Green electricity real-time transaction method and system based on AI large model, medium and equipment

Abstract

The invention discloses a green electricity real-time transaction method and system based on an AI large model, a medium and equipment. The method comprises the following steps: respectively acquiring demand characteristics of each load node, quality characteristics of each energy storage node and quality characteristics of each green electricity node by utilizing each transaction node; constructing a training sample set and a training model based on an AI large model, training the training model by using the training sample set, and taking the trained training model as a green electricity real-time transaction model; and inputting the demand feature of each load node, the quality feature of each energy storage node and the quality feature of each green power node into a green power real-time transaction model by using each transaction node to obtain an optimal transaction combination and a quotation scheme of each load node. According to the method, the optimal transaction combination and quotation scheme with the optimal economical efficiency can be given in real time according to the requirements of each load node, and the user cost is the lowest under the condition that the load requirements are met.

Description

Technical Field

[0001] The present invention relates to the field of green power trading technology, and in particular to a green power real-time trading method, system, medium and equipment based on an AI big model. Background Art

[0002] With the continuous development of the power market, green power (hereinafter referred to as green power) has become increasingly important as an energy subject to participate in power market transactions. However, green power, including photovoltaic power generation and wind power generation, is random, volatile, and intermittent, and cannot be stored, which makes it impossible to fully participate in the power trading market for real-time power transactions. At the same time, user load demand is also random, resulting in a serious mismatch between load demand and green power supply, resulting in low utilization of green power, weak market participation, and poor construction economy. Summary of the invention

[0003] In order to solve the technical problems existing in the background technology, the present invention proposes a green electricity real-time trading method, system, medium and equipment based on AI big model.

[0004] In the first aspect, the present invention proposes a green electricity real-time trading method based on an AI big model. The entities participating in the green electricity real-time trading include N green electricity nodes, K energy storage nodes, M load nodes and X trading nodes. Each trading node is in communication connection with N green electricity nodes, K energy storage nodes and M load nodes. The method comprises: Use each transaction node to obtain the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green power node; Constructing a training sample set and A training model based on the AI ​​big model is trained using a training sample set, and the trained model is used as a green electricity real-time trading model.

[0005] Each transaction node is used to input the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green electricity node into the green electricity real-time transaction model to obtain the optimal transaction combination and quotation plan for each load node.

[0006] Preferably, the demand characteristics of the load node include: location of the load node, maximum demand power, duration of maximum demand power, continuous operation power, duration of continuous operation power, and load conversion efficiency; The quality characteristics of green power nodes include the location of the green power node, the initial power generation cost coefficient, the installed age of the green power node, the real-time maximum output power, the duration of maximum output power, the rated output power, the rated output duration, and the green power conversion efficiency; The quality characteristics of energy storage nodes include the location of the energy storage node, the initial investment cost coefficient, the installed age of the energy storage node, the rated capacity of the energy storage node, the initial rated power, SOC, SOH, self-consumption and energy storage conversion efficiency.

[0007] Preferably, constructing a training sample set specifically includes: Use each transaction node to obtain the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, and the historical quality characteristics of each green power node; According to the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, the historical quality characteristics of each green power node, and each transaction node and the preset negotiation transaction strategy, the historical optimal transaction combination and quotation plan for each load node are obtained; A training sample set is constructed based on the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, the historical quality characteristics of each green power node, and each trading node and the corresponding historical optimal trading combination and quotation scheme.

[0008] Preferably, the preset negotiation transaction strategy specifically includes: According to the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, and the historical quality characteristics of each green electricity node, the demand characteristic set of each load node, the quality characteristic set of each energy storage node, and the historical quality characteristic set of each green electricity node are obtained respectively; According to the historical demand feature set of each load node, a load node demand data set is obtained; wherein the load node demand data set includes demand data of each load node participating in the transaction; According to the historical quality feature set of each green power node, a green power transaction form data set is obtained; wherein the green power transaction form data set includes green power transaction form data that each green power node can respond to the needs of each load node; According to the historical quality feature set of each energy storage node, an energy storage transaction form data set is obtained; wherein the energy storage transaction form data set includes energy storage transaction form data that each energy storage node can respond to the discharge demand of each load node or the charging demand of each green electricity node; according to the load node demand data set, the energy storage transaction form data set, and the green electricity transaction form data set, a transaction cost data set is obtained; Establish a load trading operation unit cost set, a storage trading dispatch unit cost set, and a green power trading dispatch unit cost set; According to the transaction cost data set, the load transaction operation unit cost set, the energy storage transaction dispatch unit cost set, and the green electricity transaction dispatch unit cost set, a total transaction cost set is obtained; wherein the total transaction cost set includes the total transaction cost of each load node when participating in the transaction at each transaction node; For each load node in the set of total transaction costs, prioritize the total transaction costs at each trading node separately, and take the lowest total cost as the final cost for each load node; Form the historical optimal trading portfolio for each load node by combining each load node with the trading node, green power node, and energy storage node corresponding to its final cost, and use the trading node corresponding to the final cost of each load node to make an offer according to a preset offer strategy.

