Distributed photovoltaic output scheduling method and system based on transaction mode analysis

Abstract

The application discloses a distributed photovoltaic output scheduling method and system based on transaction mode analysis, wherein the method is characterized in that: through data enhancement and clustering reduction on historical output data of the distributed photovoltaic, a double-layer scheduling model with power generation traders and energy operators as main bodies is determined according to power generation transaction modes, an upper optimization scheduling model of the power generation traders is constructed with the maximum economic benefit of the distributed photovoltaic power generation of the power generation traders as a target condition and the installed capacity of the distributed photovoltaic as a decision variable, a lower optimization scheduling model of the energy operators is constructed with the maximum operation benefit of the energy operators after connecting the photovoltaic as a target condition and the grid-connected capacity of the distributed photovoltaic and the distributed position of the distributed photovoltaic as decision variables, the double-layer scheduling model is optimized and solved, and a distributed photovoltaic output scheduling optimization scheme is determined, so that the analysis on the interest relationship of each main body under multiple transaction modes is considered, and the scheduling scheme meets the interest demands of each party.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and in particular to a method and system for dispatching distributed photovoltaic power output based on transaction pattern analysis. Background Technology

[0002] Distributed photovoltaic (PV) power generation, as an important component of new energy power generation, is experiencing rapid growth in installed capacity. In traditional power distribution, there is no power source connected to the user side, resulting in a stable power flow distribution direction. However, with large-scale distributed PV grid connection, the distribution network transforms into a multi-source structure, causing fluctuations in power flow direction. This can lead to equipment overload due to distributed PV power being fed back to the grid, or power flow reversal, resulting in increased voltage at the grid connection point of the PV power generation system, or even exceeding voltage limits. Therefore, comprehensive consideration of safety and economic efficiency in the planning and scheduling of distributed PV is a crucial research foundation for improving PV absorption capacity and economic benefits.

[0003] Current planning and scheduling for distributed photovoltaic (PV) power primarily addresses uncertainty by employing improved mathematical models or neural network prediction algorithms to mitigate the volatility of PV output, followed by optimized scheduling to achieve synergy between PV and other equipment. However, due to the inherent randomness of PV output, prediction methods struggle to fully capture the stochastic distribution of uncertainties. Furthermore, scheduling strategies often neglect the analysis of the interests of various stakeholders under different trading models, focusing solely on the economic viability of a single entity as the optimization objective. This results in scheduling schemes that fail to meet the needs of all parties involved. Summary of the Invention

[0004] This invention provides a distributed photovoltaic power output scheduling method and system based on transaction pattern analysis, solving the problem that existing technologies often fail to fully characterize the random distribution patterns of uncertain factors. Furthermore, the current scheduling strategies typically neglect the analysis of the interests of various stakeholders under multiple transaction patterns, considering only the economic efficiency of a single stakeholder as the optimization objective, resulting in scheduling schemes that cannot meet the needs of all parties.

[0005] In view of this, the first aspect of the present invention provides a distributed photovoltaic power output scheduling method based on transaction pattern analysis, comprising the following steps:

[0006] Obtain historical power output data for distributed photovoltaic systems;

[0007] The historical output data of the distributed photovoltaic system was augmented using the TimeGAN model.

[0008] Clustering was performed on the augmented historical output data of distributed photovoltaic power to obtain several typical scenarios of distributed photovoltaic power output.

[0009] In typical scenarios of distributed photovoltaic power output, a two-level scheduling model is determined based on a preset trading mode, with power generators and energy operators as the main entities respectively. The two-level scheduling model includes an upper-level optimized scheduling model for power generators and a lower-level optimized scheduling model for energy operators.

[0010] With the objective condition of maximizing the economic benefits of distributed photovoltaic power generation for power generators, and with the installed capacity of distributed photovoltaic as the decision variable, an upper-level optimization scheduling model for power generators is constructed.

[0011] With the objective of maximizing the operational benefits of energy operators after connecting to photovoltaics, and with the grid-connected capacity and distribution location of distributed photovoltaics as decision variables, a lower-level optimization scheduling model for energy operators is constructed.

[0012] By optimizing the two-layer scheduling model, an optimized scheme for distributed photovoltaic (PV) power output scheduling is determined based on the optimal solution. The optimized scheme for distributed PV power output scheduling includes the installed capacity of distributed PV, the grid-connected capacity of distributed PV, and the distribution location of distributed PV.

[0013] Preferably, the step of using the TimeGAN model to augment the historical output data of the distributed photovoltaic system specifically includes:

[0014] The historical output data of distributed photovoltaic power generation is divided into two dimensions: static characteristics and time characteristics. The static characteristics are divided as follows:

[0015]

[0016] In the formula, X represents the average value of photovoltaic power generation. t t represents the historical value of photovoltaic power generation; l represents the number of historical scenes, i∈{1,2,…,l}, where i represents the index of a single historical scene in the historical scenes of distributed photovoltaic power generation; t represents the time index.

[0017] The dimension division method for time features is as follows:

[0018]

[0019] In the formula, Y t This refers to the photovoltaic output error.

[0020] A TimeGAN model is constructed, and its learning objective is decomposed into an overall distribution learning objective and a time-dependent distribution learning objective for photovoltaic power output. The overall distribution learning objective is described as follows:

[0021]

[0022] In the formula, D represents the distance between distributions, p() represents the distribution of actual photovoltaic power output data, and f n G represents the static feature true vector. n,1:T Let T represent the true vector of time features, where T represents the length of the time series and n represents the number of vectors. This represents the distribution that the data generated at time follows. Represents the static feature prediction vector. Represents the time feature prediction vector;

[0023] The time-dependent distribution learning objective is described as follows:

[0024]

[0025] In the formula, f represents the static eigenvector, and g t Let g represent the time eigenvector at time t. n,1:t-1 This represents the true vector of time features at time t. This represents the time feature prediction vector at time t-1.

[0026] The static and temporal features of the historical output data of the distributed photovoltaic system are substituted into the TimeGAN model for encoding and training, so that... Closer to p(f) n ,g n,1:T This yields augmented historical output data for distributed photovoltaic systems.

