User demand-oriented power transaction matching recommendation method and system

Abstract

The invention is applicable to the field of power demand analysis, and provides a user demand-oriented power transaction matching recommendation method and system. The method comprises the following steps of: acquiring power consumption data of a user, and generating a resonance group and a feature packet with similar power consumption modes and risk tolerance; generating a power utilization comprehensive bidding plan in combination with the resonance group characteristics and the real-time market data; performing transaction matching based on the bidding plan, preferentially selecting the power generation resource with the closest electrical distance, introducing a green energy premium coefficient, and outputting a transaction result; and performing investment portfolio optimization analysis according to a transaction result, and generating a customized power package containing traditional energy, new energy and stored energy in combination with user risk preference. Through multi-dimensional data integration and intelligent algorithm application, accurate matching of power generation resources and user demands is realized, power transaction efficiency and user satisfaction are improved, and new energy consumption and sustainable development of the power market are promoted.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power demand analysis, and particularly relates to a power transaction matching recommendation method and system oriented to user demand. BACKGROUND

[0002] With the marketization of the electricity market, the demand of users for power services gradually changes from single power supply guarantee to individualization, economy, and environmental protection, and the rapid development of new energy and energy storage technology further enriches the energy types and transaction modes of power transactions. Currently, power transaction matching gradually changes from the traditional "generation side dominance" to the "user side demand orientation", which needs to combine multi-dimensional data such as user power consumption behavior characteristics, price sensitivity, risk preference, etc., to realize the accurate docking of power generation resources and user demand, so as to improve the efficiency of power resource allocation, promote new energy consumption, and ensure power supply stability.

[0003] In the prior art, the power transaction matching method is mostly based on user basic power load data for clustering grouping, and realizes transaction matching through a simple bidding strategy, and provides a standardized power package according to the matching result. Such a method usually includes steps of user load data acquisition, clustering grouping, bidding matching, and package generation, wherein the clustering is mostly dependent on the coincidence degree of peak and valley periods of the load curve, the bidding strategy mainly depends on the market benchmark price, and the matching focuses on whether the capacity of the power generation resource meets the user demand.

[0004] However, the prior art does not fully integrate the deep features such as user price sensitivity, risk preference, and historical performance rate, resulting in insufficient homogeneity of the user groups generated by clustering, which is difficult to reflect the real demand difference of users; the electrical distance between the power generation resource and the user is not considered in the transaction matching, which is easy to cause high power transmission loss or insufficient power supply reliability; there is a lack of differentiated incentive mechanism for green energy, which is difficult to effectively guide new energy consumption; the customized package generation does not combine portfolio optimization and user risk preference, which cannot balance the economy, stability, and environmental protection of the package, and is difficult to meet the individual needs of users. SUMMARY

[0005] The purpose of the present application is to provide a power transaction matching recommendation method and system oriented to user demand, aiming at solving the technical problems existing in the prior art determined in the background art.

[0006] The present application is implemented as follows: a power transaction matching recommendation method oriented to user demand, the method comprising: obtaining user power load characteristics, price sensitivity, and risk preference data, and integrating them into user power consumption data; performing time-space dimension aggregation on the user power consumption data to generate a resonance group with similar power consumption mode and risk bearing capacity; The bidding agent based on reinforcement learning generates a power consumption comprehensive bidding plan according to the aggregated size of the resonance group, the historical fulfillment rate and the real-time market power consumption data; Transaction matching is performed based on the power consumption comprehensive bidding plan, the closest power generation resources in terms of electrical distance are preferentially matched, and a green energy premium coefficient is introduced, and a transaction result is output; According to the transaction result, portfolio optimization analysis is performed, and a customized power package containing traditional energy, new energy and energy storage is generated combined with risk preference data.

[0007] As a further scheme of the present application, the acquisition of user power consumption load characteristics, price sensitivity and risk preference data includes: The peak-valley period distribution, load fluctuation rate standard deviation and transferable load identifier of the future 24-hour load curve are collected through the smart meter; Based on the price elasticity matrix, the load response rate of the user to the ±20% fluctuation of the time-of-use electricity price is calculated, and a price sensitivity coefficient vector is generated; Through the risk preference data, the tolerance of the user to the ±25% price fluctuation and the power supply interruption of more than 1 hour is determined, and a risk preference level is output; The user historical fulfillment rate curve is extracted from the power transaction database, and the average fulfillment rate is calculated.

[0008] As a further scheme of the present application, the spatio-temporal dimension aggregation of the user power consumption data to generate a resonance group with similar power consumption mode and risk bearing capacity includes: Based on the peak-valley period distribution of the load curve, the power consumption period coincidence degree between users is calculated, and when the period coincidence degree is greater than 80%, it is marked as a spatio-temporal correlation user group; For the spatio-temporal correlation group, the shape distance of the load curve of the users in the group and the standard deviation of the risk preference levels of the users in the group are calculated, when the shape distance is less than 15%, it is determined that the users have similar power consumption mode, and when the standard deviation is less than 0.5, it is determined that the users have similar risk bearing capacity; The user group that meets the similar power consumption mode and similar risk bearing capacity at the same time is clustered and optimized to generate a resonance group with similar power consumption mode and risk bearing capacity, and a resonance group feature package containing the load fluctuation rate mean value, risk preference level and average fulfillment rate.