[0009] Preferably, the set of transaction cost data includes a first set of transaction cost data and a second set of transaction cost data; Among them, the set of transaction cost data is obtained according to the set of load node demand data, the set of energy storage trading form data, and the set of green power trading form data, specifically including: Divide the demand data of each load node in the set of load node demand data into first demand or second demand; For each trading node, randomly extract the first demand of A1 load nodes from the set of load node demand data, and combine the first demand of A1 load nodes to form a first demand set; where A1 < M; According to the first preset condition, screen out the quality feature set of green power nodes that meet the first demand set from the set of green power trading form data; where the number of quality feature sets of the screened green power nodes is B1, B1 < N; and combine the quality feature sets of the B1 screened green power nodes to form a first green power data set; Among them, the first preset condition is and ; in the formula, represents the maximum power demand of the jth load node, ; , represents the maximum demand power of the jth load node, represents the load conversion efficiency of the jth load node; represents the duration of the maximum power demand of the jth load node, represents the real-time maximum output power of the ith green power node, ; represents the green power conversion efficiency of the ith green power node; represents the sustainable time of the maximum output power of the ith green power node; represents the load conversion efficiency of the jth load node; According to the first green power data set and the demand feature set of the load nodes in the first demand set it satisfies, obtain the first two-dimensional power generation tradable data set between each load node and each green power node for trading at this trading node; Determine whether B1 is greater than or equal to A1; if so, combine the first two-dimensional power generation tradable data sets of all trading nodes to form the first trading cost data set; For each trading node, randomly extract the second demands of A2 load nodes from the load node demand data set, and combine the second demands of the A2 load nodes to form the second demand set; where A2 < M; According to the second preset condition, screen out the quality feature set of green power nodes that meet the second demand set from the green power trading form data set; where the number of quality feature sets of green power nodes that meet the second demand set is B2, B2 < N; and combine the quality feature sets of the B2 green power nodes screened out to form the second green power data set; where the second preset condition is and ; in the formula, represents the continuous operating power demand of the jth load node, ; where, , represents the continuous operating demand power of the jth load node; represents the rated output power of the ith green power node, represents the rated output duration of the ith green power node; where, , represents the continuous operating demand power of the jth load node; According to the second green power data set and the demand feature set of the load nodes that meet the second demand set, obtain the second two-dimensional power generation tradable data set between each load node and each green power node for trading at this trading node; Determine whether B2 is greater than or equal to A2; if so, combine the second two-dimensional power generation tradable data sets of all trading nodes to form the second trading cost data set.

[0010] Preferably, when determining whether B1 is greater than or equal to A1, if not, screen out the quality feature set of energy storage nodes that meet the demand feature set of the remaining load nodes in the first demand set from the energy storage trading form data set; where the number of quality feature sets of energy storage nodes that meet the demand feature set of the remaining load nodes in the first demand set is C1; determine whether the sum of B1 and C1 is less than A1; if so, this trading node does not respond to the trading of the remaining A1 - B1 - C1 load nodes in the first demand set; if not, according to the quality feature sets of the C1 energy storage nodes screened out, obtain the first two-dimensional energy storage compensation data set between each load node and each energy storage node for energy storage compensation trading at this trading node; where the third preset condition is and ; in the formula, represents the energy storage rated power of the kth energy storage node, represents the dischargeable capacity of the kth energy storage node; All first two-dimensional power generation tradable data sets and first two-dimensional energy storage compensation data sets are combined to obtain a first transaction cost set.

[0011] Preferably, when judging whether B2 is greater than or equal to A2, if it is judged that B2 is less than A2, then according to the fourth preset condition, the quality feature set of the energy storage node that meets the demand feature set of the remaining load nodes in the second demand set is screened out from the energy storage transaction form data set; wherein, the number of the quality feature sets of the energy storage nodes that meet the demand feature set of the remaining load nodes in the second demand set is C2; judge whether the sum of B2 and C3 is less than A2; if so, the transaction node does not respond to the remaining A2-B2-C2 load nodes in the second demand set; if not, according to the quality feature sets of the screened C2 energy storage nodes, a second two-dimensional energy storage compensation data set between each load node and each energy storage node that performs energy storage compensation transactions on the transaction node is obtained; the fourth preset condition is and ; The second two-dimensional power generation tradable data set and the second two-dimensional energy storage compensation data set on each transaction node are combined to obtain a second transaction cost set.

[0012] Preferably, a total transaction cost set is obtained based on a transaction cost data set, a load transaction operation unit cost set, an energy storage transaction scheduling unit cost set, and a green electricity transaction scheduling unit cost set, specifically including: obtaining a first total transaction cost set based on a first transaction cost data set, a load transaction operation unit cost set, an energy storage transaction scheduling unit cost set, and a green electricity transaction scheduling unit cost set; obtaining a second total transaction cost set based on a second transaction cost data set, a load transaction operation unit cost set, an energy storage transaction scheduling unit cost set, and a green electricity transaction scheduling unit cost set.

[0013] Among them, the total cost of each load node in the transaction total cost set at each transaction node is prioritized respectively, and the lowest total cost is taken as the final cost of each load node, including: Prioritize the total costs of each load node in the first transaction total cost set and the second transaction total cost set at each transaction node respectively; take the lowest total cost of each load node in the first transaction total cost set as the first total cost of each load node; take the lowest total cost of each load node in the second transaction total cost set as the second total cost of each load node; and based on the first total cost and the second total cost of each load node, take the lowest total cost between the first total cost and the second total cost as the final cost of each load node.

[0014] Among them, the first set of total transaction costs is ; In the formula, represents the unit cost of load trading operation between each load node and the x-th trading node, represents the unit cost of energy storage trading scheduling between each energy storage node and the x-th trading node, represents the unit cost of green power trading scheduling between each green power node and the x-th trading node, represents the unit cost of green power generation; among them, , represents the initial generation cost coefficient, represents the installation age of green power; represents the unit cost of energy storage; among them, ; in the formula, represents the initial investment cost coefficient, represents the installation age of energy storage.

