Power transaction risk management method and system and medium

Abstract

The application discloses a power transaction risk management and control method and system and a medium, relates to the technical field of risk management and control, and comprises the following steps: acquiring a power transaction request, performing request analysis, and establishing a request analysis result; after updating a seller information base, calling transaction information analysis results in the request analysis result, performing multi-dimensional adaptive matching of the seller information base, and establishing a multi-dimensional adaptive matching result; using the request subject analysis result and the updated seller information base to establish the behavior track of the request subject and the seller subject; calling a basic risk identification network, inputting the multi-dimensional adaptive matching result into the basic risk identification network, and establishing a two-way transaction risk; using the behavior track to perform risk compensation on the two-way transaction risk, and generating a two-way transaction risk early warning. The application solves the technical problem that there is a lack of effective risk identification and early warning mechanism in the power transaction process in the prior art, and achieves the technical effects of optimizing transaction risk management and control and improving transaction security.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk control, and particularly relates to a power transaction risk control method, a system and a medium. BACKGROUND

[0002] The power transaction process involves multiple transaction subjects and complex transaction information, which contains different time, demand capacity and price factors. In this process, the identification and prediction of transaction risks are particularly critical. However, due to the lack of systematic risk identification means, it is difficult to monitor the possible performance risks, liquidity risks, abnormal bidding risks and other risks in the transaction in real time, especially under the conditions of multi-party transactions and complex transaction behaviors, the effective identification and early warning ability of risks is insufficient, which leads to the inability to take timely measures when potential risks occur. This makes the security and stability of power transaction face challenges. SUMMARY

[0003] The present application provides a power transaction risk control method, system and medium, which is used to solve the technical problem of lack of effective risk identification and early warning mechanism in the power transaction process in the prior art.

[0004] In view of the above problems, the present application provides a power transaction risk control method, system and medium.

[0005] The first aspect of the present application provides a power transaction risk control method, which comprises:

[0006] Accessing a power transaction request, performing request analysis on the power transaction request, and establishing a request analysis result, the request analysis including transaction information analysis and request subject analysis; after updating the seller information database, calling the transaction information analysis result in the request analysis result, performing multi-dimensional adaptive matching of the seller information database, and establishing a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching includes independent adaptive matching and combined adaptive matching; using the request subject analysis result and the updated seller information database to establish the behavior trajectory of the request subject and the seller subject; calling a basic risk identification network, inputting the multi-dimensional adaptive matching result into the basic risk identification network, and establishing a two-way transaction risk; using the behavior trajectory to compensate the two-way transaction risk, and generating a two-way transaction risk early warning.

[0007] The second aspect of the present application provides a power transaction risk control system, which comprises:

[0008] The request analysis module is configured to acquire an accessed power transaction request, perform request analysis on the power transaction request, and establish a request analysis result, wherein the request analysis comprises transaction information analysis and request subject analysis; the adaptive matching module is configured to, after updating the seller information database, call the transaction information analysis result in the request analysis result, perform multi-dimensional adaptive matching of the seller information database, and establish a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching comprises independent adaptive matching and combined adaptive matching; the behavior trajectory establishment module is configured to establish the behavior trajectory of the request subject and the seller subject by using the request subject analysis result and the updated seller information database; the risk identification module is configured to call a basic risk identification network, input the multi-dimensional adaptive matching result into the basic risk identification network, and establish a bidirectional transaction risk; and the risk compensation module is configured to perform risk compensation on the bidirectional transaction risk by using the behavior trajectory, and generate a bidirectional transaction risk warning.

[0009] In a third aspect, the present application provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the power transaction risk management method provided by the present application.

[0010] The one or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0011] The present application acquires an accessed power transaction request, performs request analysis on the power transaction request, and establishes a request analysis result, wherein the request analysis comprises transaction information analysis and request subject analysis; after updating the seller information database, the adaptive matching module is configured to call the transaction information analysis result in the request analysis result, perform multi-dimensional adaptive matching of the seller information database, and establish a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching comprises independent adaptive matching and combined adaptive matching; the behavior trajectory establishment module is configured to establish the behavior trajectory of the request subject and the seller subject by using the request subject analysis result and the updated seller information database; the risk identification module is configured to call a basic risk identification network, input the multi-dimensional adaptive matching result into the basic risk identification network, and establish a bidirectional transaction risk; and the risk compensation module is configured to perform risk compensation on the bidirectional transaction risk by using the behavior trajectory, and generate a bidirectional transaction risk warning. The present application solves the technical problem of lack of effective risk identification and early warning mechanism in the power transaction process in the prior art, and achieves the technical effects of optimizing transaction risk management and improving transaction security by establishing a multi-dimensional adaptive matching and risk compensation mechanism. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without any creative effort based on these drawings.

[0013] Figure 1 A power transaction risk management method flowchart is provided for the embodiments of the present application.

[0014] Figure 2 A power transaction risk management system structure diagram is provided for the embodiments of the present application.

[0015] Legend: request analysis module 11, adaptive matching module 12, behavior trajectory establishment module 13, risk identification module 14, risk compensation module 15. DETAILED DESCRIPTION

[0016] The present application provides a power transaction risk management method, system and medium, which solves the technical problem of lack of effective risk identification and early warning mechanism in the power transaction process in the prior art, establishes a multi-dimensional adaptive matching and risk compensation mechanism, and achieves the technical effects of optimizing transaction risk management and improving transaction security.