[0009] As a further scheme of the present application, the generation of the power consumption comprehensive bidding plan includes: The resonance group feature package and the real-time market power consumption data are input into the state encoder to generate a six-dimensional state vector; The DuelingDDQN network is used to process the six-dimensional state vector, and the output actions are obtained according to the price sensitivity coefficient, the transferable load proportion and the period coincidence degree, including the time-of-use pricing strategy, the standby capacity parameter and the ramp rate compensation scheme. The risk hedging parameters are dynamically set according to a historical performance rate curve, and a spread contract hedging mechanism is embedded. The output action and the risk hedging parameters are integrated to generate a comprehensive bidding plan.

[0010] As a further scheme of the application, the transaction matching is performed, the power generation resources with the closest electrical distance are preferentially matched, and a green energy premium coefficient is introduced, and the transaction result includes: The time-of-use pricing strategy in the bidding plan is analyzed, and an auction process is started, when the bidding reaches the bottom price of the pricing strategy, a technical bid application meeting the backup capacity parameters and the ramp rate compensation scheme is submitted; The electrical distance between the power generation resource and the resonance group is calculated, the resources with a topological distance <20km and meeting the ramp rate compensation scheme are preferentially selected, and a matching result data is generated; The green premium is calculated by using the price sensitivity coefficient, and the transaction result including the matching result data, the green premium and the performance clause is output.

[0011] As a further scheme of the application, the generation of the customized power package including traditional energy, new energy and energy storage includes: The matching result data is taken as input to construct a portfolio model, the optimal investment weight is solved in combination with the user risk preference level, the electricity price fluctuation scenario is simulated, the income stability is verified, and the portfolio optimization parameter is output; The energy type proportion data in the matching result data is mapped to the Black-Litterman model asset category; The constraint condition data is set based on the risk preference level data, the scenario optimization data is generated by using the load fluctuation rate data, the backup capacity service data is generated in combination with the backup capacity parameter in the bidding plan; According to the transferable load identifier, the load fluctuation rate standard deviation and the portfolio optimization parameter, the configuration capacity and the time period scheduling strategy of the energy storage are determined, and energy storage configuration data is generated; Based on the energy storage configuration data, a power package including energy ratio data, time-of-use pricing strategy data, backup capacity service data and energy storage configuration data is output.

[0012] Another object of the application is to provide a power transaction matching recommendation system for user demand, the system includes: A data acquisition and integration module is used to acquire user electricity load characteristics, price sensitivity and risk preference data, and integrate the data into user electricity data; A resonance group generation module is used to perform time and space dimension aggregation on the user electricity data, and generate a resonance group with similar electricity mode and risk bearing capacity; A bidding agent module is configured to generate an electricity comprehensive bidding plan according to the aggregated size of the resonance group, the historical performance rate and real-time market electricity data by means of reinforcement learning bidding agent. A transaction matching module is configured to perform transaction matching based on the electricity comprehensive bidding plan, preferentially match the power generation resources closest in electrical distance, introduce a green energy premium coefficient, and output a transaction result. A portfolio optimization analysis module is configured to perform portfolio optimization analysis according to the transaction result, and generate a customized power package containing traditional energy, new energy and energy storage in combination with risk preference data.