[0015] Among them, the second set of total transaction cost data is .

[0016] Preferably, after forming the first set of transaction cost data and the second set of transaction cost data, it further includes: If B1 + B2 < N and K > C1 + C2 and the rechargeable capacity of the remaining K - C1 - C2 energy storage nodes > 0, then store the power generated by the remaining N - B1 - B2 green power nodes into the remaining K - C1 - C2 energy storage nodes, and obtain the green power charging demand scheduling cost data set for the remaining N - B1 - B2 green power nodes to schedule energy storage to store electrical energy at each trading node; According to the energy storage trading scheduling unit cost set, the green power trading scheduling unit cost set, and the green power charging demand scheduling cost data set, obtain the total green power charging demand cost set; among them, the total green power charging demand cost set includes the total green power charging demand cost for each green power node to schedule energy storage to store electrical energy at each trading node; Sort the total green power charging demand costs for each green power node to schedule energy storage to store electrical energy at each trading node in the total green power charging demand cost data set respectively, and select the lowest total green power charging demand cost as the final total green power charging demand cost for each green power node to schedule energy storage to store electrical energy, and use the energy storage node corresponding to the final total green power charging demand cost as the charging response for each green power node; The transaction node corresponding to the final green electricity charging demand cost is used to quote the charging demand quotation unit price of each green electricity node according to the preset quotation strategy, and the charging energy storage scheduling unit price of the energy storage node that responds to the charging of each green electricity node is quoted.

[0017] Among them, the total cost of green electricity charging demand is ; In the formula, represents the continuous operation required power of the jth load node, ; represents the dischargeable capacity of the kth energy storage node, represents the energy storage rated power of the kth energy storage node, ;in, ; In the formula, represents the initial rated power of the kth energy storage node, represents the energy storage conversion efficiency of the kth energy storage node, Represents the energy storage life correction coefficient of the kth energy storage node.

[0018] In the second aspect, the present invention also proposes a green electricity real-time trading system based on an AI big model, wherein the entities participating in the green electricity real-time trading include N green electricity nodes, K energy storage nodes, M load nodes and X trading nodes, and each trading node is communicatively connected with N green electricity nodes, K energy storage nodes and M load nodes; the system includes: An acquisition module is used to use each transaction node to respectively acquire the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green power node; Model building module, used to build training models based on AI big models; A training set construction module is used to construct a training sample set; The training module is used to train the training model using the training sample set, and use the trained model as the green electricity real-time trading model.

[0019] The processing module stores a green electricity real-time trading model, which is used to use each trading node to input the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green electricity node into the green electricity real-time trading model to obtain the optimal trading combination and quotation plan for each load node.

[0020] In a third aspect, the present invention further proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the green electricity real-time trading method based on the AI ​​big model as described in any one of the first aspects.

[0021] In a fourth aspect, the present invention further proposes an electronic device, comprising: a processor and a memory, the memory being used to store one or more programs; when the one or more programs are executed by the processor, the green electricity real-time trading method based on the AI ​​big model as described in any one of the first aspects is implemented.

[0022] In the present invention, the proposed green electricity real-time trading method, system, medium and equipment based on AI big model proposes to use the green electricity real-time trading model based on AI big model, which can give the optimal trading combination and quotation plan with the best economic efficiency in real time according to the needs of each load node. While meeting the load demand, the user cost is the lowest and the green electricity revenue is guaranteed to be the best win-win situation, which promotes green electricity to fully participate in real-time market-oriented electricity trading and promotes the market development of green electricity. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flow chart of a green electricity real-time trading method based on an AI big model proposed in the present invention. DETAILED DESCRIPTION

[0024] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] Reference Figure 1 , a green electricity real-time trading method based on an AI big model proposed in the present invention includes: the entities participating in the green electricity real-time trading include N green electricity nodes, K energy storage nodes, M load nodes and X trading nodes, and each trading node is in communication connection with the N green electricity nodes, the K energy storage nodes and the M load nodes; The method comprises: Use each transaction node to obtain the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green power node; Construct a training sample set and a training model based on the AI ​​big model, use the training sample set to train the training model, and use the trained training model as the green electricity real-time trading model; Each transaction node is used to input the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green electricity node into the green electricity real-time transaction model to obtain the optimal transaction combination and quotation plan for each load node.

[0026] In view of the problem that green electricity cannot participate in real-time electricity trading due to its randomness, intermittency, volatility and inability to store, the present invention proposes to use a green electricity real-time trading model based on an AI big model, which can provide the most economically optimal trading combination and quotation plan in real time according to the needs of each load node. While meeting the load demand, the user cost is lowest and the green electricity revenue is guaranteed to be the best win-win situation, which promotes green electricity to fully participate in real-time market-based electricity trading and advances the market-oriented development of green electricity.

[0027] It should be understood that a trading node is a centralized single trading system or a distributed collaborative decision-making trading system. Each trading node is not restricted by national geography or other regions. Each trading node includes the geographical location of the system. , where x is the xth of all X transaction nodes.

[0028] Among them, the quality characteristics of each green power node include the location of the green power node , initial power generation cost coefficient , Green power node installed capacity , Real-time maximum output power , Maximum output power can last for a long time , Rated output power , Rated output duration and green electricity conversion efficiency .

[0029] Among them, the quality characteristics of each energy storage node include the location of the energy storage node , initial investment cost coefficient , Energy storage node installed capacity , Rated capacity of energy storage node , Initial rated power , SOC, SOH, self-consumption and energy storage conversion efficiency .

