal energy storage output of microgrid 1 during period t.

4. The method according to claim 3, characterized in that The heat load required by the microgrid in the lower layer is generated by the electric boiler, which converts the heat load into the electric load demand. In the λth iteration process, the electric equivalent load It is expressed as: The dynamic pricing mechanism considering load distribution is expressed as: Where: is the heat-to-electricity efficiency of the microgrid 1 electric boiler, Time-of-use electricity price; is the electricity price coefficient, is the dynamic electricity price of the λth iteration, is the wind-solar combined output value of the active distribution network, It is the quotient of the electric equivalent load and the wind-solar joint output value of the active distribution network during the λth iteration.

5. The method according to claim 2, characterized in that The objective function of the upper model of the active distribution network includes network loss cost and operation cost of renewable energy power generation units, minimizing the weighted average of the cumulative output of each scenario as shown in formula (10): In the formula, T is the scheduling period, S ADN For the generated active distribution network scenario, is the probability of active distribution network scenario, is the upper layer objective function, μ WT , μ PV are wind power operation and maintenance cost coefficients and photovoltaic power generation operation and maintenance cost coefficients, respectively. ij,t is the current of branch ij in period t, r ij is the resistance of branch ij, is the actual active power output of wind power and photovoltaic power generation unit at node i during period t, Ω n is the set of all nodes in the distribution network, Ω e is the set of all branches of the distribution network, μ loss is the loss coefficient.

6. The method according to claim 5, characterized in that The branch flow constraint is adopted for the branch, which is specifically formula (11): Where U i,t is the voltage of node i during period t, P ij,t , Q ij,t is the active power and reactive power of the branch ij with node i as the head in time period t, P j,t , Q j,t is the injected power of node j in time period t, L ij,t is the square of the current allowed to flow through branch ij.

7. The method according to claim 5, characterized in that The wind turbine and photovoltaic constraints are formulas (13) and (14): In the formula, are the upper limits of active power of wind power and photovoltaic power generation units at node i respectively.

8. The method according to claim 5, characterized in that The SVC constraint in the active distribution network is formula (15): In the formula, is the actual reactive power output of the SVC at node i in the active distribution network during period t, and the reactive power compensated by the SVC at node i during period t. They are the upper and lower limits of SVC reactive power compensation respectively.

9. The method according to claim 5, characterized in that The node power balance constraints are given by formulas (16) and (17): Where: are the active and reactive power consumed by the load at node i, It is the active power consumed by the i-node connected to the microgrid in the active distribution network.

10. The method according to claim 2, characterized in that The objective function of the lower layer model of multi-microgrid is expressed by formula (18): In the formula, Scenario generated for multi-microgrid, The probability of generating scenarios for multi-microgrids, is the lower layer objective function, is the gas purchase price for the microgrid, MG is the collection of multiple microgrids at the lower layer, is the compensation price for power load interruption in microgrid i during period t, is the compensation price for thermal load interruption of microgrid i during period t, is the load that can be interrupted in microgrid i during period t, is the heat load that can be interrupted by microgrid i in period t, It is a dynamic electricity price. is the purchased power, is the power consumption of the electric boiler, is the power generated by the gas turbine of microgrid i during period t.

11. The method according to claim 10, characterized in that The gas turbine constraints are expressed using formula (19): In the formula, It means the power generated by the gas turbine of microgrid i in period t. is the natural gas power input to the gas turbine of microgrid i during period t, is the minimum power generation value of the gas turbine in microgrid i, is the maximum power generation capacity of the gas turbine in the microgrid i, is the energy conversion efficiency of the gas turbine in the microgrid.

12. The method according to claim 10, characterized in that The electric boiler constraint is expressed by formula (20): In the formula, is the thermal power of the electric boiler of microgrid i in period t, is the electric power of the electric boiler of microgrid i in period t, is the electricity-to-heat efficiency of the microgrid electric boiler, is the minimum electric power of the microgrid electric boiler, It is the maximum electric power of the microgrid i electric boiler.

13. The method according to claim 10, characterized in that The microgrid also includes battery energy storage, and the battery energy storage constraint is expressed using formula (21): Where CN is the energy storage set, is the power of the CN device at time t, is the capacity of the CN equipment at time t, are the charging and discharging power of the CN device at time t, are the charging and discharging efficiency of CN equipment, are the maximum and minimum values ​​of CN equipment capacity, is a binary variable for charge and discharge, are the maximum value of the charge and discharge power of the CN equipment, is the capacity of the CN device at the first moment, is the capacity of the CN equipment at the 24th moment, is the capacity of the CN equipment at time t-1.

14. The method according to claim 10, characterized in that The objective function of the multi-microgrid lower-level model also includes a comprehensive demand response constraint, which is expressed by formula (22): In the formula, en is the energy category, E and H are the electricity demand response and heat demand response respectively. is the total power of demand response load, They are interruptible demand response and time-shiftable demand response. is the initial load, P enL,max is the maximum power that the load can use to respond to demand, is the interruptible load ratio coefficient, P enIL,max is the maximum interruptible load value, is the time-shiftable load proportionality factor, is the energy demand response load during period t, It means the demand response load during period t.