[0013] The present application has the following advantages: The present application realizes the upgrade of electricity transaction from passive matching to active adaptation through the whole-process design of multi-dimensional data integration, accurate clustering, intelligent bidding, optimized matching and customized package generation. By obtaining user electricity load characteristics, price sensitivity, risk preference and historical performance rate data, a comprehensive user electricity portrait is constructed, providing an accurate data basis for subsequent links. By aggregating in time and space dimensions to generate resonance groups, combining load curve shape similarity and risk preference consistency analysis, the homogeneity of user electricity mode and risk bearing capacity within the group is ensured, laying a foundation for unified transaction strategy formulation. Reinforcement learning bidding agent is used to generate a comprehensive bidding plan, and the price sensitivity, transferable load ratio and other characteristics are dynamically adjusted to improve the targeting and flexibility of bidding. When matching transactions, power generation resources closest in electrical distance are preferentially selected to reduce power transmission loss and power supply fluctuation, and the introduction of a green energy premium coefficient effectively encourages new energy consumption, balancing economy and environmental protection. Finally, the customized package generated based on portfolio optimization integrates traditional energy, new energy and energy storage, and realizes the dynamic balance of income-risk-environment based on user risk preference and load characteristics, meeting the individual needs of different users. The overall process improves the efficiency, accuracy and user satisfaction of electricity transactions, and promotes the development of the electricity market in a more flexible, efficient and sustainable direction. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The flowchart of the electricity transaction matching recommendation method for user demand provided by the embodiment of the present application; Figure 2 The flowchart of obtaining user electricity load characteristics, price sensitivity and risk preference data provided by the embodiment of the present application; Figure 3 The flowchart of generating a resonance group with similar electricity mode and risk bearing capacity provided by the embodiment of the present application; Figure 4 The flowchart of generating an electricity comprehensive bidding plan provided by the embodiment of the present application; Figure 5 A flowchart of outputting transaction results provided by an embodiment of the present invention; Figure 6 A flowchart for generating a customized electricity package that includes traditional energy, new energy, and energy storage, provided in an embodiment of the present invention; Figure 7 This is a structural block diagram of a user-demand-oriented power transaction matching recommendation system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0016] Figure 1 Flowchart of the power transaction matching recommendation method oriented to user needs provided by an embodiment of the present invention, such as Figure 1 As shown, the method includes: S100, obtains user electricity load characteristics, price sensitivity and risk preference data, and integrates them into user electricity consumption data; Smart meters are used to collect the peak and valley distribution of the load curve, the standard deviation of the load fluctuation rate, and the transferable load identification of the next 24 hours. The specific implementation of this step depends on the real-time monitoring and high-frequency data recording functions of the smart meters. To obtain the peak-valley period distribution, the 24-hour period must first be divided into several standard periods (such as one hour as a unit). By calculating the proportion of the load in each period to the total daily load, the specific periods of load peak (over 20%), flat period (10%-20%), and valley (less than 10%) are determined. The purpose is to capture the temporal patterns of user electricity usage behavior and provide a temporal basis for subsequent time-of-use quotation strategies and energy storage scheduling. The standard deviation of load fluctuation is calculated by calculating the square root of the sum of the squares of the deviations of the load from the average load in each period within 24 hours. This indicator can effectively reflect the stability of user electricity consumption. High volatility indicates large fluctuations in electricity consumption, requiring more backup capacity or energy storage support. Conversely, low volatility indicates stable electricity consumption, which can reduce additional configuration. The identification of transferable loads requires consideration of the type of user electrical equipment. Smart meters can identify loads that can flexibly adjust their usage time during different time periods and mark their capacity proportions. This provides clear targets for subsequent energy storage configuration and load transfer strategies, thereby improving the utilization efficiency of power resources.

[0017] The load response rate of the user to the time-of-use electricity price ±20% fluctuation is calculated based on the price elasticity matrix to generate a price sensitivity coefficient vector. The price elasticity matrix is a matrix reflecting the quantitative relationship between the price change of different time periods and the load change. The element of the matrix represents the load change proportion when the price of a certain time period changes by 1%. In the specific implementation, the time periods with ±20% price fluctuation in the user's historical electricity consumption data are selected, the corresponding load change is counted, the response rate (load change rate / price change rate) is calculated, and the response rates of different time periods are integrated into a vector. The larger the absolute value of the value in the vector, the more sensitive the user is to the price of the time period. The purpose of this step is to quantify the reaction strength of the user to the price signal. For example, a high sensitivity user will significantly reduce the load when the price rises, while a low sensitivity user will be less affected. The advantage is that the subsequent time-of-use pricing strategy is more targeted, avoiding user loss or revenue loss caused by uniform pricing, and matching the price adjustment with the actual bearing capacity of the user.

[0018] The tolerance of the user to the electricity price fluctuation ±25% and power supply interruption >1 hour is determined through the risk preference data to output the risk preference level. The user's tolerance to the electricity price fluctuation ±25% is classified into categories such as completely unacceptable, reluctantly acceptable, basically acceptable, and completely acceptable, and the tolerance to the power supply interruption >1 hour is also classified. The risk preference level (the higher the level, the stronger the tolerance) is determined by combining the two types of tolerance. The user is classified from the risk dimension to ensure that the electricity package generated subsequently strikes a balance between revenue and risk. The advantage is to avoid recommending high-fluctuation energy combinations to risk-averse users and overly conservative packages to risk-preference users, thereby improving the user's acceptance of the package.

[0019] The historical compliance rate curve of the user is extracted from the electricity trading database, and the average compliance rate is calculated. The historical compliance rate curve is a curve drawn with time as the horizontal axis and the compliance rate (the number of compliance times / total transaction times) as the vertical axis by recording the user's past compliance with electricity or payment in each transaction. The average compliance rate is the arithmetic mean of the points on the curve. This step is used to evaluate the credit level and cooperation stability of the user. For example, a user with a high average compliance rate is more trustworthy and can be a high-quality member in the resonance group aggregation, reducing the risk of default in transactions and providing a basis for setting risk hedging parameters to ensure the rights and interests of both parties in the transaction.

[0020] As shown in Figure 2 The acquisition of the user's electricity load characteristics, price sensitivity, and risk preference data includes: S110, the peak-valley period distribution, load fluctuation rate standard deviation, and transferable load identifier of the future 24-hour load curve are collected by the smart meter. S120, calculate the load response rate of the user to the time-of-use electricity price ±20% fluctuation based on the price elasticity matrix, and generate a price sensitivity coefficient vector; S130, determine the tolerance of the user to the electricity price fluctuation ±25% and power supply interruption >1 hour through the risk preference data, and output the risk preference level; S140, extract the user historical compliance rate curve from the electricity trading database, and calculate the average compliance rate.