[0017] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0018] It should be noted that any variation of the terms "comprise" and "have" is intended to cover non-exclusive inclusion, for example, a process, method, system, product or server comprising a series of steps or units does not have to be limited to those clearly listed steps or units, but can include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0019] Embodiment one, as shown in the present application provides a power transaction risk management method, which comprises: Figure 1

[0020] Step S100: acquiring an accessed power transaction request, performing request analysis on the power transaction request, establishing a request analysis result, the request analysis including transaction information analysis and request subject analysis.

[0021] In the embodiments of the present application, first, transaction request data from the power market is acquired, these requests including basic information such as transaction power, transaction price, demand time, etc. Next, the power transaction request is analyzed, wherein the request analysis includes transaction information analysis and request subject analysis. When performing transaction information analysis, power demand, transaction time period, etc. information is extracted from the power transaction request. When performing request subject analysis, the identity of the buyer, historical transaction records, etc. information is parsed from the power transaction request.​

[0022] The extracted information is summarized by transaction information analysis and request subject analysis to obtain a request analysis result.

[0023] Step S200: After updating the seller information database, the transaction information analysis result in the request analysis result is called to perform multi-dimensional adaptive matching of the seller information database to establish a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching includes independent adaptive matching and combined adaptive matching.

[0024] In the embodiments of the present application, the seller information database is first updated, that is, the latest information of the seller is manually entered or automatically collected to ensure the accuracy and real-time performance of the seller information database. The updated information includes the latest supply capacity, historical transaction records, and power prices of the seller. The updating process includes importing these information from external data sources (such as transaction records, market reports, etc.) into the seller information database, or automatically updating these data by real-time monitoring of the performance of the seller.

[0025] Next, the transaction information analysis result in the request analysis result is called to perform multi-dimensional adaptive matching of the seller information database. In this process, the transaction information analysis data in the request analysis result is first called to extract the demand time and demand capacity of the buyer and other information as matching features. Then, according to these matching features, the multi-dimensional adaptive matching of the seller information database is performed, first the independent adaptive matching is performed, and the data in the seller information database is matched according to the demand time and capacity and other single dimensions to obtain a preliminary matching result. Then, by configuring the combined number constraint, the combined adaptive matching is performed, and the information of multiple dimensions (such as the power supply capacity, historical transaction records, etc. of the seller) is considered to generate a more accurate matching result. Finally, the multi-dimensional adaptive matching result is established by integrating the independent adaptive matching result and the combined adaptive matching result.

[0026] Further, the method provided by the embodiments of the present application further comprises:

[0027] According to the request analysis result, the demand time and demand capacity are obtained; the demand time and demand capacity are taken as associated matching features to perform independent seller adaptive matching of the seller information database to establish a first-dimensional adaptive matching result; after configuring the combined number constraint, the demand time and demand capacity are taken as associated matching features to perform combined seller adaptive matching of the seller information database to establish a second-dimensional adaptive matching result; and the multi-dimensional adaptive matching result is established according to the first-dimensional adaptive matching result and the second-dimensional adaptive matching result.

[0028] In the embodiment of the present application, the demand time and the demand capacity are first obtained from the request analysis result. The demand time refers to a specific time period in which the buyer hopes to conduct power transaction, and the demand capacity refers to the amount of power required by the buyer.

[0029] Next, the demand time and the demand capacity are taken as the associated matching features to perform independent seller adaptation matching of the seller information library. In this step, the power supply time of the seller is matched with the demand time of the buyer, and the power supply capacity of the seller is compared with the demand capacity of the buyer. That is, if the power supply time of a seller completely matches the demand time of the buyer, and the seller can provide sufficient power to meet the demand capacity, then the seller can meet the demand of the buyer. Through this matching process, a first-dimension adaptation matching result is generated. The first-dimension adaptation matching result contains information of all sellers that can simultaneously meet the demand time and the demand capacity of the buyer.

[0030] Next, the combination quantity constraint is configured, that is, the number of sellers participating in matching is limited, which is set by the buyer. Subsequently, the demand time and the demand capacity are taken as the associated matching features to perform combined seller adaptation matching of the seller information library. In this process, according to the demand time and the demand capacity, the power supply capacities of multiple sellers are searched in the seller information library, and if the power supply capacities of the multiple sellers can meet the demand time and the demand capacity of the buyer when added together, then the demand of the buyer can be met through the combination of the sellers, while it is also required to be constrained by the combination quantity constraint to ensure that the number of sellers does not exceed the combination quantity constraint. In this way, a second-dimension adaptation matching result, that is, a matching result in which multiple sellers meet the demand of the buyer, is generated.

[0031] Finally, the first-dimension adaptation matching result and the second-dimension adaptation matching result are combined to generate a multi-dimension adaptation matching result.

[0032] Step S300: The behavior trajectories of the request subject and the seller subject are established by using the request subject analysis result and the updated seller information library.

[0033] In the embodiment of the present application, the behavior features of the buyer and the seller are extracted by configuring a crawling time window by using the request subject analysis result and the updated seller information library. In this process, the behavior data of the buyer and the seller are crawled within a specified time window. These data include time-series bidding behavior features, time-series capacity supply behavior features, and time-series fund behavior features. The time-series bidding behavior features reflect the changes in the bids of the seller at different time points, the time-series capacity supply behavior features reveal the changes in the power supply capacity of the seller at different time periods, and the time-series fund behavior features reflect the performance of the buyer and the seller in fund flow. Through the extraction of these time-series behavior features, the behavior trajectories of the request subject and the seller subject are integrated and formed.