[0027] Preferably, the step of clustering the augmented historical output data of distributed photovoltaic power generation to obtain several typical scenarios of distributed photovoltaic power generation specifically includes:

[0028] Using daily historical output data of distributed photovoltaic power as a sample, a sample set is constructed based on the data-enhanced historical output data of distributed photovoltaic power.

[0029] The sample set was divided into multiple clusters using the cosine similarity metric to obtain the clustering results.

[0030] The silhouette coefficient is calculated as an evaluation index of clustering results using the following formula:

[0031]

[0032] In the formula, z c (i) represents the silhouette coefficient, a(i) is the average cosine similarity between a sample and other samples in the same cluster; b(i) is the average cosine similarity between a sample and all samples in other clusters, and i represents the sample index;

[0033] The overall silhouette coefficient Z of the clustering result is obtained by averaging the silhouette coefficients of all samples. C for:

[0034]

[0035] In the formula, K represents the number of silhouette coefficients, i.e., the number of cluster centers, and Z... C The value range of is [-1, 1];

[0036] The enumeration method is used to change the value of K, and the overall profile coefficient under the current K value is calculated. The K cluster centers in the clustering result corresponding to the largest overall profile coefficient are selected as typical scenarios of distributed photovoltaic power output.

[0037] Preferably, the preset trading modes include direct trading, power grid sales on behalf of others, and benchmark purchase mode.

[0038] Preferably, the upper-level optimization scheduling model of the generator includes an objective function and constraints for upper-level optimization scheduling; wherein, the objective function for upper-level optimization scheduling is:

[0039]

[0040] In the formula, F DGPV Let γ represent the economic benefits of distributed photovoltaic power generation for power generators, and P represent the revenue coefficient of power generators selling electricity. s For electricity sold, P is the electricity subsidy coefficient for power generators. b W1 represents the total power generation, W2 represents the investment and construction cost of photovoltaic equipment, W3 represents the maintenance cost of photovoltaic equipment, and N represents the grid fees paid by the power generators.

[0041] The constraints for upper-level optimization scheduling include:

[0042] The profit constraint for distributed photovoltaic power generators participating in the electricity market is:

[0043]

[0044] In the formula, M j This represents the profit of distributed photovoltaic (PV) power generator j participating in the electricity market, g represents the PV equipment index, b represents the power generator's price range index, and ω represents the profit of distributed PV power generator j participating in the electricity market. j Let σ represent the set of generators j. n This represents the set of photovoltaic devices connected to node n. Let n be the marginal price of the region at time t. Let i be the marginal cost of photovoltaic equipment i in the power generator's quotation segment b. Let i be the power generation of photovoltaic equipment i in the power generator's bid segment b during time period t;

[0045] The capacity constraints for power generation companies are:

[0046] 0≤U' DGPV ≤U DGPVmax

[0047] In the formula, U DGPVmax U' is the maximum installed photovoltaic capacity in the region. DGPV This refers to the installed photovoltaic capacity.

[0048] The revenue restrictions for generators are as follows:

[0049] S DGPVmin ≤S DGPV

[0050] In the formula, S DGPV S represents the total investment return of power generators in photovoltaics. DGPVmin This represents the lower limit of revenue for e-commerce sellers.

[0051] Preferably, the lower-level optimization scheduling model of the energy operator includes the objective function and constraints of the lower-level optimization scheduling of the energy operator; wherein, the objective function of the lower-level optimization scheduling of the energy operator is:

[0052] maxF = D l +D sell.down +D sell.up +D ex -(E loss +E buy +E con +E tran )

[0053] In the formula, F represents the operating benefits of the energy operator after connecting to photovoltaic power, and D... l To delay the benefits of grid upgrades and renovations after grid connection to photovoltaic power, D sell.down For the revenue that energy operators generate from supplying electricity to users, D sell.up For energy operators to generate revenue from supplying electricity to the upstream power grid, D ex To absorb the quota revenue, E loss For network loss costs, E buy E contributes to the acquisition cost of photovoltaic power. con For voltage control costs, E tran For energy operators' grid access fees;

[0054] in,

[0055]

[0056] In the formula, δ con The cost of constructing a substation per unit capacity; max(W) down (t) represents the maximum active power load of the entire network; W PV(t) represents the active power load of the photovoltaic equipment; max(W down (t)-W PV (t) represents the maximum net active power load of the entire network; δ sell The electricity price; Q down Q represents the system's total annual power consumption. pv-down This represents the total direct transaction volume for the year; m and n are binary flags indicating whether data is transmitted; δ r The electricity price; Let T be the power exchanged between the distribution network and the upstream grid at time t. A positive value represents the power sold to the upstream grid, and a negative value represents the power purchased. t The number of time periods that run in the same way as the t-th time period; δ s The electricity price sold by the upstream power grid; δ p_up The transmission and distribution price is the voltage level of the upstream power grid; δ ex The price for each unit of renewable energy consumption; N pv This represents the total number of photovoltaic nodes that can be connected. Photovoltaic power output is connected at node v at time t;

[0057]

[0058] In the formula, θ is the set of branches; r ij Let be the resistance of branch ij; Let λ be the current in branch ij at time t; λ be the discount rate; T con For the service life of the control system; E fixed For fixed costs; E cable For the cost of optical fiber; E ter For terminal costs;

[0059] The constraints on the lower-level optimal dispatch of energy operators include:

[0060] Output and ramp-up power constraints of photovoltaic equipment

[0061]

[0062] In the formula, and These represent the lower and upper limits of the output power of the photovoltaic device g, respectively. and These represent the lower and upper limits of the ramp power for photovoltaic equipment g, respectively. Let g be the power output of the photovoltaic device in hour h. The power output of photovoltaic device g in hour h+1;

[0063] Constraints on energy storage equipment for energy operators:

[0064]

[0065] In the formula, K o and K g These represent the energy release and charging power of the energy storage device, δ o and δ g δ represents the states of energy release and energy charge, respectively. o and δ g All are 0-1 variables. and These represent the maximum energy release and charging power, and the State of Charge (SOC). h SOC h+1 Let be the remaining energy of the energy storage device at hour h and hour h+1, and α be the self-loss rate of the energy storage. These refer to the charging and discharging efficiencies of energy storage devices. These represent the maximum energy release and charging power of the energy storage device in hour h+1, respectively. and These are the lower and upper limits of the remaining energy stored, respectively.