[0021] S200, aggregate the user electricity consumption data in space-time dimensions to generate a resonance group with similar electricity consumption patterns and risk tolerance; Based on the peak-valley period distribution of the load curve, the electricity consumption period coincidence degree between users is calculated. When the period coincidence degree exceeds a certain threshold, it is marked as a space-time associated user group. The peak-valley period of each user's 24-hour load curve needs to be converted into a standardized time interval (peak: 8:00-12:00, 18:00-22:00, valley: 0:00-6:00), and then the overlap length of the peak-valley periods of different users is compared to obtain the coincidence degree. For example, if the peak period of two users both includes 18:00-22:00, and the valley period both includes 0:00-6:00, and only the flat section has some differences, then the coincidence degree calculation result will be higher. The core purpose of this step is to select users with similar electricity consumption rhythm from the time dimension, and combine the spatial attributes of the distribution area (users in the same area are close in geographical distance and the power grid is closely connected), to preliminarily narrow the aggregation range. Its advantage is to avoid including users with completely mismatched electricity consumption periods in the same group through time-space double constraints, reduce the redundancy of subsequent data processing, and focus on effective samples for similarity analysis.

[0022] For the space-time associated user group, the similarity of electricity consumption patterns and risk tolerance needs to be determined by calculating the load curve shape distance and the standard deviation of the risk preference level. The load curve shape distance is calculated using the dynamic time warping (DTW) algorithm, which stretches or compresses local fragments of time series to find the optimal matching path of two load curves, which is more suitable for handling load curves with time offset or length difference than traditional Euclidean distance.

[0023] When calculating, the user's hourly load data is taken as a time series, and the minimum cumulative distance is iteratively solved by the DTW formula. The smaller the distance, the more similar the curve shape, i.e. the more similar the electricity consumption patterns. This step breaks through the surface association of simple period coincidence and deeply captures the internal law of load fluctuation (same electricity consumption in the daytime, some users have stable load, and some users have frequent load fluctuation), ensuring that the users in the group not only have consistent electricity consumption periods, but also have highly similar behavior patterns.

[0024] The calculation of the risk preference level standard deviation is based on the risk preference level (1-5 level) of the users in the group. The formula quantifies the dispersion of the level distribution, and the smaller the standard deviation, the more consistent the users in the group are in tolerating price fluctuations and power interruptions. This avoids aggregating users with significantly different risk tolerance, eliminates internal contradictions for subsequent unified risk hedging strategies and package design, ensures the consistency of the group's decision-making from the risk dimension, and reduces the coordination cost in subsequent transactions.

[0025] The user group that meets the similar power consumption mode and risk tolerance is clustered and optimized to generate a resonance group and a feature package. The group DTW distance and risk preference standard deviation are used as constraint conditions, and the user group is continuously merged or split until all groups meet the termination condition of shape distance < 15% and standard deviation < 0.5. The final output of the resonance group feature package integrates the average load fluctuation rate (reflecting the overall power consumption stability), the risk preference level (representing the risk attitude of the group), and the average compliance rate (reflecting the historical credit level). These feature data will be directly used as input parameters for subsequent bid plan generation, further improving the homogeneity of the group and ensuring that each resonance group becomes an organic whole with similar power consumption behavior, consistent risk attitude, and predictable credit level.

[0026] As shown in Figure 3 , the space-time dimension aggregation of the user power consumption data to generate a resonance group with similar power consumption mode and risk tolerance includes: S210, calculating the power consumption period coincidence degree between users based on the peak-valley period distribution of the load curve, and marking as a space-time associated user group when the period coincidence degree > 80%; S220, for the space-time associated group, calculating the shape distance of the load curve within the group and the standard deviation of the risk preference level of the users within the group , determining that the users have similar power consumption modes when the shape distance < 15%, and determining that the users have similar risk tolerance when the standard deviation < 0.5; ; In the formula, , are the load values of user and user at the same time scale, is the minimum shape distance of the first , data points of the load curves of the two users; ; In the formula, is the risk preference level (1-5 level) of the th user in the group, is the average value of the risk preference level of the users in the group, the number of users in the group; S230, clustering optimization is performed on the user group that meets the similar power consumption mode and similar risk tolerance capacity at the same time, to generate a resonance group with similar power consumption mode and risk tolerance capacity and a resonance group characteristic package containing load fluctuation rate mean, risk preference level and average compliance rate. S300, an enhanced learning bidding agent generates a comprehensive power consumption bidding plan according to the aggregate size of the resonance group, historical compliance rate and real-time market power consumption data; The load fluctuation rate mean in the resonance group characteristic package reflects the overall stability of group power consumption, the risk preference level embodies the tolerance attitude of the group to market fluctuations, and the average compliance rate represents the credit level of historical transactions; real-time market power consumption data includes dynamic information such as current period market benchmark price, supply-demand balance index, new energy generation proportion and transmission line load rate. The state encoder compresses these multi-source heterogeneous data into a six-dimensional vector through normalization processing (mapping different magnitude characteristic values to the 0-1 interval) and feature fusion, and each dimension corresponds to group risk level, power consumption stability, credit level, market benchmark price, supply-demand relationship and new energy proportion. The purpose of this operation is to eliminate redundant information and retain key features, so that high-dimensional complex data is converted into low-dimensional structured data suitable for reinforcement learning network processing, which has the advantage of reducing network computation complexity while ensuring the integrity of input information.