[0034] Further, the method provided by the application embodiment further comprises the following steps:

[0035] After the capture time window is configured, time sequence behavior characteristics extraction is performed in the capture time window, and the time sequence behavior characteristics extraction includes time sequence bidding behavior characteristics, time sequence capacity supply behavior characteristics, and time sequence fund behavior characteristics; and a behavior trajectory is established by using the time sequence behavior characteristics extraction result.

[0036] In the application embodiment, first, a capture time window is configured, that is, a specific time period is defined for collecting transaction data of the buying and selling parties. For example, a time window is set as every hour, every week, or adjusted according to the transaction period of the electricity market, such as from 9:00 to 12:00 in the morning of each trading day, and all electricity transaction activities occurring in this period are focused on.

[0037] Next, time sequence behavior characteristics extraction is performed in the capture time window. In this process, the captured data is processed, and key features reflecting the transaction behaviors of the buying and selling parties are extracted, and these features are arranged in chronological order. Specifically, time sequence bidding behavior characteristics, time sequence capacity supply behavior characteristics, and time sequence fund behavior characteristics are extracted.

[0038] In the extraction of time sequence bidding behavior characteristics, the change of the bid of the seller in the capture time window is analyzed to extract the characteristics. For example, if the power demand of the buyer increases, the seller may increase the bid in a specific period, or lower the bid when the power supply is surplus. Assuming that a seller adjusts the price from 200 yuan / MWh to 250 yuan / MWh in the time period from 9:00 to 10:00, the bid change data in this period is extracted as the time sequence bidding behavior characteristics.

[0039] In the extraction of time sequence capacity supply behavior characteristics, the power capacity provided by the seller in the capture time window is analyzed to describe the power supply capacity of the seller in different time periods. For example, in a period with high demand, such as from 16:00 to 18:00, a seller may increase the power supply from 100 MW to 150 MW; and in a period with low demand, the seller may reduce the power supply. In this way, the change of the power supply capacity of the seller in different time periods is extracted, and the time sequence capacity supply behavior characteristics are formed.

[0040] Then, the time sequence fund behavior characteristics are extracted, which reflect the fund flow of the buyer and the seller in the transaction. In the extraction of the time sequence fund behavior characteristics, whether the buyer pays the payment on time and whether the seller receives the payment on time are tracked. In some cases, the buyer may delay payment, and the seller may receive payment after the agreed time. Assuming that the buyer pays the payment that should have been paid at 9:00 at 10:00, this information is recorded as the fund behavior characteristics.

[0041] Finally, the extraction results of the time-series quote behavior feature, the time-series capacity supply behavior feature, and the time-series fund behavior feature are integrated into a behavior trajectory. The behavior trajectory shows the interaction mode of the buyer and the seller within a specific time window. For example, a comprehensive data trajectory of the seller's quote change, the power supply adjustment, and the fund flow within a time period (e.g., 9:00 to 10:00) is constructed.

[0042] Further, the method provided by the application embodiment further comprises:

[0043] configuring an external event correlator; taking the time-series behavior feature extraction result as a matching feature, performing event correlation matching based on the external event correlator, establishing an event-sensitive behavior trajectory; and using the event-sensitive behavior trajectory to compensate and correct the behavior trajectory.

[0044] In the application embodiment, an external event correlator is first configured to ensure that external events related to power trading can be captured in real time, and these external events are associated with the trading behavior of the buyer and the seller. Specifically, the external event correlator is implemented through pre-set rule configuration. It is defined which external events will affect power trading, such as weather changes (e.g., high-temperature weather leading to increased demand for electricity), power supply interruption (e.g., shutdown of a power plant), etc. Through these pre-set rules, the data interface is connected with external event data sources (such as weather information, market demand data, etc.) to capture these external events in real time and mark them as event data. When an external event occurs, the event data is matched with the power trading data in the corresponding time period according to the set time window, ensuring that the external event can be associated with the power trading behavior synchronously.

[0045] Next, the time-series behavior feature extraction result is taken as a matching feature, and event correlation matching based on the external event correlator is performed. In this step, the previously extracted time-series behavior feature data, such as the time-series quote behavior feature, the time-series capacity supply behavior feature, and the time-series fund behavior feature, is used as the basic feature for external event correlation. Through the external event correlator, the external event (such as weather change) is matched with the behavior feature data to reveal how the external event affects the behavior of the buyer and the seller. For example, high-temperature weather may lead to a sharp increase in demand for electricity, thereby prompting the seller to increase the electricity price and increase the power supply, while the buyer may reduce demand or pay more fees. Matching these changes with the external event generates an event-sensitive behavior trajectory.

[0046] Finally, the event-sensitive behavior trajectory is used to compensate and correct the behavior trajectory. The goal of this process is to correct the abnormal behavior fluctuations caused by external events, so that the final behavior trajectory is more in line with the actual market operation rules. For example, suppose that during a certain period, due to external events (such as extremely high temperature weather), the seller's quotation fluctuates abnormally, and the price is much higher than the normal market fluctuation range, resulting in abnormal behavior trajectory. At this time, compensation and correction are carried out. The compensation process includes adjusting the seller's quotation behavior characteristics according to the normal fluctuation range of the market, and removing the abnormally high peak caused by external events. For example, if the seller raises the electricity price from 200 yuan / MWh to 400 yuan / MWh during the extreme weather, but the fluctuation exceeds the normal range of the market, identify this fluctuation and adjust it to a reasonable price range (such as adjust to 250 yuan / MWh) through the compensation and correction mechanism, so that the final behavior trajectory is more smooth and true.