[0066] The load transfer constraint is:

[0067]

[0068] In the formula, and These are the lower and upper limits of the electrical load transfer, ΔW. h It is the amount of electrical load transfer.

[0069] Preferably, by optimizing the two-layer scheduling model, a distributed photovoltaic (PV) power output scheduling optimization scheme is determined based on the optimal solution. Specifically, the steps of defining the distributed PV power output scheduling optimization scheme, which includes the installed capacity of the distributed PV, the grid-connected capacity of the distributed PV, and the distribution location of the distributed PV include:

[0070] Based on the improved non-dominated sorting multi-objective genetic algorithm, the Pareto optimal solution set is obtained by solving the upper-level optimization scheduling model of the power generator.

[0071] Using the Pareto optimal solution set as the search space, the lower-level optimal scheduling model of the energy operator is solved, yielding multiple frontier solutions, which constitute the frontier solution set. A fuzzy decision algorithm is then used to fuzzify the objective function value of the lower-level optimal scheduling model of the energy operator. The fuzzification process is as follows:

[0072]

[0073] In the formula, θ w For the fuzzy result of the w-th objective function value, F w (w) represents the w-th frontier solution of the objective function. These are the maximum and minimum values ​​of the frontier solution set of the objective function, respectively.

[0074] The weighted average of the fuzzy objective function values ​​is obtained as follows:

[0075]

[0076] In the formula, θ(w) is the weighted average, and β ω Here, θ represents the weight coefficients of the objective function, M is the number of objective functions, ω is the index of the objective function, W is the number of frontier solutions corresponding to the objective function, and θ is the weight coefficient of the objective function. ω (w) represents the fuzzification result of the w-th objective function value for each ω-th objective function;

[0077] The weighted averages are sorted, and the frontier solution corresponding to the largest objective function value is determined as the optimal compromise solution.

[0078] The optimal compromise solution is used to determine the distributed photovoltaic power output scheduling optimization scheme, wherein the distributed photovoltaic power output scheduling optimization scheme includes the installed capacity of distributed photovoltaic, the grid-connected capacity of distributed photovoltaic, and the distribution location of distributed photovoltaic.

[0079] Secondly, the present invention also provides a distributed photovoltaic power output scheduling system based on transaction pattern analysis, comprising:

[0080] The data acquisition module is used to acquire historical power output data of distributed photovoltaic systems.

[0081] The data augmentation module is used to augment the historical output data of the distributed photovoltaic system using the TimeGAN model.

[0082] The clustering module is used to cluster the data-enhanced historical output data of distributed photovoltaic power generation to obtain several typical scenarios of distributed photovoltaic power generation.

[0083] A two-layer model construction module is used to determine two-layer scheduling models based on preset trading modes in typical distributed photovoltaic power output scenarios, with power generators and energy operators as the main entities respectively. The two-layer scheduling model includes an upper-layer optimized scheduling model for power generators and a lower-layer optimized scheduling model for energy operators.

[0084] The upper-level model construction module is used to construct an upper-level optimization scheduling model for power generators with the objective condition of maximizing the economic benefits of distributed photovoltaic power generation and the installed capacity of distributed photovoltaic as the decision variable.

[0085] The lower-level model construction module is used to construct a lower-level optimization scheduling model for energy operators with the objective condition of maximizing the operational benefits after energy operators connect to photovoltaics, and with the grid-connected capacity of distributed photovoltaics and the distribution location of distributed photovoltaics as decision variables.

[0086] The optimization solution module is used to optimize the solution of the two-level scheduling model and determine the optimal solution for the distributed photovoltaic power output scheduling. The optimal solution for the distributed photovoltaic power output scheduling includes the installed capacity of the distributed photovoltaic, the grid-connected capacity of the distributed photovoltaic, and the distribution location of the distributed photovoltaic.

[0087] Thirdly, the present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described above.

[0088] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0089] As can be seen from the above technical solutions, the present invention has the following advantages:

[0090] This invention utilizes the TimeGAN model to augment historical distributed photovoltaic (PV) output data. By clustering and reducing the distributed PV output data, typical distributed PV output scenarios are obtained, comprehensively depicting the random output distribution under uncertainties. Within these typical scenarios, a two-tiered scheduling model is established, with power generators and energy operators as the main stakeholders, based on power generation trading modes. The upper-level optimization scheduling model for power generators is constructed with the installed capacity of distributed PV as the objective condition and the grid-connected capacity and location of distributed PV as the decision variables. The lower-level optimization scheduling model for energy operators is constructed with the grid-connected capacity and location of distributed PV as the objective condition and the grid-connected capacity and location of distributed PV as the decision variables. By optimizing and solving the two-tiered scheduling model, an optimal distributed PV output scheduling scheme is determined based on the optimal solution. This approach considers the analysis of the interests of various stakeholders under multiple trading modes, ensuring that the scheduling scheme meets the needs of all parties. Attached Figure Description

[0091] Figure 1 A flowchart of a distributed photovoltaic power output scheduling method based on transaction pattern analysis is provided for an embodiment of the present invention;

[0092] Figure 2 This is a schematic diagram of a distributed photovoltaic power output scheduling system based on transaction pattern analysis, provided as an embodiment of the present invention. Detailed Implementation

[0093] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0094] For easier understanding, please refer to Figure 1 The present invention provides a distributed photovoltaic power output scheduling method based on transaction pattern analysis, comprising the following steps:

[0095] S1. Obtain historical power output data of distributed photovoltaic systems.

[0096] S2. Use the TimeGAN model to augment the historical output data of distributed photovoltaic power generation.

[0097] S3. Cluster the augmented historical output data of distributed photovoltaic power generation to obtain several typical scenarios of distributed photovoltaic power generation.