[0027] The DuelingDDQN network can more accurately evaluate the actual value of different actions in the current state by separating the state value function and the action advantage function, avoiding the bias of traditional DQN network in action value evaluation. Specifically, the network first extracts features from the six-dimensional state vector, and then designs branch calculation for different output actions: the generation of time-sharing bidding strategy is based on the price sensitivity coefficient, which couples the market benchmark price with the user's sensitivity to price through a formula. When the absolute value of the user's price sensitivity coefficient is large (i.e. the response to price fluctuations is strong), the bid will be close to the benchmark price to improve the transaction probability, while a lower sensitivity can adjust the bid appropriately to pursue higher returns, so that the bidding strategy is highly matched with the user's price tolerance, avoiding missing transactions or losing income due to over-high bidding or over-low bidding; The determination of the standby capacity parameter is related to the transferable load proportion. The higher the transferable load proportion, the stronger the flexibility of user power consumption, and the standby capacity required can be appropriately reduced, and vice versa. The formula quantifies this correlation into a specific standby capacity value through the product of the standby coefficient and the transferable load proportion, ensuring that the configuration of standby resources is neither excessively redundant nor insufficient; The ramp rate compensation scheme is generated according to the time period coincidence degree. The user group with high time period coincidence degree has strong synchronism in the load peak period. When the load fluctuation (peak-valley difference) is large, a higher ramp rate compensation is needed to stimulate the power generation side to quickly respond to the load change. The formula matches the compensation scheme to the load fluctuation characteristics of the group through the negative correlation between the compensation coefficient and the time period coincidence degree. The advantage of this sub-step is that the autonomous decision-making ability of the reinforcement learning algorithm is realized, and the dynamic optimization of the bidding strategy is realized. Compared with the traditional fixed strategy, it can better adapt to the differences in market changes and group characteristics.

[0028] The historical performance curve records the performance of the resonance group in the past transactions. If the curve shows a sustained high position (the performance rate is stable at a high level), the group credit is good, and the risk hedging parameter can be appropriately reduced, and only a small number of spread contracts are used to lock the price of the core load; if the curve fluctuates greatly or is generally low, the risk hedging parameter needs to be increased, and the coverage of the spread contract needs to be expanded to avoid losses caused by user default or market price crash.

[0029] The embedding of the spread contract specifically means that a part of the transaction volume at a fixed price is agreed in the bidding plan, and the remaining part participates in market bidding. The purpose is to pursue market income while providing a safe cushion for price risk for both parties. The historical credit data is converted into a quantifiable risk control basis, so that the risk response strategy matches the actual credit level of the group.

[0030] The integrated output action and risk hedging parameter generate a comprehensive bidding plan. The plan not only includes transaction core terms such as time-of-use pricing, backup capacity, and ramp rate compensation, but also includes specific risk hedging schemes (transaction volume and price locking period of spread contract), forming a strategy-risk integrated bidding scheme. It provides complete and clear bidding basis for the transaction matching link, ensures that the power generation side can fully understand the needs and risk preferences of the resonance group, and reduces the information asymmetry between the two parties.

[0031] As shown in Figure 4 , the method for generating a comprehensive power bidding plan comprises the following steps: S310, inputting the resonance group characteristic package and real-time market power data into a state encoder to generate a six-dimensional state vector; S320, processing the six-dimensional state vector using a DuelingDDQN network to obtain output actions including time-of-use pricing strategy, backup capacity parameter, and ramp rate compensation scheme according to the price sensitivity coefficient, transferable load ratio, and time period coincidence degree, respectively; The time-of-use pricing strategy calculation formula is: ; In the formula, is the time-of-use price, is the spread contract price, and is the spread contract price. Market benchmark electricity price for the time period, is the price adjustment coefficient (range 0.1-0.3), is the price sensitivity coefficient (calculated from the price elasticity matrix, ranging from -1 to 1); The calculation formula for the spare capacity parameter is: ; Where, is the spare capacity value, is the reserve coefficient (value range is 0.15-0.25), is the transferable load ratio (the ratio of the load corresponding to the transferable load identifier to the total load), is the average load of the resonance group; The calculation formula for the slope compensation scheme is: ; Where, is the compensation amount for the ramp rate, is the compensation coefficient (value range is 0.05-0.1), is the time period overlap (value range is 0-1), is the load fluctuation (the difference between the peak and valley of the load curve); S330 dynamically sets risk hedging parameters based on historical performance rate curves and embeds a CFD hedging mechanism; S340, integrate the output actions and risk hedging parameters to generate a comprehensive bidding plan.

[0032] S400: performing transaction matching based on the comprehensive electricity bidding plan, giving priority to matching power generation resources with the closest electrical distance, introducing a green energy premium coefficient, and outputting a transaction result; The time-of-use bidding strategy in the bid proposal is parsed and the auction process is initiated. When the bid reaches the bidding strategy's floor price, a technical bid application is submitted. This is the market-based pricing phase of transaction matching. The time-of-use bidding strategy includes floor prices for different time periods (such as peak, flat, and off-peak). The parsing process requires extracting the price threshold and corresponding load for each time period. For example, the peak period from 8:00 AM to 8:00 PM may have a higher floor price than other periods to match the higher electricity demand during this period.