[0030] The demand characteristics of each load node include the location of the load node , Maximum required power , Duration of maximum power demand , continuous operating power , Continuous operating power duration and load conversion efficiency .

[0031] In this embodiment, the training model is trained using the training sample set, specifically including: The training model is trained using the training sample set to obtain a training model that passes the training once, so that the training model that passes the training once can be used as a green electricity real-time trading model.

[0032] In a further embodiment, after obtaining a training model that has passed training once, the method further includes: The fine-tuning network is used to strengthen and correct the training model that has passed the first training, so as to obtain the training model that has passed the second training, so as to use the training model that has passed the second training as the green electricity real-time trading model, so as to further improve the accuracy of the green electricity real-time trading model.

[0033] In order to further improve the accuracy of the green electricity real-time trading model, it is necessary to improve the accuracy of the data in the training sample set. In order to improve the accuracy of the data in the training sample set, in this embodiment, constructing the training sample set specifically includes: Use each transaction node to obtain the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, and the historical quality characteristics of each green power node; According to the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, the historical quality characteristics of each green power node, and each transaction node and the preset negotiation transaction strategy, the historical optimal transaction combination and quotation plan for each load node are obtained; A training sample set is constructed based on the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, the historical quality characteristics of each green power node, and each trading node and the corresponding historical optimal trading combination and quotation scheme.

[0034] The preset negotiation and transaction strategies specifically include: According to the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, and the historical quality characteristics of each green electricity node, the demand characteristic set of each load node, the quality characteristic set of each energy storage node, and the historical quality characteristic set of each green electricity node are obtained respectively; According to the historical demand feature set of each load node, a load node demand data set is obtained; according to the historical quality feature set of each energy storage node, a storage transaction form data set is obtained; according to the historical quality feature set of each green electricity node, a green electricity transaction form data set is obtained; and according to the load node demand data set, the storage transaction form data set, and the green electricity transaction form data set, a transaction cost data set is obtained; Establish a load transaction operation unit cost set, an energy storage transaction dispatch unit cost set, and a green electricity transaction dispatch unit cost set; obtain a total transaction cost set based on the transaction cost data set, the load transaction operation unit cost set, the energy storage transaction dispatch unit cost set, and the green electricity transaction dispatch unit cost set; The total costs of each load node in the transaction total cost set at each transaction node are prioritized and the lowest total cost is taken as the final cost of each load node; Each load node and the transaction nodes, green electricity nodes and energy storage nodes corresponding to its final cost are combined to form the historical optimal transaction combination of each load node, and the transaction nodes corresponding to the final cost of each load node are used to make quotations according to the preset quotation strategy.

[0035] Among them, the quality feature set of each green power node is ; In the formula, i represents the i-th node in each green power node; ; In the formula, represents the unit cost of green electricity generation at the i-th green electricity node, represents the initial power generation cost coefficient, Indicates the age of green power installed capacity.

[0036] Among them, the quality feature set of each energy storage node is , k represents the kth of all K energy storage nodes, represents the unit cost of energy storage, Indicates the dischargeable capacity of energy storage; represents the rated power of energy storage. Among them, the unit cost of energy storage is ; In the formula, represents the initial investment cost coefficient, Indicates the installed age of energy storage.

[0037] Among them, the energy storage discharge capacity is ; In the formula, Represents the self-consumption of energy storage. , represents the conversion efficiency of the kth energy storage node, It is the correction factor of energy storage life.

[0038] Among them, the demand characteristic set of each load node is , where j is the jth load node among all N load nodes, Indicates the maximum power demand, Indicates the continuous operating power requirement, Indicates the duration of the maximum power demand. Indicates the duration of continuous operation power demand. , where represents the maximum power demand of the jth load node, represents the maximum required power of the jth load node, represents the load conversion efficiency of the jth load node. , where represents the continuous operation power demand of the jth load node, Represents the continuous operation required power of the jth load node.

[0039] Among them, taking any load node as the benchmark, according to the physical distance between any transaction node around load node j and load node j, the transaction operation unit cost between the transaction node and the jth load node is obtained; wherein the transaction operation unit cost includes the operating cost of the load participating in the transaction and the line transmission cost between the two; according to the transaction operation unit cost between each load node and any transaction node, the load transaction operation unit cost set is obtained. Among them, ; In the formula, represents the set of unit costs of load trading operation, It represents the unit cost of load trading operation between the j-th load node and the x-th trading node.

[0040] Based on the load transaction operation unit cost set, taking any x-th transaction node as the benchmark, the transaction dispatch unit cost between the k-th energy storage node and the x-th transaction node is obtained; among which, the transaction dispatch unit cost between the k-th energy storage node and the x-th transaction node includes the line transmission cost during the period and the transaction pair energy storage operation cost; according to the transaction dispatch unit cost between the k-th energy storage node and the x-th transaction node, the energy storage transaction dispatch unit cost set is obtained. Among which, ; In the formula, represents the unit cost set of energy storage transaction scheduling, Represents the unit cost of energy storage transaction scheduling between the kth energy storage node and the xth transaction node.

[0041] Based on the load trading operation unit cost set, taking any x-th trading node as the benchmark, the green electricity trading dispatch unit cost between the i-th green electricity node and the x-th trading node is obtained; among which, the green electricity trading dispatch unit cost includes the line transmission cost during the period and the energy storage operation cost of the transaction pair; according to the green electricity trading dispatch unit cost between the i-th green electricity node and the x-th trading node, the green electricity trading dispatch unit cost set is obtained. Among which, ; In the formula, represents the unit cost set of green power trading scheduling, Represents the unit cost of green electricity trading scheduling between the i-th green electricity node and the x-th trading node.