[0047] Step S400: calling the basic risk identification network, inputting the multi-dimensional adaptive matching result into the basic risk identification network, and establishing a two-way transaction risk.

[0048] In the embodiments of the present application, first, the updated seller information base and the request subject analysis result are subjected to static feature extraction, and the key feature data of the seller and the buyer are extracted. These data include the power supply capacity, quotation history, credit status and other information of the seller, and the demand time, demand capacity and other features of the buyer. Then the extracted static feature data is sent to the basic risk identification network, which is used to analyze and initialize the risk identification database.

[0049] Next, the multi-dimensional adaptive matching result is input into the basic risk identification network, and the matching result is combined and adapted by activating the discrimination layer to determine whether the matching of the buyer and the seller meets the independent adaptation condition. If the matching result is independent adaptation, the corresponding seller static feature, independent adaptation result and request subject static feature are called, and these data are transmitted to the first two-way abnormality identification layer for analysis to evaluate the potential transaction risk between the buyer and the seller, and finally the two-way transaction risk is output.

[0050] When the discrimination result is a combined adaptation result, the power supply capacity of multiple sellers is combined according to the time period and the allocation capacity based on the static features to form a seller allocation combination set that meets the demand of the buyer. On this basis, combined performance coordination features are established, including combined performance coordination covariance, time splicing continuity, performance redundancy and combined performance risk weight, which are used to evaluate the performance ability, stability and potential risk of the seller combination. Then these coordination features, the number of seller combinations and the buyer static features are sent to the second two-way abnormality identification layer for risk analysis, and finally the two-way transaction risk is output.

[0051] Further, the method provided by the application embodiment further comprises:

[0052] After the updated seller information base and the request subject analysis result are subjected to static feature extraction, the updated seller information base and the request subject analysis result are sent to the basic risk identification network for network analysis database initialization. After the multi-dimensional adaptive matching result is input to the basic risk identification network, a discrimination layer is activated to perform combined adaptive discrimination. If the discrimination result is an independent adaptive result, the seller static feature corresponding to the independent adaptive result is called, and the seller static feature, the independent adaptive result, and a request subject static feature are sent to a first bidirectional anomaly identification layer to output a bidirectional transaction risk.

[0053] In the application embodiment, first, the updated seller information base and the request subject analysis result are subjected to static feature extraction, i.e., static features such as the power supply capacity of a seller, historical transaction records, and bidding history are extracted from the seller information base, and demand information such as demand time and demand capacity is extracted from the request subject analysis result. For example, the power supply capacity of a seller is extracted as a static feature, and the demand power of a buyer is extracted from the request subject analysis result as a static feature.

[0054] After the static feature extraction is completed, the feature data is sent to the basic risk identification network, and network analysis database initialization is performed in the network. In this process, the extracted static features are stored in the database according to a predetermined format and structure. The initialization process involves storing the static feature data of the seller and the buyer in a standardized format and establishing an index for quick query and retrieval. For example, the power supply capacity of a seller is divided into different capacity levels, and the demand data of a buyer is classified and stored according to demand time periods and demand capacity.

[0055] Next, the multi-dimensional adaptive matching result is input to the basic risk identification network, and a discrimination layer is activated to perform combined adaptive discrimination. In this step, it is determined according to the multi-dimensional adaptive matching result whether the matching of the buyer and the seller conforms to an independent adaptive result or a combined adaptive result. The independent adaptive result refers to a situation in which one seller can meet all the demands of a buyer, and the combined adaptive result refers to a situation in which multiple sellers are combined to meet the demands of a buyer. The discrimination layer determines whether it is independent adaptation or combined adaptation according to the input information.

[0056] When the determination result is the independent adaptation result, the seller static feature corresponding to the independent adaptation result is called, and the seller static feature, the independent adaptation result, and the request subject static feature are sent to the first bidirectional anomaly identification layer, a four-dimensional risk index is called for risk identification, and the four-dimensional risk index includes a performance risk, a liquidity risk, a quotation anomaly risk, and a time sequence matching risk. Through the four-dimensional risk index, the possible risks in the power transaction process are comprehensively evaluated, and a bidirectional transaction risk is output.

[0057] Further, the method provided by the application embodiment further comprises the following steps:

[0058] The four-dimensional risk index of the first bidirectional anomaly identification layer is called for risk identification, the four-dimensional risk index includes a performance risk, a liquidity risk, a quotation anomaly risk, and a time sequence matching risk, and the bidirectional transaction risk is output according to the risk identification result.

[0059] In the application embodiment, the four-dimensional risk index of the first bidirectional anomaly identification layer is first called for risk identification. In this process, by analyzing the data from the seller static feature, the independent adaptation result, and the request subject static feature, the performance risk, the liquidity risk, the quotation anomaly risk, and the time sequence matching risk are quantified, and their respective risk values are obtained through preset rule matching.

[0060] In the risk identification, the performance risk is evaluated by analyzing the power supply capacity and the historical performance record of the seller. According to the preset rule, whether the seller has the situation of not supplying power on time or not performing the contract according to the agreement in the past transaction history is checked. For example, if the number of times that a certain seller fails to perform the contract on time within the past three months exceeds 3 times, according to this historical record, the performance risk value is set to 0.8, indicating that the seller has a higher performance risk.