[0033] The starting of the auction relies on the real-time price adjustment mechanism of the trading platform. The initial price is set slightly higher than the bidding bottom price, and then decreases by a fixed gradient (a certain percentage per minute) until it reaches the bidding bottom price of a certain period. At this time, the system automatically triggers the technical bid application and submits the corresponding backup capacity parameters and ramp rate compensation scheme (such as the compensation standard of the maximum hourly load change rate) of the period. Through the market-oriented bidding mechanism, a reasonable transaction price is found, which not only avoids the idle of resources caused by high bidding, but also prevents the damage to the interests of users caused by low bidding. The advantage is that the price formation process is transparent and efficient, and through the time-sharing auction, the electricity consumption characteristics of different periods are adapted to ensure that the supply and demand of each period are accurately matched. For example, for the evening peak period electricity consumption of the residential user resonance group, the auction mechanism can quickly match the power generation resources to ensure the stability of power supply.

[0034] The electrical distance is calculated by the formula, which is the sum of the product of the resistance value and impedance value of each section of the transmission line from the power generation resource to the distribution area where the resonance group is located. The core function of this formula is to quantify the energy loss and voltage stability in the transmission process. The smaller the electrical distance, the lower the loss of the transmission line and the smaller the voltage fluctuation, and the higher the reliability of power supply. In specific implementation, the transmission line parameters between the power generation resource and the distribution area are retrieved from the grid topology database, and the electrical distance value is calculated by substituting the formula. Then, combined with the topological distance (physical spatial distance), the resource with the smallest electrical distance and the topological distance within 20 km is selected as the priority, and it is verified whether the ramp rate of the resource meets the load fluctuation demand of the resonance group (if the load of the resonance group increases by 10% per hour during the peak period, the matching power generation resource must have a ramp rate not less than this proportion). Among the many power generation resources, the optimal solution with short distance, low loss, and fast response is found, which not only reduces the transmission cost, but also ensures that the load fluctuation of the group can be quickly responded. The advantage is that it breaks through the limitation of pure physical distance and selects resources from the perspective of grid operation characteristics. For example, although a certain wind farm is slightly far away in physical distance, it has a smaller electrical distance (because the line parameters are better), and it can still be a priority to achieve the balance between efficiency and cost.

[0035] The price sensitivity coefficient is applied to calculate the green premium, which is an incentive design link that takes into account environmental orientation and user preference. The calculation formula of the green premium combines the price sensitivity coefficient, the green energy coefficient, and the unit cost of new energy generation. The price sensitivity coefficient reflects the user's sensitivity to price. When the user's price sensitivity is low, the green premium can be appropriately increased to encourage more new energy access. Conversely, for price-sensitive user groups, the premium needs to be controlled at a lower level to avoid excessive burden.

[0036] In specific implementation, the price sensitivity coefficient vector of the resonance group is extracted, and the actual generation cost of new energy and the preset green energy coefficient are combined to calculate the green premium value of each period. Through economic incentives, the priority matching of new energy resources is guided, and the green transformation of energy structure is promoted. The advantage is that the green premium is linked with the price bearing capacity of users, avoiding resource mismatch caused by one-size-fits-all subsidies.

[0037] The matching result data determines the type of traded power generation resources, the proportion of each type, and the corresponding power supply period. The green premium is used as a price supplement for new energy transactions and is listed separately in the transaction price. The performance clause covers power supply reliability commitment, price adjustment mechanism, and default handling method. The specific implementation of this operation depends on the contract generation module of the transaction platform, which integrates the above data into standardized transaction certificates.

[0038] As shown in Figure 5 , the transaction matching prioritizes the power generation resources closest in electrical distance and introduces a green energy premium coefficient to output the transaction results, including: S410, analyze the time-of-use pricing strategy in the bidding plan and start the auction process. When the bidding reaches the bottom price of the pricing strategy, submit a technical bid application that meets the reserve capacity parameters and the ramp rate compensation scheme; S420, calculate the electrical distance between the power generation resource and the resonance group, and prioritize resources with a topological distance <20 km and a ramp rate compensation scheme to generate matching result data; ; In the formula, is the electrical distance, is the resistance value of the transmission line in the first segment (unit: Ω), is the impedance value of the transmission line in the first segment (unit: Ω), is the number of transmission line segments from the power generation resource to the distribution area where the resonance group is located; S430, calculate the green premium using the price sensitivity coefficient and output the transaction results including matching result data, green premium, and performance clause.

[0039] ; In the formula, is the green energy premium, is the green energy coefficient (value 1.1-1.3), is the price sensitivity coefficient, is the unit generation cost of new energy; S500, perform portfolio optimization analysis based on the transaction results and generate a customized electricity package including traditional energy, new energy, and energy storage in combination with risk preference data.