[0042] Specifically, the load node demand data set includes the demand data of each load node participating in the transaction. Specifically, the load node demand data set is ; In the formula, represents the load node demand data set, j represents the jth load node, Indicates the maximum power demand, Indicates the continuous operating power requirement, Indicates the duration of the maximum power demand. Indicates the duration of continuous operating power demand.

[0043] Since the i-th green power node participates in the transaction in two ways, including power Sustainable uptime Or less than the rated power Minimum continuous operation time Therefore, the green electricity transaction form data set includes the green electricity transaction form data that each green electricity node can respond to the load node demand. Specifically, the green electricity transaction form data set is ; In the formula, Represents the data set of green electricity transaction form, represents the unit cost of green electricity generation, Indicates the real-time maximum output power. Indicates the duration of maximum output power. Indicates the rated output power, Indicates rated output duration, Indicates green electricity conversion efficiency.

[0044] The energy storage transaction form data set includes energy storage transaction form data that each energy storage node can respond to the discharge demand of each load node or the charging demand of each green power node. Specifically, the energy storage transaction form data set is ; In the formula, k represents the kth energy storage node among all K energy storage nodes, represents the unit cost of energy storage, Indicates the dischargeable capacity involved in the transaction, Indicates the chargeable capacity. Represents the energy storage rated power. Among them, the energy storage rated power is ; In the formula, Indicates the initial rated power of energy storage, represents the energy storage conversion efficiency, Represents the energy storage life correction factor. Among them, the discharge capacity of the kth energy storage node The corresponding discharge power is less than Any power of; where the rechargeable capacity is .

[0045] In order to promote green electricity to fully participate in real-time market-based electricity trading, in this embodiment, the transaction cost data set includes a first transaction cost data set and a second transaction cost data set; The transaction cost data set is obtained based on the load node demand data set, the energy storage transaction form data set, and the green electricity transaction form data set, including: Divide the demand data of each load node in the load node demand data set into first demand or second demand; For each trading node, randomly extract the first demands of A1 load nodes from the load node demand data set, and combine the first demands of the A1 load nodes to form a first demand set; where, A1 < M; According to the first preset condition, from the green power trading form data set Filter out the quality feature set of green power nodes that meet the first demand set; where, the number of the quality feature sets of the filtered green power nodes is B1, B1 < N; and combine the quality feature sets of the B1 filtered green power nodes to form a first green power data set ; where, the first preset condition is and ; According to the first green power data set and the demand feature set of the load nodes in the first demand set it satisfies, obtain the first two-dimensional power generation tradable data set between each load node and each green power node trading on this trading node; Judge whether B1 is greater than or equal to A1; if so, combine the first two-dimensional power generation tradable data sets of all trading nodes to form a first trading cost data set; For each trading node, randomly extract the second demands of A2 load nodes from the load node demand data set, and combine the second demands of the A2 load nodes to form a second demand set; where, A2 < M; According to the second preset condition, from the green power trading form data set Filter out the quality feature set of green power nodes that meet the second demand set; where, the number of the quality feature sets of the green power nodes that meet the second demand set is B2, B2 < N; and combine the quality feature sets of the B2 filtered green power nodes to form a second green power data set ; where, the second preset condition is ; According to the second green power data set and the demand feature set of the load nodes in the second demand set it satisfies, obtain the second two-dimensional power generation tradable data set between each load node and each green power node trading on this trading node; judge whether B2 is greater than or equal to A2; if so, combine the second two-dimensional power generation tradable data sets of all trading nodes to form a second trading cost data set.

[0046] Among them, the first two-dimensional power generation tradable data set is ; represents the first two-dimensional power generation tradable data set between the jth load node and the ith green power node; the second two-dimensional power generation tradable data set is ; Represents the second two-dimensional power generation tradable data set between the j-th load node and the i-th green power node.

[0047] In a further embodiment, when determining whether B1 is greater than or equal to A1, if not, then according to the third preset condition, the energy storage transaction form data set is selected. The quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the first demand set are screened out; wherein the number of quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the first demand set is C1; Determine whether the sum of B1 and C1 is less than A1; if so, the trading node does not respond to the transactions of the remaining A1-B1-C1 load nodes in the first demand set; if not, based on the quality feature set of the screened C1 energy storage nodes, obtain the first two-dimensional energy storage compensation data set between each load node and each energy storage node that conducts energy storage compensation transactions on the trading node; wherein the third preset condition is ; Combining the first two-dimensional power generation tradable data set and the first two-dimensional energy storage compensation data set on each transaction node to obtain a first transaction cost set; Similarly, when determining whether B2 is greater than or equal to A2, if it is determined that B2 is less than A2, then according to the fourth preset condition, the energy storage transaction form data set The quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the second demand set are screened out; wherein the number of quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the second demand set is C2; Determine whether the sum of B2 and C3 is less than A2; if so, the transaction node does not respond to the remaining A2-B2-C2 load nodes in the second demand set; if not, based on the quality feature set of the screened C2 energy storage nodes, obtain the second two-dimensional energy storage compensation data set between each load node and each energy storage node that conducts energy storage compensation transactions on the transaction node; the fourth preset condition is and ; and combining the second two-dimensional power generation tradable data set and the second two-dimensional energy storage compensation data set on each transaction node to obtain a second transaction cost set.