[0098] It is understandable that by clustering the data-enhanced historical output data of distributed photovoltaics, distributed photovoltaic scenarios can be reduced and simplified to obtain typical scenarios that meet the requirements of distribution network planning and operation.

[0099] S4. In typical distributed photovoltaic power output scenarios, a two-layer scheduling model is determined based on a preset trading mode, with power generators and energy operators as the main entities respectively. The two-layer scheduling model includes an upper-layer optimized scheduling model for power generators and a lower-layer optimized scheduling model for energy operators.

[0100] The pre-set trading models include direct trading, grid-based electricity sales, and benchmark purchase models.

[0101] Under the direct trading model, energy operators experience a reduction in revenue from selling some of their photovoltaic (PV) power, and when PV power generation is substantial, network losses and integration costs also increase. From the perspective of power generators, although the revenue from the direct trading model is lower than that from the grid-based power sales model, choosing the direct trading model will bring greater economic returns compared to the lower electricity prices of the benchmark model.

[0102] In the grid-based electricity sales model, power generators sell electricity at a composite price, thereby obtaining revenue exceeding that of the benchmark model. However, for energy operators, when distributed photovoltaic (PV) generation is small, it may lead to delays in upgrade plans and reduced profits from reduced network losses. When PV generation is large, network losses and integration costs increase significantly. Furthermore, when the energy operator's average electricity sales price is lower than the composite price level, its economic efficiency will be further compromised. Under the benchmark purchase model, energy operators recover costs by purchasing electricity at the benchmark price and reselling it at the consumption price, a method that is more economically efficient than purchasing electricity directly from the upper-level grid.

[0103] Considering the characteristics of different trading models, it can be concluded that for power generators, cross-voltage level power flow reversal does not significantly affect their economic benefits, while the size of the photovoltaic (PV) capacity will more significantly determine their final economic return. Conversely, for energy operators, the capacity and location of PV will impact network loss costs and equipment investment. There is a mutual influence between the two stakeholders, especially in deciding on the final PV adoption rate. Therefore, to achieve optimal decision-making regarding distributed PV adoption capacity, a two-level optimization model needs to be established to encourage energy operators and power generators to make multifaceted economic decisions.

[0104] S5. With the objective condition of maximizing the economic benefits of distributed photovoltaic power generation by power generators, and with the installed capacity of distributed photovoltaic power generation as the decision variable, an upper-level optimization scheduling model for power generators is constructed.

[0105] The upper-level optimization scheduling model for power generators includes the objective function and constraints of the upper-level optimization scheduling; the objective function of the upper-level optimization scheduling is:

[0106]

[0107] In the formula, F DGPV Let γ represent the economic benefits of distributed photovoltaic power generation for power generators, and P represent the revenue coefficient of power generators selling electricity. s For electricity sold, P is the electricity subsidy coefficient for power generators. b W1 represents the total power generation, W2 represents the investment and construction cost of photovoltaic equipment, W3 represents the maintenance cost of photovoltaic equipment, and N represents the grid fees paid by the power generators.

[0108] The constraints for upper-level optimization scheduling include:

[0109] The profit constraint for distributed photovoltaic power generators participating in the electricity market is:

[0110]

[0111] In the formula, Mj This represents the profit of distributed photovoltaic (PV) power generator j participating in the electricity market, g represents the PV equipment index, b represents the power generator's price range index, and ω represents the profit of distributed PV power generator j participating in the electricity market. j Let σ represent the set of generators j. n This represents the set of photovoltaic devices connected to node n. Let n be the marginal price of the region at time t. Let i be the marginal cost of photovoltaic equipment i in the power generator's quotation segment b. Let i be the power generation of photovoltaic equipment i in the power generator's bid segment b during time period t;

[0112] The capacity constraints for power generation companies are:

[0113] 0≤U' DGPV ≤U DGPVmax

[0114] In the formula, U DGPVmax U' is the maximum installed photovoltaic capacity in the region. DGPV This refers to the installed photovoltaic capacity.

[0115] The revenue restrictions for generators are as follows:

[0116] S DGPVmin ≤S DGPV

[0117] In the formula, S DGPV S represents the total investment return of power generators in photovoltaics. DGPVmin This represents the lower limit of revenue for e-commerce sellers.

[0118] Among them, the total investment revenue of power generators in photovoltaics S DGPV It consists of the revenue from electricity sales by power generators and subsidies per kilowatt-hour.

[0119] S6. With the objective condition of maximizing the operational benefits of energy operators after connecting to photovoltaics, and with the grid-connected capacity and distribution location of distributed photovoltaics as decision variables, a lower-level optimization scheduling model for energy operators is constructed.

[0120] The lower-level optimal scheduling model for energy operators includes the objective function and constraints of the lower-level optimal scheduling; the objective function of the lower-level optimal scheduling for energy operators is:

[0121] maxF = D l +D sell.down +D sell.up +D ex -(E loss +E buy +E con +E tran )

[0122] In the formula, F represents the operating benefits of the energy operator after connecting to photovoltaic power, and D... l To delay the benefits of grid upgrades and renovations after grid connection to photovoltaic power, D sell.down For the revenue that energy operators generate from supplying electricity to users, D sell.up For energy operators to generate revenue from supplying electricity to the upstream power grid, D ex To absorb the quota revenue, E loss For network loss costs, E buy E contributes to the acquisition cost of photovoltaic power. con For voltage control costs, E tran For energy operators' grid access fees;

[0123] in,

[0124]

[0125] In the formula, δ con The cost of constructing a substation per unit capacity; max(W) down (t) represents the maximum active power load of the entire network; W PV (t) represents the active power load of the photovoltaic equipment; max(W down (t)-W PV (t) represents the maximum net active power load of the entire network; δ sell The electricity price; Q down Q represents the system's total annual power consumption. pv-down This represents the total direct transaction volume for the year; m and n are binary flags indicating whether data is transmitted; δ r The electricity price; Let T be the power exchanged between the distribution network and the upstream grid at time t. A positive value represents the power sold to the upstream grid, and a negative value represents the power purchased. t The number of time periods that run in the same way as the t-th time period; δ s The electricity price sold by the upstream power grid; δ p_up The transmission and distribution price is the voltage level of the upstream power grid; δ ex The price for each unit of renewable energy consumption; N pv This represents the total number of photovoltaic nodes that can be connected. Photovoltaic power output is connected at node v at time t;

[0126]

[0127] In the formula, θ is the set of branches; r ij Let be the resistance of branch ij; Let λ be the current in branch ij at time t; λ be the discount rate; T con For the service life of the control system; E fixed For fixed costs; E cable For the cost of optical fiber; E ter For terminal costs;

[0128] Among these measures, energy operators consider optimizing the grid-connected capacity and distribution location of photovoltaic power at multiple voltage levels to achieve the best economic benefits while meeting relevant constraints.