15. The method according to claim 10, characterized in that The microgrid power balance constraint is expressed by formula (23): is the actual photovoltaic power generation power of microgrid i in period t, is the equivalent load of microgrid i in period t, is the thermal power of the electric boiler in microgrid i during period t.

16. The method according to claim 2, characterized in that The joint optimization objective function is expressed as C JO : In the formula, K upper , K lower are the weight coefficients of the upper and lower layers respectively, is the upper objective function, is the lower objective function.

17. Active distribution network control equipment considering wind and solar randomness and dynamic pricing, characterized by: Including generation module and output module, The generation module is used for: Step 1, in view of the uncertainty of wind and solar power output, a conditional generative adversarial network is used to generate scenarios, and then scenario reduction is performed; The output module is used for: Step 2, through the two-layer optimization scheduling model with the optimal joint objective function of multiple microgrids and active distribution networks, the comprehensive dynamic pricing mechanism considering the matching degree of load renewable energy guides the power interaction between microgrids and active distribution networks on the basis of the upper and lower energy and load distribution conditions, effectively coordinates the distribution of benefits between distribution networks and multiple microgrids, and improves the voltage distribution of distribution networks; Based on the CWGAN-GPCT generation scenario, CGAN generates a network based on GAN, retaining the game structure of the GAN generator G and the discriminator D, which conforms to The distributed random noise z and conditional label y are input to the generator G; The goal of the generator G is to generate a distribution scenario that conforms to the given label by learning the parameters G(z|y). The resulting distribution is denoted by The input of the discriminator D is historical data x, random noise z and conditional label y. The goal of the discriminator D is to distinguish and And determine whether the random noise z meets the conditional label y. In CWGAN-GP based on Wasserstein distance and adding GP term, the objective function L G And the loss function L D They are described as follows: In the formula, is the Wasserstein distance, is the upper bound of the discriminator D expectation, ‖D‖L≤1 is the 1-Lipschitz constraint, and the observed data point Exp is the expectation, is the expectation of generating data, [D(x|y)] is the input x to the discriminator under label y, is the differential, yes Input discriminator D, is the expectation of the original data, is the expectation of the generated noise, time period t, and U is a set of 0-1 parameters; The consistency term CT is introduced in CWGAN-GP by using two x′ and x″ close to the observed data point The perturbed data points improve the continuity condition, which can be expressed as follows: Where M′ is the boundary, d is the two-norm input space, D(x′) is the perturbed data point x′ input to the discriminator D, D(x″) is the perturbed data point x″ input to the discriminator D, and D - (x′) is the perturbed data point x″ input into the penultimate layer of the discriminator D, D - (x″) is the perturbed data point x″ input into the penultimate layer of the discriminator D; The loss function L of the discriminator D D It is expressed as: Where, proportional parameter λ1, proportional parameter λ2.

18. The device according to claim 17, characterized in that Step 2 specifically includes: power interaction includes coordinating the power interaction between the upper model of the active distribution network and the lower model of the multiple microgrids through a joint optimization objective function.

19. The device according to claim 18, characterized in that The equivalent load is used to describe the load demand of the lower layer in each period. The lower equivalent load includes the electric and thermal load of the microgrid and the electric and thermal energy storage power. The lower equivalent load is expressed as: is the equivalent load of microgrid 1 in period t, is the equivalent load of microgrid 2 in period t, is the equivalent load of microgrid 3 in period t, is the thermal equivalent load of microgrid 1 in period t, is the initial load of microgrid 1 in period t, is the load that can be shifted in time in microgrid 1 during period t, is the energy storage output of microgrid 1 in period t, is the load that can be interrupted in microgrid 1 during period t, is the power consumption of the electric boiler in microgrid 1 during period t, is the initial load of microgrid 2 in period t, is the energy storage output of microgrid 2 during period t, is the initial load of microgrid 3 in period t, is the load that can be shifted in time in microgrid 3 during period t, is the load that can be interrupted in microgrid 3 during period t, is the initial heat load of microgrid 1 in period t, is the heat load that can be shifted in time in microgrid 1 during period t, The microgrid 1 can interrupt the heat load in the period t. is the thermal energy storage output of microgrid 1 during period t.

20. The device according to claim 19, characterized in that The heat load required by the microgrid in the lower layer is generated by the electric boiler, which converts the heat load into the electric load demand. In the λth iteration process, the electric equivalent load It is expressed as: The dynamic pricing mechanism considering load distribution is expressed as: Where: is the heat-to-electricity efficiency of the microgrid 1 electric boiler, Time-of-use electricity price; is the electricity price coefficient, is the dynamic electricity price of the λth iteration, is the wind-solar combined output value of the active distribution network, It is the quotient of the electric equivalent load and the wind-solar joint output value of the active distribution network during the λth iteration.