[0061] The liquidity risk evaluates the financial situation of the buyer and the seller in the transaction, especially whether the liquidity is sufficient. By analyzing the payment history of the buyer and the financial situation of the seller through the preset rule, for example, whether the buyer has the behavior of frequently delaying the payment of electricity charges, or whether the seller frequently faces the difficulty of fund turnover. If the frequency of payment delay of the buyer exceeds the preset standard, and the proportion of the amount of each payment to the account balance is relatively high (for example, more than 30%), the liquidity risk value is set to 0.7, indicating that the liquidity risk is high.

[0062] The offer abnormality risk reflects the abnormality of the offer of the seller in market fluctuation. By analyzing the historical data of the offer of the seller and comparing with the normal fluctuation range of the market, if the fluctuation range of the offer of the seller exceeds the normal fluctuation range of the market (for example, the fluctuation range of the market offer is ±10%, and the fluctuation range of the offer of the seller reaches ±30%), according to the preset rule, it is identified that the offer of the seller has abnormality risk, and a higher risk value is given. For example, if the offer fluctuation of a seller is abnormal, the offer abnormality risk value of the seller is set to 0.85, indicating that the seller has a greater risk of offer abnormality.

[0063] The time sequence matching risk assesses the matching degree of the buyer and the seller in the transaction time. By analyzing the power supply time of the seller and the demand time of the buyer, it is checked whether the two can completely match. If the power supply time of the seller matches the demand time of the buyer at the beginning, but the time does not match during the transaction process, the time sequence matching risk is triggered. For example, it is assumed that the demand time of the buyer is from 1 pm to 3 pm, and the seller can provide power in this time period, but in the actual transaction process, the seller fails to provide power in this time period due to some reasons, resulting in that the buyer fails to obtain the required power on time. By monitoring and analyzing this time sequence mismatching risk, if it is found that the frequency of time sequence mismatching is high, a higher risk value, for example, 0.6, is assigned to it, indicating that the transaction has a time sequence matching problem.

[0064] Through the risk identification, the quantified values of the performance risk, the liquidity risk, the offer abnormality risk and the time sequence matching risk are obtained. Then, according to the preset weights corresponding to the performance risk, the liquidity risk, the offer abnormality risk and the time sequence matching risk, the quantified values of the performance risk, the liquidity risk, the offer abnormality risk and the time sequence matching risk are weighted and calculated to obtain a comprehensive risk value as the risk identification result.

[0065] After obtaining the risk identification result, the performance risk, the liquidity risk, the offer abnormality risk and the time sequence matching risk are interactively calculated with the key characteristics of the request subject and the seller subject to generate risk cross amplification characteristics. These characteristics include fund performance interaction amplification characteristics, offer performance interaction amplification characteristics and time sequence performance interaction amplification characteristics, which reflect the intensifying influence of the interaction between different risk factors on the transaction risk. Through these cross amplification characteristics, the risk identification result is compensated and corrected, and finally the bidirectional transaction risk is output.

[0066] Further, in the method provided by the application embodiment, the output of the bidirectional transaction risk according to the risk identification result further includes:

[0067] The calculated performance risk, liquidity risk, bid abnormality risk, and timing matching risk are respectively interacted with the request subject and the seller subject to calculate risk cross amplification features, including fund performance interaction amplification features, bid performance interaction amplification features, and timing performance interaction amplification features. After the risk cross amplification features are used to compensate the risk identification result, the bidirectional transaction risk is output.

[0068] In the embodiments of the present application, the calculated performance risk, liquidity risk, bid abnormality risk, and timing matching risk are respectively interacted with the request subject and the seller subject to calculate risk cross amplification features. In this process, the fund performance interaction amplification features are calculated first, that is, the seller performance risk is multiplied by the buyer liquidity risk to obtain. When the bid performance interaction amplification features are calculated, the seller bid abnormality risk is multiplied by the buyer bid volatility to obtain. The buyer bid volatility is quantified by analyzing the bid change of the buyer in the historical transaction. For example, in the past 30 days, the buyer provided bids of 250 yuan / MWh, 270 yuan / MWh, 240 yuan / MWh, 260 yuan / MWh, 280 yuan / MWh, etc. The standard deviation (i.e. fluctuation amplitude) of these bids is calculated, and the fluctuation amplitude is assumed to be 30 yuan. According to a preset rule, the bid volatility of the buyer can be quantified as 0.7, wherein the preset rule is that each fluctuation amplitude is preset with a corresponding volatility. Finally, the timing performance interaction amplification features are calculated, that is, the seller performance risk is multiplied by the timing matching risk to obtain. Through this calculation process, the risk cross amplification features are established.

[0069] Finally, the risk identification result is compensated by using the risk cross amplification features, that is, the fund performance interaction amplification features, the bid performance interaction amplification features, the timing performance interaction amplification features, and the risk identification result are multiplied to calculate the bidirectional transaction risk.

[0070] Further, in the method provided by the embodiments of the present application, the activation discrimination layer performs combined adaptive discrimination, which further includes:

[0071] If the discrimination result is a combined adaptive result, the seller allocation combination based on the time period and the allocated capacity is called by using the static features to establish a seller allocation combination set. The combined performance coordination features are established for the seller allocation combination set, including combined performance covariance, time splicing continuity, performance redundancy, and combined performance risk weight. The combined performance coordination features, the number of combinations, and the request subject static features are sent to the second bidirectional abnormality identification layer to output the bidirectional transaction risk.

[0072] In the embodiments of the present application, when the discrimination result is the combination fitting result, first, the static characteristics are used to perform time period and allocation capacity based allocation combination of the sellers. By analyzing the static characteristics of the sellers, including the power supply capacity of the sellers, in combination with the demand time period and demand capacity of the buyers, a combination of multiple sellers is generated, and a seller allocation combination set is established.