[0048] Among them, the first two-dimensional energy storage compensation data set is ; In the formula, represents the first two-dimensional energy storage compensation data set; where r is the node without i in the load node, and k is any energy storage node. The second two-dimensional energy storage compensation data set is ; In the formula, represents the second two-dimensional energy storage compensation data set.

[0049] Therefore, in this embodiment, according to the transaction cost data set, the load transaction operation unit cost set, the energy storage transaction dispatch unit cost set, and the green power transaction dispatch unit cost set, a total transaction cost set is obtained, which specifically includes: Obtaining a first transaction total cost set according to the first transaction cost data set, the load transaction operation unit cost set, the energy storage transaction dispatch unit cost set, and the green electricity transaction dispatch unit cost set; The second transaction total cost set is obtained according to the second transaction cost data set, the load transaction operation unit cost set, the energy storage transaction scheduling unit cost set, and the green electricity transaction scheduling unit cost set.

[0050] In a further embodiment, the total costs of each load node in the transaction total cost set at each transaction node are prioritized respectively, and the lowest total cost is used as the final cost of each load node, specifically including: Prioritize the total costs of each load node in the first transaction total cost set and the second transaction total cost set at each transaction node respectively; take the lowest total cost of each load node in the first transaction total cost set as the first total cost of each load node; take the lowest total cost of each load node in the second transaction total cost set as the second total cost of each load node; and based on the first total cost and the second total cost of each load node, take the lowest total cost between the first total cost and the second total cost as the final cost of each load node.

[0051] In this embodiment, in the preset bidding strategy, the trading node in each historical optimal trading combination uses the demand of its load node, the power generation price of the green power node and the discharge of the energy storage node to bid; The unit price of demand quotation for each load node is the sum of the final cost and the preset first profit coefficient; the unit price of power generation sale price for each green electricity node is the sum of the unit price of power generation and the preset second profit coefficient; the unit price of discharge quotation for each energy storage node is the sum of the unit price of energy storage and the preset third profit coefficient; the first profit coefficient is greater than the sum of the second profit coefficient and the third profit coefficient.

[0052] Among them, the first transaction total cost set is ; In the formula, It represents the unit cost of the load transaction between each load node and the xth transaction node. represents the unit cost of energy storage transaction scheduling between each energy storage node and the xth transaction node, Represents the unit cost of green electricity transaction scheduling between each green electricity node and the xth transaction node.

[0053] Among them, the second total transaction cost data set is .

[0054] To further ensure the best green power revenue and promote the full participation of green power in real-time market-based power trading, in a further embodiment, if B1 + B2 < N and K > C1 + C2 and the rechargeable capacity of the remaining K - C1 - C2 energy storage nodes > 0, then the electricity generated by the remaining N - B1 - B2 green power nodes is stored in the remaining K - C1 - C2 energy storage nodes, and the green power charging demand scheduling cost data set for the remaining N - B1 - B2 green power nodes to schedule energy storage to store electrical energy at each trading node is obtained; According to the energy storage trading scheduling unit cost set, the green power trading scheduling unit cost set, and the green power charging demand scheduling cost data set, the green power charging demand total cost set is obtained; among them, the green power charging demand total cost set includes the total green power charging demand cost for each green power node to schedule energy storage to store electrical energy at each trading node; Respectively perform priority sorting on the total green power charging demand cost for each green power node to schedule energy storage to store electrical energy at each trading node in the green power charging demand scheduling cost data set, and select the lowest total green power charging demand cost as the final total green power charging demand cost for each green power node to schedule energy storage to store electrical energy, and use the energy storage node corresponding to the final total green power charging demand cost as the charging response for each green power node; Use the trading node corresponding to the final green power charging demand cost to quote the unit price of the charging demand for each green power node according to the preset quoting strategy, and quote the unit price of the charging energy storage scheduling for the charging response of each green power node.

[0055] Among them, the cost data set for the remaining N - B1 - B2 green power nodes to schedule energy storage to store electrical energy at each trading node is .

[0056] Specifically, in the preset quoting strategy, the unit price of the charging demand for each green power node is the sum of the final cost for each green power node to schedule energy storage to store electrical energy and the preset first profit coefficient; the unit price quote of the charging energy storage scheduling for each energy storage node is the sum of the unit price of the charging demand for each energy storage node and the preset second profit coefficient.

[0057] In a second aspect, the present invention also proposes a green power real-time trading system based on an AI large model. The main bodies participating in the green power real-time trading include N green power nodes, K energy storage nodes, M load nodes, and X trading nodes. Each trading node is communicatively connected to N green power nodes, K energy storage nodes, and M load nodes; the system includes: An acquisition module is used to use each transaction node to respectively acquire the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green power node; Model building module, used to build training models based on AI big models; A training set construction module is used to construct a training sample set; The training module is used to train the training model using the training sample set, and use the trained model as the green electricity real-time trading model.

[0058] The processing module stores a green electricity real-time trading model, which is used to use each trading node to input the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green electricity node into the green electricity real-time trading model to obtain the optimal trading combination and quotation plan for each load node.

[0059] It should be understood that the method for constructing a training sample set by the training set construction module in this embodiment is the same as the construction process of the training sample set described in the first aspect, and will not be described in detail here.

[0060] In a third aspect, the present invention further proposes a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the green electricity real-time trading method based on the AI ​​big model as described in any one of the first aspects.