[0129] The constraints on the lower-level optimal dispatch of energy operators include:

[0130] The output and ramp-up power constraints of photovoltaic equipment are as follows:

[0131]

[0132] In the formula, and These represent the lower and upper limits of the output power of the photovoltaic device g, respectively. and These represent the lower and upper limits of the ramp power for photovoltaic equipment g, respectively. Let g be the power output of the photovoltaic device in hour h. The power output of photovoltaic device g in hour h+1;

[0133] Constraints on energy storage equipment for energy operators:

[0134]

[0135] In the formula, K o and K g These represent the energy release and charging power of the energy storage device, δ o and δ g δ represents the states of energy release and energy charge, respectively. o and δ g All are 0-1 variables. and These represent the maximum energy release and charging power, and the State of Charge (SOC). h SOC h+1 Let be the remaining energy of the energy storage device at hour h and hour h+1, and α be the self-loss rate of the energy storage. These refer to the charging and discharging efficiencies of energy storage devices. These represent the maximum energy release and charging power of the energy storage device in hour h+1, respectively. and These are the lower and upper limits of the remaining energy stored, respectively.

[0136] Where, δ o =0 indicates charging, δ o When δ = 1, it indicates energy release, δ g =0 indicates energy release, δ g =1 indicates that it is charged.

[0137] The load transfer constraint is:

[0138]

[0139] In the formula, and These are the lower and upper limits of the electrical load transfer, ΔW. h It is the amount of electrical load transfer.

[0140] S7. By optimizing the solution of the two-level scheduling model, the optimal solution is used to determine the distributed photovoltaic power output scheduling optimization scheme. The distributed photovoltaic power output scheduling optimization scheme includes the installed capacity of distributed photovoltaic, the grid-connected capacity of distributed photovoltaic, and the distribution location of distributed photovoltaic.

[0141] It should be noted that this invention utilizes the TimeGAN model to augment historical distributed photovoltaic (PV) output data. By clustering and reducing the distributed PV output data, typical scenarios of distributed PV output are obtained, thus comprehensively depicting the random output distribution under uncertain factors. In these typical scenarios, a two-tiered scheduling model is determined based on the power generation trading mode, with power generators and energy operators as the main entities. The upper-level optimization scheduling model for power generators is constructed with the installed capacity of distributed PV as the decision variable, while the lower-level optimization scheduling model for energy operators is constructed with the grid-connected capacity and distribution location of distributed PV as the objective condition. By optimizing and solving the two-tiered scheduling model, an optimal distributed PV output scheduling scheme is determined based on the optimal solution. This approach considers the analysis of the interests of various entities under multiple trading modes, ensuring that the scheduling scheme meets the needs of all parties.

[0142] In one specific embodiment, step S2 specifically includes:

[0143] 201. The historical output data of distributed photovoltaic power generation is divided into two dimensions: static characteristics and time characteristics. The division method for static characteristics is as follows:

[0144]

[0145] In the formula, X represents the average value of photovoltaic power generation. t t represents the historical value of photovoltaic power generation; l represents the number of historical scenes, i∈{1,2,…,l}, where i represents the index of a single historical scene in the historical scenes of distributed photovoltaic power generation; t represents the time index.

[0146] The dimension division method for time features is as follows:

[0147]

[0148] In the formula, Y t This refers to the photovoltaic output error.

[0149] 202. Construct a TimeGAN model, decomposing the learning objective of the TimeGAN model into an overall distribution learning objective for photovoltaic power output and a time-dependent distribution learning objective. The overall distribution learning objective is described as follows:

[0150]

[0151] In the formula, D represents the distance between distributions, p() represents the distribution of actual photovoltaic power output data, and f n G represents the static feature true vector. n,1:T Let T represent the true vector of time features, where T represents the length of the time series and n represents the number of vectors. This represents the distribution that the data generated at time follows. Represents the static feature prediction vector. Represents the time feature prediction vector;

[0152] The time-dependent distribution learning objective is described as follows:

[0153]

[0154] In the formula, f represents the static eigenvector, and g t Let g represent the time eigenvector at time t. n,1:t-1 This represents the true vector of time features at time t. This represents the time feature prediction vector at time t-1.

[0155] 203. Substitute the static and temporal features of the historical output data of distributed photovoltaic power into the TimeGAN model for encoding and training, so that... Closer to p(f) n ,g n,1:T This yields augmented historical output data for distributed photovoltaic systems.

[0156] The learning objective of the TimeGAN model is to use the training data to make... Closer to p(f) n ,g n,1:T The TimeGAN model is jointly trained by an autoencoder network (embedding function and recovery function) and a generative adversarial network (generator and discriminator), enabling the model to simultaneously learn distributed features, generate static and temporal features, and perform alternating training.

[0157] In one specific embodiment, step S3 specifically includes:

[0158] 301. Take the daily historical output data of distributed photovoltaic power as a sample, and construct a sample set based on the data-enhanced historical output data of distributed photovoltaic power.

[0159] 302. The sample set is divided into multiple clusters using the cosine similarity metric to obtain the clustering results.

[0160] In one example, let H i =[h i1 ,h i2 ,…,h iM [H] represents the load data sequence for day i. j =[h j1 ,h j2 ,…,h jM Let H be the load data sequence for day j. i With H j The cosine similarity between them is:

[0161]

[0162] In the formula, M is the number of daily sampling points for the load sequence, and S... cos The value range is [0, 1]. S cos The closer it is to 1, the better the sequence H is. i With H j The more similar the shapes of the load curves, the more the sample set can be divided into multiple clusters by comparing the cosine similarity using a threshold.