[0073] Next, the combination performance coordination characteristics of the seller allocation combination set are established. The combination performance coordination characteristics include combination performance covariance, time splicing continuity, performance redundancy, and combination performance risk weight. Among them, the combination performance covariance is used to measure the consistency of the performance capabilities of multiple sellers, and reflects the mutual relationship between the performance capabilities of the sellers. To calculate the combination performance covariance, first, the performance data of each seller in multiple trading periods is collected, which is represented in the form of a binary variable, where 1 represents timely performance, and 0 represents untimely performance. Then, the average value of the performance data of each seller, i.e., the performance rate, is calculated. Next, the difference between the performance data of each seller and its performance rate is calculated. Then, the differences of all sellers in each time period are multiplied to obtain the difference products between them. Finally, the products are summed and the average value is calculated to obtain the combination performance covariance.

[0074] The time splicing continuity reflects whether the sellers in the combination can provide continuous power supply within the demand time period of the buyer. If the power supply time periods of multiple sellers can be seamlessly connected or overlapped, it means that the time splicing continuity of the combination is strong, and is recorded as 1. By analyzing the overlap between the power supply time period of each seller and the demand time period of the buyer, the idle time between each pair of sellers, i.e., the blank part of the power supply time period of the seller and the time period of the other seller, is calculated. By calculating the proportion of these idle times in the demand time of the buyer, and by subtracting the proportion from 1, the time splicing continuity is obtained.

[0075] The performance redundancy is used to measure whether the seller combination has sufficient redundant power supply when performing to cope with the situation of untimely performance of the seller. By calculating the difference between the total power supply capacity of the seller combination and the demand capacity of the buyer, the amount of redundant power is obtained. That is, by calculating the difference between the total power supply capacity of the seller combination and the demand capacity of the buyer, and subtracting the demand capacity of the buyer, the performance redundancy is obtained.

[0076] The combination performance risk weight is the overall performance risk of the seller combination calculated by the weighted average method. The performance risk of each seller in the combination is weighted and averaged according to the proportion of its power supply capacity, so as to obtain a combination performance risk weight.

[0077] Finally, the combination performance coordination feature, the combination quantity, and the request subject static feature are sent to the second bidirectional abnormality identification layer, and bidirectional transaction risk is output. In this process, first, the combination performance coordination feature (such as combination performance covariance, time splicing continuity, performance redundancy, and combination performance risk weight), the combination quantity (that is, the number of seller combinations), and the request subject static feature (such as the demand time, demand capacity, payment history, and the like of the buyer) are standardized, and different types of data are converted into a unified scale. Subsequently, the second bidirectional abnormality identification layer is used to weight and sum all input features according to a preset weight, and bidirectional transaction risk is obtained.

[0078] Step S500: The bidirectional transaction risk is compensated for risk using the behavior trajectory, and bidirectional transaction risk warning is generated.

[0079] In the embodiment of the present application, when the behavior trajectory is used to compensate for the bidirectional transaction risk, first, the behavior trajectory is compared with a preset rule to determine a compensation factor. The compensation factor is used to adjust the bidirectional transaction risk coefficient according to the time sequence behavior feature and external events. The time sequence bidding behavior feature reflects the fluctuation of the seller's bid. If the bid fluctuation exceeds the preset range, the compensation factor is correspondingly increased. The time sequence capacity supply behavior feature focuses on the change of the seller's power supply capacity. When the power supply capacity fluctuation exceeds a certain threshold, the compensation factor is also increased. The time sequence fund behavior feature involves fund flow, such as buyer payment delay, and the compensation factor is increased. The external event compensation factor reflects the influence of external factors such as weather and market demand. For example, in extreme weather, the compensation factor is increased. By adding all the compensation factors, a comprehensive compensation factor is obtained. For example, if the bid fluctuation is 15% (the compensation factor is increased by 0.1), the power supply capacity change is 18% (the compensation factor is increased by 0.15), the buyer payment delay is 30 minutes (the compensation factor is increased by 0.1), and the high-temperature weather causes the power demand to surge (the compensation factor is increased by 0.2).

[0080] Then, the comprehensive compensation factor is multiplied by the bidirectional transaction risk to obtain the compensated bidirectional transaction risk, and compared with a preset risk coefficient threshold. When the compensated bidirectional transaction risk is greater than the preset risk coefficient threshold, it is considered that there is a transaction risk, and bidirectional transaction risk warning is performed at this time.

[0081] In the embodiment of the present application, as described above, the embodiment of the present application at least has the following technical effects:

[0082] The application obtains an accessed power transaction request, performs request analysis on the power transaction request, and establishes a request analysis result, wherein the request analysis includes transaction information analysis and request subject analysis; after updating a seller information library, the application calls a transaction information analysis result in the request analysis result, performs multi-dimensional adaptive matching of the seller information library, and establishes a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching includes independent adaptive matching and combined adaptive matching; the application uses the request subject analysis result and the updated seller information library to establish a behavior track of a request subject and a seller subject; the application calls a basic risk identification network, inputs the multi-dimensional adaptive matching result into the basic risk identification network, establishes a two-way transaction risk, and uses the behavior track to perform risk compensation on the two-way transaction risk to generate a two-way transaction risk warning. The application solves the technical problem of lack of effective risk identification and warning mechanism in the power transaction process in the prior art, and achieves the technical effects of optimizing transaction risk control and improving transaction security by establishing a multi-dimensional adaptive matching and risk compensation mechanism.