[0040] The matching result data is used as input to build a portfolio model, combined with the user's risk preference level to solve the optimal investment weight, and simulate the electricity price fluctuation scenario to verify the income stability, and output the portfolio optimization parameters. The matching result data includes the types of traditional energy and new energy traded, the quantity-price relationship and proportion of each type, which constitute the basic asset information of the portfolio. The portfolio model uses the mean-variance model, which focuses on balancing between income and risk: through the formula, the expected return rate of each type of energy (based on the transaction price and historical data estimation), risk aversion coefficient (determined by the risk preference level, such as conservative users taking higher values to reduce fluctuations) and return rate covariance (reflecting the correlation of different energy returns) are included in the calculation, to solve the investment weight (i.e. the proportion of traditional energy, new energy and energy storage) that optimizes the balance of income and risk. Simulating the electricity price fluctuation scenario is to generate a large number of possible electricity price trends to test the income stability of the portfolio under different scenarios, ensuring that the package can maintain its expected performance in market fluctuations.

[0041] The core value of the Black-Litterman model lies in combining the market equilibrium income with the user's subjective opinion (here embodied in the energy preferences of the resonance group) to correct the traditional mean-variance model's over-reliance on historical data. The mapping process needs to convert the actual traded energy types and their proportions into asset categories that the model can recognize and assign appropriate weights and expected returns. The specific implementation of this operation needs to establish a corresponding rule between energy types and financial asset categories, with the goal of making portfolio optimization more consistent with the actual characteristics of the electricity market. Electricity, as a special commodity, is greatly influenced by non-market factors, and the introduction of the Black-Litterman model can effectively integrate these subjective opinions, avoiding the misjudgment of market mutations by purely data-driven models, and making asset allocation both respect historical laws and adapt to reality.

[0042] The risk preference level determines the strictness of the constraint condition: for groups with low risk tolerance, the constraint condition requires higher standby capacity redundancy, while for groups with strong risk tolerance, the constraint can be appropriately relaxed to reduce costs. The load fluctuation rate data ensures that standby capacity services can cover electricity demand under different fluctuation conditions by generating multi-scenario optimization data. For groups with high fluctuation rates, scenario optimization data will include more extreme load growth situations, and the response speed requirement for standby capacity will be more stringent.

[0043] The standby capacity parameters (from the bidding plan) are converted into specific service terms, such as the activation conditions of standby capacity (triggered when the actual load exceeds the planned load by 10%), duration, and compensation standards.

[0044] As Figure 6As shown, the generation of the customized power package containing traditional energy, new energy and energy storage includes: S510, taking the matching result data as input, constructing a portfolio model, and solving the optimal investment weight combined with the user risk preference level: The optimal investment weight is solved by using the mean-variance model, and the calculation formula is: ; ; In the formula, is the investment weight of the first energy (including traditional energy, new energy and energy storage), is the expected return rate of the first energy, is the risk aversion coefficient (determined by the risk preference level, 1.5 for conservative type, 1.0 for stable type, and 0.5 for aggressive type), is the yield rate covariance of the first energy and the first energy, is the number of energy types; and simulate the electricity price fluctuation scenario to verify the yield stability and output portfolio optimization parameters; S520, mapping the energy type proportion data in the matching result data to the Black-Litterman model asset category; S530, setting constraint condition data based on the risk preference level data, and generating scenario optimization data by using the load fluctuation rate data, combined with the standby capacity parameter in the bid plan, to generate standby capacity service data; S540, according to the transferable load identifier, the load fluctuation rate standard deviation and the portfolio optimization parameters, determining the configuration capacity and time period scheduling strategy of the energy storage, and generating the energy storage configuration data; S550, based on the energy storage configuration data, outputting the power package containing energy ratio data, time-of-use pricing strategy data, standby capacity service data and energy storage configuration data.

[0045] Figure 7 The structural block diagram of the power transaction matching recommendation system for user demand provided by the embodiment of the present application is as shown in Figure 7 The system comprises: A data acquisition and integration module 100 is used to acquire user electricity load characteristics, price sensitivity and risk preference data, and integrate them into user electricity data; A resonance group generation module 200 is used to aggregate the user electricity data in time and space dimensions, and generate a resonance group with similar electricity consumption mode and risk bearing capacity; The bidding agent module 300 is configured to generate an electricity consumption comprehensive bidding plan according to the aggregate size of the resonance group, the historical fulfillment rate and real-time market electricity data by means of a reinforcement learning bidding agent; The transaction matching module 400 is configured to perform transaction matching based on the electricity consumption comprehensive bidding plan, preferentially match the power generation resources closest in electrical distance, introduce a green energy premium coefficient, and output a transaction result; The portfolio optimization analysis module 500 is configured to perform portfolio optimization analysis according to the transaction result, and generate a customized electricity package containing traditional energy, new energy and energy storage in combination with risk preference data.

[0046] The technical features of the above-described embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described, but it should be understood that any combination of the technical features that does not cause contradiction shall be considered within the scope of the present disclosure.

[0047] The above-described embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the patent scope of the present application. It should be noted that, for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which shall be within the protection scope of the present application. Therefore, the patent protection scope of the present application shall be subject to the appended claims.