[0163] 303. The silhouette coefficient is calculated using the following formula as an evaluation index for clustering results. The formula for calculating the silhouette coefficient is:

[0164]

[0165] In the formula, z c (i) represents the silhouette coefficient, a(i) is the average cosine similarity between a sample and other samples in the same cluster; b(i) is the average cosine similarity between a sample and all samples in other clusters, and i represents the sample index;

[0166] 304. Calculate the mean of the silhouette coefficients of all samples to obtain the overall silhouette coefficient Z of the clustering result. C for:

[0167]

[0168] In the formula, K represents the number of silhouette coefficients, i.e., the number of cluster centers, and Z... C The value range of is [-1, 1];

[0169] 305. Use the enumeration method to change the K value, calculate the overall profile coefficient under the current K value, and select the K cluster centers in the clustering result corresponding to the largest overall profile coefficient as typical scenarios of distributed photovoltaic power output.

[0170] Among them, Z C The closer the value is to 1, the better the clustering effect.

[0171] In one specific embodiment, step S7 specifically includes:

[0172] 701. Based on the improved non-dominated sorting multi-objective genetic algorithm, the Pareto optimal solution set is obtained by solving the upper-level optimization scheduling model of the generator.

[0173] Specifically, the process of solving the upper-level optimal scheduling model of power generators and the lower-level optimal scheduling model of energy operators using an improved non-dominated sorting multi-objective genetic algorithm is as follows:

[0174] 1) Initialize the population by randomly generating a parent population Pt.

[0175] 2) Calculate the objective function value for the current population individuals.

[0176] 3) Perform non-dominated stratification and ranking of individuals in the population.

[0177] 4) Use binary tournament selection, crossover, and mutation operations to generate N offspring populations Qt.

[0178] 5) Populations Pt and Qt are merged into Rt, and Rt = Pt∪Qt.

[0179] 6) Calculate the objective function value for individuals in the new population Rt.

[0180] 7) Perform a non-dominated sort on the individuals in the population, and select the top u individuals to generate the parent population Pt+1;

[0181] 8) If the convergence condition is met (the fitness of the generated population is less than the set value), then terminate; otherwise, increase the iteration count by 1, go to step 2), and output the Pareto optimal solution set.

[0182] The Pareto optimal solution set consists of non-dominated solutions.

[0183] 702. Using the Pareto optimal solution set as the search space, the lower-level optimal scheduling model of the energy operator is solved, yielding multiple frontier solutions, which constitute the frontier solution set. A fuzzy decision algorithm is then used to fuzzify the objective function value of the lower-level optimal scheduling model of the energy operator. The fuzzification process is as follows:

[0184]

[0185] In the formula, θ w For the fuzzy result of the w-th objective function value, F w (w) represents the w-th frontier solution of the objective function. These are the maximum and minimum values ​​of the frontier solution set of the objective function, respectively.

[0186] 703. By taking a weighted average of the fuzzy objective function values, we obtain:

[0187]

[0188] In the formula, θ(w) is the weighted average, and β ω Here, θ represents the weight coefficients of the objective function, M is the number of objective functions, ω is the index of the objective function, W is the number of frontier solutions corresponding to the objective function, and θ is the weight coefficient of the objective function. ω (w) represents the fuzzification result of the w-th objective function value for each ω-th objective function;

[0189] 704. Sort the weighted averages and determine the frontier solution corresponding to the largest objective function value as the optimal compromise solution;

[0190] 705. Determine the optimal scheme for dispatching distributed photovoltaic power output based on the optimal compromise solution. The optimal scheme for dispatching distributed photovoltaic power output includes the installed capacity of distributed photovoltaic power, the grid-connected capacity of distributed photovoltaic power, and the distribution location of distributed photovoltaic power.

[0191] The above is a detailed description of an embodiment of a distributed photovoltaic power output scheduling method based on transaction pattern analysis provided by the present invention. The following is a detailed description of an embodiment of a distributed photovoltaic power output scheduling system based on transaction pattern analysis provided by the present invention.

[0192] For easier understanding, please refer to Figure 2 The distributed photovoltaic power output scheduling system based on transaction pattern analysis provided by this invention includes:

[0193] Data acquisition module 100 is used to acquire historical power output data of distributed photovoltaic systems;

[0194] Data augmentation module 200 is used to augment historical output data of distributed photovoltaic power using the TimeGAN model;

[0195] Clustering module 300 is used to cluster the data-enhanced historical output data of distributed photovoltaic power generation to obtain several typical scenarios of distributed photovoltaic power generation.

[0196] The two-layer model construction module 400 is used to determine two-layer scheduling models based on preset trading modes in typical distributed photovoltaic power output scenarios, with power generators and energy operators as the main entities respectively. The two-layer scheduling model includes an upper-layer optimized scheduling model for power generators and a lower-layer optimized scheduling model for energy operators.

[0197] The upper-level model construction module 500 is used to construct the upper-level optimization scheduling model of the power generator with the objective condition of maximizing the economic benefits of distributed photovoltaic power generation and the installed capacity of distributed photovoltaic as the decision variable.

[0198] The lower-level model construction module 600 is used to construct a lower-level optimization scheduling model for energy operators with the objective condition of maximizing the operational benefits after energy operators connect to photovoltaics, and with the grid-connected capacity of distributed photovoltaics and the distribution location of distributed photovoltaics as decision variables.

[0199] The optimization solution module 700 is used to optimize the two-level scheduling model and determine the optimal scheme for distributed photovoltaic power output scheduling based on the optimal solution. The optimal scheme for distributed photovoltaic power output scheduling includes the installed capacity of distributed photovoltaic, the grid-connected capacity of distributed photovoltaic, and the distribution location of distributed photovoltaic.

[0200] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described above.

[0201] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method.