[0061] In a fourth aspect, the present invention further proposes an electronic device, comprising: a processor and a memory, the memory being used to store one or more programs; when the one or more programs are executed by the processor, the green electricity real-time trading method based on the AI ​​big model as described in any one of the first aspects is implemented.

[0062] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A green electricity real-time trading method based on AI big model, characterized in that: The entities participating in the real-time green electricity transaction include N green electricity nodes, K energy storage nodes, M load nodes and X transaction nodes. Each transaction node is connected to N green electricity nodes, K energy storage nodes and M load nodes. The method comprises: Use each transaction node to obtain the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green power node; Construct a training sample set and a training model based on the AI ​​big model, use the training sample set to train the training model, and use the trained training model as the green electricity real-time trading model; Each transaction node is used to input the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green electricity node into the green electricity real-time transaction model to obtain the optimal transaction combination and quotation plan for each load node.

2. The green electricity real-time trading method based on AI big model according to claim 1 is characterized in that: Construct a training sample set, including: Use each transaction node to obtain the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, and the historical quality characteristics of each green power node; According to the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, the historical quality characteristics of each green power node, and each transaction node and the preset negotiation transaction strategy, the historical optimal transaction combination and quotation plan for each load node are obtained; A training sample set is constructed based on the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, the historical quality characteristics of each green power node, and each trading node and the corresponding historical optimal trading combination and quotation scheme.

3. The green electricity real-time trading method based on AI big model according to claim 2 is characterized in that: The preset negotiation and transaction strategies include: According to the historical demand characteristics of each load node, the historical quality characteristics of each energy storage node, and the historical quality characteristics of each green electricity node, the demand characteristic set of each load node, the quality characteristic set of each energy storage node, and the historical quality characteristic set of each green electricity node are obtained respectively; According to the historical demand feature set of each load node, a load node demand data set is obtained; wherein the load node demand data set includes demand data of each load node participating in the transaction; According to the historical quality feature set of each green power node, a green power transaction form data set is obtained; wherein the green power transaction form data set includes green power transaction form data that each green power node can respond to the needs of each load node; According to the historical quality feature set of each energy storage node, a set of energy storage transaction form data is obtained; wherein the set of energy storage transaction form data includes energy storage transaction form data that each energy storage node can respond to the discharge demand of each load node or the charging demand of each green power node; Obtain a transaction cost data set based on a load node demand data set, an energy storage transaction form data set, and a green electricity transaction form data set; Establish a load trading operation unit cost set, a storage trading dispatch unit cost set, and a green power trading dispatch unit cost set; According to the transaction cost data set, the load transaction operation unit cost set, the energy storage transaction dispatch unit cost set, and the green electricity transaction dispatch unit cost set, a total transaction cost set is obtained; wherein the total transaction cost set includes the total transaction cost of each load node when participating in the transaction at each transaction node; The total transaction costs of each load node in the total transaction cost set at each transaction node are prioritized respectively, and the lowest total cost is taken as the final cost of each load node; Each load node and the transaction nodes, green electricity nodes and energy storage nodes corresponding to its final cost are combined to form the historical optimal transaction combination of each load node, and the transaction nodes corresponding to the final cost of each load node are used to make quotations according to the preset quotation strategy.

4. The green electricity real-time trading method based on AI big model according to claim 3 is characterized in that: The transaction cost data set includes a first transaction cost data set and a second transaction cost data set; wherein the transaction cost data set is obtained according to the load node demand data set, the energy storage transaction form data set, and the green electricity transaction form data set, and specifically includes: Classifying the demand data of each load node in the load node demand data set into a first demand or a second demand; For each transaction node, the first demands of A1 load nodes are randomly extracted from the load node demand data set, and the first demands of A1 load nodes are combined to form a first demand set; wherein A1 <M; According to the first preset condition, screen out the quality feature set of green power nodes that meet the first demand set from the green power trading form data set; among them, the number of the screened quality feature sets of green power nodes is B1, and B1 < N; and combine the screened quality feature sets of B1 green power nodes to form the first green power data set; where the first preset condition is and ; in the formula represents the maximum power demand of the jth load node, ; represents the duration of the maximum power demand of the jth load node, represents the real-time maximum output power of the ith green power node, ; represents the green power conversion efficiency of the ith green power node; represents the sustainable time of the maximum output power of the ith green power node And according to the first green electricity data set and the demand feature set of the load nodes in the first demand set satisfied by the first green electricity data set, a first two-dimensional power generation tradable data set between each load node and each green electricity node traded on the transaction node is obtained; Determine whether B1 is greater than or equal to A1; if so, combine the first two-dimensional power generation tradable data sets of all transaction nodes to form a first transaction cost data set; For each transaction node, the second demands of A2 load nodes are randomly extracted from the load node demand data set, and the second demands of A2 load nodes are combined to form a second demand set; wherein A2 <M; According to the second preset condition, screen out the quality feature set of green power nodes that meet the second requirement set from the green power trading form data set; wherein, the number of quality feature sets of green power nodes that meet the second requirement set is B2, and B2 < N; and combine the quality feature sets of the B2 green power nodes screened out to form a second green power data set; wherein, the second preset condition is and ; where represents the continuous operating power demand of the j-th load node, ; represents the rated output power of the i-th green power node, represents the continuous operating power duration of the j-th load node, represents the rated output duration of the i-th green power node; According to the second green electricity data set and the demand feature set of the load node of the second demand set satisfied by the second green electricity data set, a second two-dimensional power generation tradable data set between each load node and each green electricity node that performs transactions on the transaction node is obtained; Determine whether B2 is greater than or equal to A2; if so, combine the second two-dimensional power generation tradable data sets of all transaction nodes to form a second transaction cost data set.