21. The device according to claim 18, characterized in that The objective function of the upper model of the active distribution network includes network loss cost and operation cost of renewable energy power generation units, minimizing the weighted average of the cumulative output of each scenario as shown in formula (10): Where T is the scheduling period, S ADN For the generated active distribution network scenario, is the probability of active distribution network scenario, is the upper layer objective function, μ WT , μ PV are wind power operation and maintenance cost coefficients and photovoltaic power generation operation and maintenance cost coefficients, respectively. ij,t is the current of branch ij in period t, r ij is the resistance of branch ij, is the actual active power output of wind power and photovoltaic power generation unit at node i during period t, Ω n is the set of all nodes in the distribution network, Ω e is the set of all branches of the distribution network, μ loss is the loss coefficient.

22. The device according to claim 21, characterized in that The branch flow constraint is adopted for the branch, which is specifically formula (11): Where U i,t is the voltage of node i during period t, P ij,t , Q ij,t is the active power and reactive power of the branch ij with node i as the head in time period t, P j,t , Q j,t is the injected power of node j in time period t, L ij,t is the square of the current allowed to flow through branch ij.

23. The device according to claim 21, characterized in that The wind turbine and photovoltaic constraints are formulas (13) and (14): In the formula, are the upper limits of active power of wind power and photovoltaic power generation units at node i respectively.

24. The device according to claim 21, characterized in that The SVC constraint in the active distribution network is formula (15): In the formula, is the actual reactive power output of the SVC at node i in the active distribution network during period t, and the reactive power compensated by the SVC at node i during period t. They are the upper and lower limits of SVC reactive power compensation respectively.

25. The device according to claim 21, characterized in that The node power balance constraints are given by formulas (16) and (17): Where: are the active and reactive power consumed by the load at node i, It is the active power consumed by the i-node connected to the microgrid in the active distribution network.

26. The device of claim 18, wherein: The objective function of the multi-microgrid lower-level model includes the power interaction cost between the microgrid and the distribution network, the natural gas purchase cost, and the demand response compensation fee, which is expressed by formula (18): In the formula, Scenario generated for multi-microgrid, The probability of generating scenarios for multi-microgrids, is the lower layer objective function, is the gas purchase price of the microgrid, MG is the collection of multiple microgrids at the lower level, is the compensation price for power load interruption in microgrid i during period t, is the compensation price for thermal load interruption of microgrid i during period t, is the load that can be interrupted in microgrid i during period t, is the heat load that can be interrupted by microgrid i in period t, It is a dynamic electricity price. is the purchased power, is the power consumption of the electric boiler, is the power generated by the gas turbine of microgrid i during period t.

27. The device according to claim 26, characterized in that The gas turbine constraints are expressed using formula (19): In the formula, It means the power generated by the gas turbine of microgrid i in period t. is the natural gas power input to the gas turbine of microgrid i during period t, is the minimum power generation value of the gas turbine in microgrid i, is the maximum power generation capacity of the gas turbine in the microgrid i, is the energy conversion efficiency of the gas turbine in the microgrid.

28. The device according to claim 26, characterized in that The electric boiler constraint is expressed by formula (20): In the formula, is the thermal power of the electric boiler of microgrid i in period t, is the electric power of the electric boiler of microgrid i in period t, is the electricity-to-heat efficiency of the microgrid electric boiler, is the minimum electric power of the microgrid electric boiler, It is the maximum electric power of the microgrid electric boiler.

29. The device according to claim 26, characterized in that The microgrid also includes battery energy storage, and the battery energy storage constraint is expressed using formula (21): Where CN is the energy storage set, is the power of the CN device at time t, is the capacity of the CN equipment at time t, are the charging and discharging power of the CN device at time t, are the charging and discharging efficiency of CN equipment, are the maximum and minimum values ​​of CN equipment capacity, is a binary variable for charge and discharge, are the maximum value of the charge and discharge power of the CN equipment, is the capacity of the CN device at the first moment, is the capacity of the CN equipment at the 24th moment, is the capacity of the CN equipment at time t-1.

30. The apparatus of claim 26, wherein: The objective function of the multi-microgrid lower-level model also includes a comprehensive demand response constraint, which is expressed by formula (22): In the formula, en is the energy category, E and H are the electricity demand response and heat demand response respectively. is the total power of demand response load, They are interruptible demand response and time-shiftable demand response. is the initial load, P enL,max is the maximum power that the load can use to respond to demand, is the interruptible load ratio coefficient, P enIL,max is the maximum interruptible load value, is the time-shiftable load proportionality factor, is the energy demand response load during period t, It means the demand response load during period t.

31. The apparatus of claim 26, wherein: The microgrid power balance constraint is expressed by formula (23): is the actual photovoltaic power generation power of microgrid i in period t, is the equivalent load of microgrid i in period t, is the thermal power of the electric boiler in microgrid i during period t.

32. The apparatus of claim 18, wherein: The joint optimization objective function is expressed as In the formula, K upper , K lower are the weight coefficients of the upper and lower layers respectively, is the upper objective function, is the lower objective function.

33. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method according to any one of claims 1 to 16 when executing the computer program.

34. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 16 are implemented.