[0202] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, electronic devices, and computer-readable storage media described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0203] In the embodiments provided by this invention, it should be understood that the disclosed systems, electronic devices, computer-readable storage media, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of devices or units may be electrical, mechanical, or other forms.

[0204] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0205] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0206] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods described in the various embodiments of the present invention through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0207] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A distributed photovoltaic power output scheduling method based on transaction pattern analysis, characterized in that, Includes the following steps: Obtain historical power output data for distributed photovoltaic systems; The historical output data of the distributed photovoltaic system was augmented using the TimeGAN model. Clustering was performed on the augmented historical output data of distributed photovoltaic power generation to obtain several typical scenarios of distributed photovoltaic power generation. In typical distributed photovoltaic (PV) power generation scenarios, a two-tier scheduling model is determined based on a preset trading mode, with power generators and energy operators as the main entities respectively. The two-tier scheduling model includes an upper-level optimization scheduling model for power generators and a lower-level optimization scheduling model for energy operators. The upper-level optimization scheduling model for power generators includes an objective function and constraints for upper-level optimization scheduling. The objective function for upper-level optimization scheduling is: In the formula, To improve the economic benefits of distributed photovoltaic power generation for power generators, This refers to the electricity sales revenue coefficient for power generators. For electricity sales, The electricity subsidy coefficient for power generators. Total power generation For the investment and construction costs of photovoltaic equipment, For the maintenance costs of photovoltaic equipment, The electricity generator pays the grid fees, where N is the installed capacity of the distributed photovoltaic system. The constraints for upper-level optimization scheduling include: The profit constraint for distributed photovoltaic power generators participating in the electricity market is: In the formula, This represents the profit of distributed photovoltaic (PV) power generator j participating in the electricity market; g represents the PV equipment index; and b represents the power generator's price range index. Let j represent the set of generators. This represents the set of photovoltaic devices connected to node n. Let n be the marginal price of the region at time t. Let i be the marginal cost of photovoltaic equipment i in the power generator's quotation segment b. Let i be the power generation of photovoltaic equipment i in the power generator's bid segment b during time period t; The capacity constraints for power generation companies are: In the formula, This represents the maximum capacity of photovoltaic systems that can be installed in the area. This refers to the installed photovoltaic capacity. The revenue restrictions for generators are as follows: In the formula, The total return on investment in photovoltaic power generation by power generators. This serves as the lower limit for the revenue of electricity distributors; The lower-level optimal scheduling model of the energy operator includes the objective function and constraints of the lower-level optimal scheduling; wherein, the objective function of the lower-level optimal scheduling of the energy operator is: In the formula, F represents the operating benefits for energy operators after connecting to photovoltaic power. To delay the benefits of grid upgrades and renovations after grid connection to photovoltaic power, For the revenue that energy operators generate from supplying electricity to users, For energy operators to generate revenue from supplying electricity to the upper-level power grid, To absorb the quota revenue, For network loss costs, To contribute to the acquisition costs of photovoltaic power, For voltage control cost, For energy operators' grid access fees; in, In the formula, Cost of construction and investment for a substation per unit capacity; This represents the maximum active power load of the entire network. This represents the active load value of the photovoltaic equipment. This represents the maximum net active power load of the entire network. The electricity price; This represents the system's total annual electricity consumption. This represents the total direct transaction volume for the year; m and n are binary flags indicating whether data is transmitted. The electricity price; Let t be the power exchanged between the distribution network and the upstream power grid at time t. A positive value represents the power sold to the upstream power grid, and a negative value represents the power purchased. This is the number of time periods that run in the same way as the t-th time period; The electricity price is the price charged by the higher-level power grid. The transmission and distribution price is based on the voltage level of the upstream power grid. The price for the unit of renewable energy consumption electricity trading; This represents the total number of photovoltaic nodes that can be connected. Photovoltaic power output is connected at node v at time t; In the formula, For branch set; Let be the resistance of branch ij; Let be the current in branch ij at time t; The discount rate; The service life of the control system; For fixed costs; For the cost of optical fiber; For terminal costs; The constraints on the lower-level optimal dispatch of energy operators include: Output and ramp-up power constraints of photovoltaic equipment: In the formula, and These represent the lower and upper limits of the output power of the photovoltaic device g, respectively. and These represent the lower and upper limits of the ramp power for photovoltaic equipment g, respectively. Let g be the power output of the photovoltaic device in hour h. The power output of photovoltaic device g in the (h+1)th hour; Constraints on energy storage equipment for energy operators: In the formula, and These refer to the energy release and charging power of the energy storage device, respectively. and These represent the states of energy release and energy charging, respectively. and All are 0-1 variables. and These are the maximum energy release and charging power, respectively. , The remaining energy of the energy storage device at hour h and hour h+1. The self-loss rate of energy storage, , These refer to the charging and discharging efficiencies of energy storage devices. , These represent the maximum energy release and charging power of the energy storage device in hour h+1, respectively. and These are the lower and upper limits of the remaining energy stored, respectively. The load transfer amount constraint is: In the formula, and These are the lower and upper limits of the electrical load transfer amount, respectively. It is the amount of electrical load transfer; With the objective condition of maximizing the economic benefits of distributed photovoltaic power generation for power generators, and with the installed capacity of distributed photovoltaic as the decision variable, an upper-level optimization scheduling model for power generators is constructed. With the objective of maximizing the operational benefits of energy operators after connecting to photovoltaics, and with the grid-connected capacity and distribution location of distributed photovoltaics as decision variables, a lower-level optimization scheduling model for energy operators is constructed. By optimizing the two-layer scheduling model, an optimized scheme for distributed photovoltaic (PV) power output scheduling is determined based on the optimal solution. The optimized scheme for distributed PV power output scheduling includes the installed capacity of distributed PV, the grid-connected capacity of distributed PV, and the distribution location of distributed PV.