5. The green electricity real-time trading method based on AI big model according to claim 4 is characterized in that: When judging whether B1 is greater than or equal to A1, if not, the quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the first demand set are screened out from the energy storage transaction form data set according to the third preset condition; wherein the number of the quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the first demand set is C1; Determine whether the sum of B1 and C1 is less than A1; if not, obtain the first two-dimensional energy storage compensation data set between each load node and each energy storage node that conducts energy storage compensation transactions on the transaction node based on the quality feature set of the screened C1 energy storage nodes; wherein the third preset condition is and ; In the formula, represents the energy storage rated power of the kth energy storage node, represents the dischargeable capacity of the kth energy storage node; Combining the first two-dimensional power generation tradable data set and the first two-dimensional energy storage compensation data set on each transaction node to obtain a first transaction cost set; When determining whether B2 is greater than or equal to A2, if it is determined that B2 is less than A2, then according to the fourth preset condition, a quality feature set of energy storage nodes that meet the demand feature sets of the remaining load nodes in the second demand set is screened out from the energy storage transaction form data set; among them, the number of quality feature sets of energy storage nodes that meet the demand feature sets of the remaining load nodes in the second demand set is C2; Determine whether the sum of B2 and C3 is less than A2; if not, obtain the second two-dimensional energy storage compensation data set between each load node and each energy storage node that conducts energy storage compensation transactions on the transaction node based on the quality feature set of the screened C2 energy storage nodes; the fourth preset condition is and ; And combine the second two-dimensional power generation tradable data set and the second two-dimensional energy storage compensation data set on each trading node to obtain a second transaction cost set.

6. The green electricity real-time trading method based on AI big model according to claim 4 or 5 is characterized in that: According to the transaction cost data set, the load trading operation unit cost set, the energy storage transaction scheduling unit cost set, and the green power transaction scheduling unit cost set, obtain a total transaction cost set, specifically including: According to the first transaction cost data set, the load trading operation unit cost set, the energy storage transaction scheduling unit cost set, and the green power transaction scheduling unit cost set, obtain a first total transaction cost set; According to the second transaction cost data set, the load trading operation unit cost set, the energy storage transaction scheduling unit cost set, and the green power transaction scheduling unit cost set, obtain a second total transaction cost set; Perform priority sorting on the total costs of each load node at each trading node in the total transaction cost set, and take the lowest total cost as the final cost of each load node, specifically including: Perform priority sorting on the total costs of each load node at each trading node in the first total transaction cost set and the second total transaction cost set respectively; Take the lowest total cost of each load node in the first total transaction cost set as the first total cost of each load node; Take the lowest total cost of each load node in the second total transaction cost set as the second total cost of each load node; According to the first total cost and the second total cost of each load node, take the lowest total cost among the first total cost and the second total cost as the final cost of each load node.

7. The green electricity real-time trading method based on AI big model according to claim 6 is characterized in that: After forming the first transaction cost data set and the second transaction cost data set, it also includes: If B1 + B2 < N and K > C1 + C2 and the rechargeable capacity of the remaining K - C1 - C2 energy storage nodes > 0, then store the power generated by the remaining N - B1 - B2 green power nodes into the remaining K - C1 - C2 energy storage nodes, and obtain a green power charging demand scheduling cost data set for the remaining N - B1 - B2 green power nodes to schedule energy storage to store electrical energy at each trading node; According to the energy storage transaction scheduling unit cost set, the green power transaction scheduling unit cost set, and the green power charging demand scheduling cost data set, obtain a total green power charging demand cost set; among them, the total green power charging demand cost set includes the total green power charging demand cost for each green power node to schedule energy storage to store electrical energy at each trading node; Prioritize the total cost of green electricity charging demand that each green electricity node in the green electricity charging demand total cost data set needs to dispatch energy storage to store electric energy at each transaction node, and select the lowest total cost of green electricity charging demand as the final total cost of green electricity charging demand that each green electricity node needs to dispatch energy storage to store electric energy, and use the energy storage node corresponding to the final total cost of green electricity charging demand as the charging response of each green electricity node; The transaction node corresponding to the final green electricity charging demand cost is used to quote the charging demand quotation unit price of each green electricity node according to the preset quotation strategy, and the charging energy storage scheduling unit price of the energy storage node that responds to the charging of each green electricity node is quoted.

8. A green electricity real-time trading system based on AI big model, characterized in that: The entities participating in the real-time green electricity transaction include N green electricity nodes, K energy storage nodes, M load nodes and X transaction nodes. Each transaction node is connected to N green electricity nodes, K energy storage nodes and M load nodes, including: An acquisition module is used to use each transaction node to respectively acquire the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green power node; Model building module, used to build training models based on AI big models; A training set construction module is used to construct a training sample set; A training module is used to train the training model using the training sample set, and use the trained model as a green electricity real-time trading model; The processing module stores a green electricity real-time trading model, which is used to use each trading node to input the demand characteristics of each load node, the quality characteristics of each energy storage node, and the quality characteristics of each green electricity node into the green electricity real-time trading model to obtain the optimal trading combination and quotation plan for each load node.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the green electricity real-time trading method based on the AI ​​big model as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: include: A processor and a memory, the memory is used to store one or more programs; when the one or more programs are executed by the processor, the green electricity real-time trading method based on the AI ​​big model as described in any one of claims 1-7 is implemented.

Images