[0083] Embodiment two, based on the same inventive concept as the power transaction risk control method in the foregoing embodiments, as shown in the following table, the application provides a power transaction risk control system, and the system and method embodiments in the application are based on the same inventive concept. The system includes: Figure 2

[0084] The request analysis module 11 is configured to obtain an accessed power transaction request, perform request analysis on the power transaction request, and establish a request analysis result, wherein the request analysis includes transaction information analysis and request subject analysis; the adaptive matching module 12 is configured to update a seller information library, call a transaction information analysis result in the request analysis result, perform multi-dimensional adaptive matching of the seller information library, and establish a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching includes independent adaptive matching and combined adaptive matching; the behavior track establishment module 13 is configured to use the request subject analysis result and the updated seller information library to establish a behavior track of a request subject and a seller subject; the risk identification module 14 is configured to call a basic risk identification network, input the multi-dimensional adaptive matching result into the basic risk identification network, establish a two-way transaction risk, and use the behavior track to perform risk compensation on the two-way transaction risk to generate a two-way transaction risk warning.

[0085] Further, the system is also configured to implement the following functions:

[0086] ​The updated seller information base and the request subject analysis result are sent to the basic risk identification network for performing network analysis database initialization after static feature extraction. After the multi-dimensional adaptive matching result is input into the basic risk identification network, the combination adaptive discrimination is performed by activating the discrimination layer. If the discrimination result is an independent adaptive result, the seller static feature corresponding to the independent adaptive result is called, and the seller static feature, the independent adaptive result and the request subject static feature are sent to the first bidirectional abnormality identification layer to output bidirectional transaction risk.

[0087] Further, the system is also used to implement the following functions:

[0088] If the discrimination result is a combination adaptive result, the seller allocation combination based on time period and allocation capacity is performed by calling the static feature to establish a seller allocation combination set. The combination performance cooperation feature is established for the seller allocation combination set, and the combination performance cooperation feature includes combination performance cooperation covariance, time splicing continuity, performance redundancy, and combination performance risk weight. The combination performance cooperation feature, the number of combinations, and the request subject static feature are sent to the second bidirectional abnormality identification layer to output bidirectional transaction risk.

[0089] Further, the system is also used to implement the following functions:

[0090] The first bidirectional abnormality identification layer is called to perform risk identification, and the four-dimensional risk index includes performance risk, liquidity risk, quotation abnormality risk, and time sequence matching risk. The bidirectional transaction risk is output according to the risk identification result.

[0091] Further, the system is also used to implement the following functions:

[0092] The calculated performance risk, liquidity risk, quotation abnormality risk, and time sequence matching risk are respectively interactively calculated with the request subject and the seller subject to establish risk cross amplification features, and the risk cross amplification features include fund performance interactive amplification feature, quotation performance interactive amplification feature, and time sequence performance interactive amplification feature. After the risk identification result is compensated by using the risk cross amplification features, the bidirectional transaction risk is output.

[0093] Further, the system is also used to implement the following functions:

[0094] According to the request resolution result, requirement time and requirement capacity are obtained; the requirement time and requirement capacity are taken as associated matching features, independent seller adaptation matching of the seller information base is executed, a first dimension adaptation matching result is established; after configuring combination quantity constraint, the requirement time and requirement capacity are taken as associated matching features, combination seller adaptation matching of the seller information base is executed, a second dimension adaptation matching result is established; and a multi-dimension adaptation matching result is established according to the first dimension adaptation matching result and the second dimension adaptation matching result.

[0095] Further, the system is further used to realize the following functions:

[0096] After configuring a grabbing time window, time sequence behavior feature extraction is performed in the grabbing time window, and the time sequence behavior feature extraction includes time sequence bidding behavior feature, time sequence capacity supply behavior feature and time sequence fund behavior feature; and a behavior track is established by using the time sequence behavior feature extraction result.

[0097] Further, the system is further used to realize the following functions:

[0098] An external event correlator is configured; the time sequence behavior feature extraction result is taken as matching features, event correlation matching based on the external event correlator is executed, and an event sensitive behavior track is established; and the behavior track is compensated and corrected by using the event sensitive behavior track.

[0099] Embodiment three, based on the power transaction risk control method in the foregoing embodiments, the same inventive concept is provided, and the application also provides a computer readable storage medium, the computer readable storage medium stores a computer program, and the computer program realizes the steps of the method in any one of the foregoing embodiments when executed.

[0100] It should be noted that the above-mentioned sequence of the embodiments of the application is only for description, and does not represent the advantages and disadvantages of the embodiments. The above describes a specific embodiment of the present application. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0101] The above only describes the preferred embodiments of the application and does not limit the application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application should be included in the protection scope of the application.

[0102] The specification and drawings are, of course, to be regarded in an illustrative rather than a restrictive sense. It is to be understood that any such modifications, variations, combinations or equivalents that fall within the scope of the application are intended to be embraced herein.