[0048] The above-described only is the preferred embodiment of the present application, and does not limit the present application. Any modification, equivalent replacement and improvement within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A user-demand-oriented power transaction matching recommendation method, characterized in that: The method comprises: Obtain user electricity load characteristics, price sensitivity, and risk preference data, and integrate them into user electricity consumption data; Aggregating the user's electricity consumption data in time and space to generate a resonance group with similar electricity consumption patterns and risk tolerance; Generate a comprehensive electricity bidding plan based on the aggregate size, historical contract fulfillment rate, and real-time market electricity consumption data of the resonance group through a reinforcement learning bidding agent; Transaction matching is performed based on the comprehensive electricity bidding plan, with priority given to power generation resources with the closest electrical distance, and a green energy premium coefficient is introduced to output the transaction results; Based on the transaction results, we conduct portfolio optimization analysis and generate customized electricity packages that include traditional energy, new energy and energy storage, combined with risk preference data.

2. The method according to claim 1, characterized in that The acquisition of user electricity load characteristics, price sensitivity and risk preference data includes: The peak and valley period distribution, load fluctuation standard deviation and transferable load identification of the load curve for the next 24 hours are collected through smart meters; Based on the price elasticity matrix, the user's load response rate to the ±20% fluctuation of the time-of-use electricity price is calculated to generate the price sensitivity coefficient vector; Determine the user's tolerance for electricity price fluctuations of ±25% or power outages >1 hour through risk preference data, and output the risk preference level; The user's historical compliance rate curve is extracted from the power transaction database, and the average compliance rate is calculated.

3. The method according to claim 2, characterized in that Aggregating the user electricity consumption data in time and space dimensions to generate a resonance group with similar electricity consumption patterns and risk tolerance includes: Based on the peak and valley period distribution of the load curve, the overlap of electricity consumption periods among users is calculated. When the overlap is greater than 80%, it is marked as a spatiotemporally associated user group. For spatiotemporal correlation groups, the load curve shape distance and the standard deviation of the risk preference level of users within the group are calculated. When the shape distance is less than 15%, they are judged to have similar power consumption patterns; when the standard deviation is less than 0.5, they are judged to have similar risk tolerance. Cluster optimization is performed on user groups that meet both similar electricity consumption patterns and similar risk tolerance, generating resonance groups with similar electricity consumption patterns and risk tolerance and resonance group feature packages containing load volatility mean, risk preference level and average fulfillment rate.

4. The method according to claim 3, characterized in that The comprehensive bidding plan for power generation and consumption includes: Input the resonance group feature package and real-time market electricity consumption data into the state encoder to generate a six-dimensional state vector; The DuelingDDQN network is used to process the six-dimensional state vector and obtain output actions based on the price sensitivity coefficient, the proportion of transferable load, and the time overlap, including the time-of-use quotation strategy, spare capacity parameters, and ramp rate compensation scheme; Dynamically set risk hedging parameters based on historical performance rate curves and embed CFD hedging mechanisms; Integrate output actions and risk hedging parameters to generate a comprehensive bidding plan.

5. The method according to claim 4, characterized in that The transaction matching is performed by giving priority to the power generation resources with the closest electrical distance and introducing the green energy premium coefficient. The output transaction results include: Analyze the time-sharing bidding strategy in the bid plan and initiate the auction process. When the bid reaches the bidding strategy floor price, submit a technical bid application that meets the spare capacity parameters and ramp rate compensation plan. Calculate the electrical distance between the power generation resource and the resonance group, give priority to resources with a topological distance less than 20km and that meet the ramp rate compensation scheme, and generate matching result data; Apply the price sensitivity coefficient to calculate the green premium and output the transaction results including matching result data, green premium and performance terms.

6. The method according to claim 5, characterized in that The customized electricity package generated, which includes traditional energy, new energy and energy storage, includes: The matching result data is used as input to build an investment portfolio model. The optimal investment weight is solved based on the user's risk preference level. The electricity price fluctuation scenario is simulated to verify the stability of returns and output the portfolio optimization parameters. Mapping the energy type proportion data in the matching result data to the Black-Litterman model asset category; Setting constraint condition data based on the risk preference level data, generating scenario optimization data using the load fluctuation rate data, and generating reserve capacity service data in combination with the reserve capacity parameters in the bidding plan; Determine the energy storage configuration capacity and time-slot scheduling strategy based on the transferable load identifier, load volatility standard deviation, and portfolio optimization parameters, and generate energy storage configuration data; Based on the energy storage configuration data, the output includes an electricity package containing energy ratio data, time-of-use quotation strategy data, spare capacity service data and energy storage configuration data.

7. The power transaction matching recommendation system oriented to user needs is characterized by: The system comprises: The data acquisition and integration module is used to obtain user electricity load characteristics, price sensitivity and risk preference data, and integrate them into user electricity consumption data; A resonance group generation module is used to aggregate the user electricity consumption data in time and space dimensions to generate a resonance group with similar electricity consumption patterns and risk tolerance; A bidding agent module is used to generate a comprehensive electricity bidding plan based on the aggregate size, historical fulfillment rate and real-time market electricity consumption data of the resonance group through reinforcement learning bidding agent; A transaction matching module is used to match transactions based on the comprehensive electricity bidding plan, give priority to matching power generation resources with the closest electrical distance, introduce a green energy premium coefficient, and output a transaction result; The investment portfolio optimization analysis module is used to perform investment portfolio optimization analysis based on the transaction results and generate customized electricity packages including traditional energy, new energy and energy storage in combination with risk preference data.

Images