2. The distributed photovoltaic power output scheduling method based on transaction pattern analysis according to claim 1, characterized in that, The specific steps for data augmentation of the historical output data of distributed photovoltaic power using the TimeGAN model include: The historical output data of distributed photovoltaic power generation is divided into two dimensions: static characteristics and time characteristics. The static characteristics are divided as follows: In the formula, This represents the average value of photovoltaic power generation. t represents the historical value of photovoltaic power generation; l represents the number of historical scenes, i∈{1,2,…,l}, where i represents the index of a single historical scene in the historical scenes of distributed photovoltaic power generation; t represents the time index. The dimensional division of the time feature is as follows: In the formula, This refers to the photovoltaic output error. A TimeGAN model is constructed, and its learning objective is decomposed into an overall distribution learning objective and a time-dependent distribution learning objective for photovoltaic power output. The overall distribution learning objective is described as follows: In the formula, D represents the distance between distributions. () represents the distribution of actual photovoltaic power output data. n Represents the static feature true vector. Let T represent the true vector of time features, where T represents the length of the time series and n represents the number of vectors. () indicates that the data generated at that time follows a distribution. Represents the static feature prediction vector. Represents the time feature prediction vector; The time-dependent distribution learning objective is described as follows: In the formula, Represents a static eigenvector. This represents the time feature vector at time t. This represents the true vector of time features at time t. This represents the time feature prediction vector at time t-1. The static and temporal features of the historical output data of the distributed photovoltaic system are substituted into the TimeGAN model for encoding and training, so that... Closer to We obtained the historical output data of distributed photovoltaic power after data enhancement.

3. The distributed photovoltaic power output scheduling method based on transaction pattern analysis according to claim 1, characterized in that, The specific steps for clustering the augmented historical output data of distributed photovoltaic (PV) power generation to obtain several typical distributed PV power generation scenarios include: Using daily historical output data of distributed photovoltaic power as a sample, a sample set is constructed based on the data-enhanced historical output data of distributed photovoltaic power. The sample set was divided into multiple clusters using the cosine similarity metric to obtain the clustering results. The silhouette coefficient is calculated as an evaluation index of clustering results using the following formula: In the formula, Represents the profile coefficient. The average cosine similarity between a given sample and other samples in the same cluster; Let i be the average cosine similarity between a given sample and all samples in other clusters, where i represents the sample index. The overall silhouette coefficient of the clustering result is obtained by averaging the silhouette coefficients of all samples. for: In the formula, K represents the number of silhouette coefficients, i.e., the number of cluster centers. The value range of is [-1, 1]; The enumeration method is used to change the value of K, and the overall profile coefficient under the current K value is calculated. The K cluster centers in the clustering result corresponding to the largest overall profile coefficient are selected as typical scenarios of distributed photovoltaic power output.

4. The distributed photovoltaic power output scheduling method based on transaction pattern analysis according to claim 1, characterized in that, The preset trading modes include direct trading, grid-based electricity sales, and benchmark purchase models.

5. The distributed photovoltaic power output scheduling method based on transaction pattern analysis according to claim 1, characterized in that, By optimizing the two-level scheduling model, an optimized scheme for distributed photovoltaic (PV) power output scheduling is determined based on the optimal solution. Specifically, the optimized scheme includes steps related to the installed capacity of distributed PV, the grid-connected capacity of distributed PV, and the distribution location of distributed PV. Based on the improved non-dominated sorting multi-objective genetic algorithm, the Pareto optimal solution set is obtained by solving the upper-level optimization scheduling model of the power generator. Using the Pareto optimal solution set as the search space, the lower-level optimal scheduling model of the energy operator is solved, yielding multiple frontier solutions, which constitute the frontier solution set. A fuzzy decision algorithm is then used to fuzzify the objective function value of the lower-level optimal scheduling model of the energy operator. The fuzzification process is as follows: In the formula, For the first The fuzzing result of the objective function value For the objective function A cutting-edge solution. , These are the maximum and minimum values ​​of the frontier solution set of the objective function, respectively. The weighted average of the fuzzy objective function values ​​is obtained as follows: In the formula, For weighted average, These are the weight coefficients of the objective function. The number of objective functions. Here, W is the index of the objective function, and W is the number of frontier solutions corresponding to the objective function. For the first The first objective function The fuzzification result of the objective function value; The weighted averages are sorted, and the frontier solution corresponding to the largest objective function value is determined as the optimal compromise solution. The optimal compromise solution is used to determine the distributed photovoltaic power output scheduling optimization scheme, wherein the distributed photovoltaic power output scheduling optimization scheme includes the installed capacity of distributed photovoltaic, the grid-connected capacity of distributed photovoltaic, and the distribution location of distributed photovoltaic.

6. A distributed photovoltaic power output scheduling system based on transaction pattern analysis, used to execute the distributed photovoltaic power output scheduling method based on transaction pattern analysis as described in any one of claims 1 to 5, characterized in that, include: The data acquisition module is used to acquire historical power output data of distributed photovoltaic systems. The data augmentation module is used to augment the historical output data of the distributed photovoltaic system using the TimeGAN model. The clustering module is used to cluster the data-enhanced historical output data of distributed photovoltaic power generation to obtain several typical scenarios of distributed photovoltaic power generation. A two-layer model construction module is used to determine two-layer scheduling models based on preset trading modes in typical distributed photovoltaic power output scenarios, with power generators and energy operators as the main entities respectively. The two-layer scheduling model includes an upper-layer optimized scheduling model for power generators and a lower-layer optimized scheduling model for energy operators. The upper-level model construction module is used to construct an upper-level optimization scheduling model for power generators with the objective condition of maximizing the economic benefits of distributed photovoltaic power generation and the installed capacity of distributed photovoltaic as the decision variable. The lower-level model construction module is used to construct a lower-level optimization scheduling model for energy operators with the objective condition of maximizing the operational benefits after energy operators connect to photovoltaics, and with the grid-connected capacity of distributed photovoltaics and the distribution location of distributed photovoltaics as decision variables. The optimization solution module is used to optimize the solution of the two-level scheduling model and determine the optimal solution for distributed photovoltaic power output scheduling. The optimal solution for distributed photovoltaic power output scheduling includes the installed capacity of distributed photovoltaic, the grid-connected capacity of distributed photovoltaic, and the distribution location of distributed photovoltaic.

7. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-5.

Images