Claims

1. A method for power transaction risk management, characterized in that, The method comprises: acquiring an accessed power transaction request, performing request analysis on the power transaction request, and establishing a request analysis result, wherein the request analysis comprises transaction information analysis and request subject analysis; after updating the seller information database, calling the transaction information analysis result in the request analysis result, performing multi-dimensional adaptive matching of the seller information database, and establishing a multi-dimensional adaptive matching result, wherein the multi-dimensional adaptive matching comprises independent adaptive matching and combined adaptive matching; establishing the behavior track of the request subject and the seller subject by using the request subject analysis result and the updated seller information database; calling a basic risk identification network, inputting the multi-dimensional adaptive matching result into the basic risk identification network, and establishing a two-way transaction risk; performing risk compensation on the two-way transaction risk by using the behavior track, and generating a two-way transaction risk warning; through the risk identification, obtaining the quantified values of the performance risk, the liquidity risk, the quotation abnormal risk, and the time sequence matching risk, then performing weighted calculation on the quantified values of the performance risk, the liquidity risk, the quotation abnormal risk, and the time sequence matching risk according to the preset weights corresponding to the performance risk, the liquidity risk, the quotation abnormal risk, and the time sequence matching risk, and obtaining a comprehensive risk value as a risk identification result; outputting the two-way transaction risk according to the risk identification result, comprising: performing key feature interaction calculation on the calculated performance risk, the liquidity risk, the quotation abnormal risk, and the time sequence matching risk and the request subject and the seller subject respectively, establishing risk cross amplification features, and the risk cross amplification features comprise fund performance interaction amplification features, quotation performance interaction amplification features, and time sequence performance interaction amplification features; after compensating the risk identification result by using the risk cross amplification features, outputting the two-way transaction risk; calling the transaction information analysis result in the request analysis result, performing multi-dimensional adaptive matching of the seller information database, and establishing a multi-dimensional adaptive matching result, comprising: obtaining a demand time and a demand capacity according to the request analysis result; taking the demand time and the demand capacity as associated matching features, performing independent seller adaptive matching of the seller information database, and establishing a first-dimensional adaptive matching result; after configuring a combined quantity constraint, taking the demand time and the demand capacity as associated matching features, performing combined seller adaptive matching of the seller information database, and establishing a second-dimensional adaptive matching result; establishing a multi-dimensional adaptive matching result according to the first-dimensional adaptive matching result and the second-dimensional adaptive matching result.

2. The power transaction risk management method of claim 1, wherein, the calling of the basic risk identification network, the inputting of the multi-dimensional adaptive matching result into the basic risk identification network, and the establishment of a two-way transaction risk, comprising: after static feature extraction of the updated seller information database and the request subject analysis result, sending the updated seller information database and the request subject analysis result to the basic risk identification network to perform network analysis database initialization; after inputting the multi-dimensional adaptive matching result into the basic risk identification network, activating a discrimination layer to perform combined adaptive discrimination; if the discrimination result is an independent adaptive result, calling a seller static feature corresponding to the independent adaptive result, and sending the seller static feature, the independent adaptive result, and a request subject static feature to a first two-way abnormality identification layer to output a two-way transaction risk. 3.The power transaction risk management method of claim 2, wherein, The activation discrimination layer performs combined adaptation discrimination, including: If the discrimination result is a combined adaptation result, a seller allocation combination based on a time period and an allocated capacity is called using the static characteristics, and a seller allocation combination set is established; Combined performance coordination characteristics are established for the seller allocation combination set, including combined performance coordination covariance, time splicing continuity, performance redundancy, and combined performance risk weight; The combined performance coordination characteristics, the number of combinations, and the static characteristics of the request subject are sent to a second two-way abnormality identification layer, and a two-way transaction risk is output.

4. The power transaction risk management method of claim 2, wherein, The seller static characteristics, the independent adaptation result, and the static characteristics of the request subject are sent to a first two-way abnormality identification layer, and a two-way transaction risk is output. A four-dimensional risk index of the first two-way abnormality identification layer is called to perform risk identification, including performance risk, liquidity risk, quotation abnormality risk, and time sequence matching risk; The two-way transaction risk is output according to the risk identification result.

5. The power transaction risk management method of claim 1, wherein, The request subject and the seller subject are used to establish a behavior track using the request subject analysis result and the updated seller information base, including: After configuring a capture time window, time sequence behavior characteristic extraction is performed in the capture time window, including time sequence quotation behavior characteristics, time sequence capacity supply behavior characteristics, and time sequence fund behavior characteristics; A behavior track is established using the time sequence behavior characteristic extraction result.

6. The power transaction risk management method of claim 5, wherein, The behavior track is established using the time sequence behavior characteristic extraction result, including: An external event correlator is configured; The time sequence behavior characteristic extraction result is used as a matching feature, event correlation matching based on the external event correlator is performed, an event-sensitive behavior track is established, and behavior track compensation correction is performed using the event-sensitive behavior track. The system is used to perform the power transaction risk control method according to any one of claims 1-6, and the system includes:

7. The power transaction risk management system, characterized in that, A request analysis module is configured to obtain an accessed power transaction request, analyze the power transaction request, establish a request analysis result, and perform transaction information analysis and request subject analysis; An adaptation matching module is configured to update a seller information base, call a transaction information analysis result in a request analysis result, perform multi-dimensional adaptation matching of the seller information base, and establish a multi-dimensional adaptation matching result, wherein the multi-dimensional adaptation matching includes independent adaptation matching and combined adaptation matching; A behavior track establishment module is configured to establish a behavior track of a request subject and a seller subject using a request subject analysis result and an updated seller information base; A risk identification module is configured to call a basic risk identification network, input the multi-dimensional adaptation matching result into the basic risk identification network, and establish a two-way transaction risk; A risk compensation module is configured to perform risk compensation on the two-way transaction risk using the behavior track, and generate a two-way transaction risk warning. The program is executed by a processor to implement the power transaction risk control method according to any one of claims 1-6.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that ​

Images