Full-scene cost analysis method and system for virtual power plant to participate in electric power commodity transaction

Abstract

The invention relates to the technical field of virtual power plant operation, in particular to a full-scene benefit analysis method and system for a virtual power plant to participate in electric power commodity transaction, and the method comprises the steps: collecting a new energy output curve and a load demand power curve, and constructing a scene feature distance matrix based on the spatial and temporal distribution features of the new energy output and the load demand; taking the distance matrix as input, finding a neuron closest to the sample as a winning neuron based on a self-organizing mapping neural network, judging the activation or inhibition effect of the neuron near the winning neuron, performing dynamic adjustment, and generating a new energy and load typical scene set; considering that the virtual power plant provides various electric power commodities such as electric energy, peak regulation and frequency modulation for power grid operation, and constructing a virtual power plant multi-type electric power commodity benefit analysis index system; a typical scene set and a benefit analysis index system are combined, probability distribution of each scene is considered, and a virtual power plant full-scene benefit analysis method fusing an entropy weight method and an analytic hierarchy process is provided.

Description

Technical Field

[0001] This invention relates to the field of virtual power plant operation technology, and in particular to a method and system for full-scenario cost analysis of virtual power plants participating in electricity commodity trading. Background Technology

[0002] With the acceleration of energy transition and the continuous improvement of electricity market mechanisms, Virtual Power Plants (VVPs), as key carriers aggregating heterogeneous resources such as distributed energy sources, energy storage systems, and controllable loads, are gradually becoming core participants in multi-level electricity market transactions. By coordinating internal resources to participate in the trading of various electricity commodities such as electrical energy, peak shaving, frequency regulation, and reserves, VVPs can significantly improve system flexibility and create synergistic benefits. However, the accurate assessment of their benefits faces the following systemic challenges.

[0003] Distributed resource output exhibits unique spatiotemporal distribution characteristics and variable output properties. Within virtual power plants, renewable energy generation and load demand exhibit strong randomness, intermittency, and geographical dispersion. Their output curves are non-stationary in the time dimension, showing significant intraday fluctuations, weather variations, and seasonal differences. Spatially, they are influenced by resource distribution and grid topology, forming complex relationships. Traditional scenario generation methods, such as typical day selection and Monte Carlo simulation, struggle to effectively characterize these high-dimensional, nonlinear spatiotemporal characteristics. This results in generated scenario sets that are insufficient in representing the actual operational situation, impacting the reliability of subsequent assessments.

[0004] Meanwhile, virtual power plants involve the evaluation of various electricity commodity transactions, including energy, peak shaving, frequency regulation, and reserve. These transactions have significantly different rules, pricing mechanisms, and settlement methods. Existing evaluation models mostly focus on a single energy market or simplified trading instruments, and resource dispatchability conflicts exist. The same resource struggles to adapt to the differentiated requirements of frequency regulation (millisecond-level response) and energy dispatch (hourly-level), hindering a comprehensive and accurate assessment of the combined benefits of various virtual power plant transactions. The information disclosed in this background section is intended only to enhance understanding of the overall background of this invention and should not be construed as an admission or implication in any way that such information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] This invention provides a method and system for full-scenario cost analysis of virtual power plants participating in electricity commodity trading, thereby effectively solving the problems in the background technology.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a full-scenario cost analysis method for virtual power plants participating in electricity commodity trading, comprising the following steps:

[0007] Collect the renewable energy output curves and load demand power curves from the historical operation data of the virtual power plant, and construct a scenario feature distance matrix for each typical scenario based on the spatiotemporal distribution characteristics of renewable energy output and load demand.

[0008] Using the distance matrix as input and one of its elements as a sample, a neuron that is closest to the sample is found as the winning neuron based on a self-organizing map neural network. The activation or inhibition effect of neurons near the winning neuron is judged and dynamically adjusted to generate a set of typical scenarios of new energy and load.

[0009] Considering that virtual power plants provide various types of electricity products such as electrical energy, peak shaving, and frequency regulation for grid operation, a benefit analysis index system for various types of electricity products of virtual power plants is constructed.

[0010] By combining a set of typical scenarios with a benefit analysis index system and considering the probability distribution of each scenario, a virtual power plant full-scenario benefit analysis method integrating entropy weight method and analytic hierarchy process is proposed.

[0011] Furthermore, the construction of a scene feature distance matrix based on the spatiotemporal distribution characteristics of new energy output and load demand includes:

[0012] The values ​​and trends of new energy power generation and user electricity load in different time periods are divided by day. T time segments are defined with a preset duration as each time node. Each day is used as a sample feature, and vector data are used to... It means that among them It is a feature value of the sample after the input data has been standardized;

[0013] In a self-organizing map neural network, each neuron is a vector composed of features in different proportions. , The weight of a feature is represented by an arbitrary neuron vector as input.

[0014] The Euclidean distance d between the neuron and the sample is calculated using the following formula:

[0015] ;

[0016] In the formula, x i Belongs to X, W i It belongs to W.

[0017] Further, the step of using the distance matrix as input, taking one of its elements as a sample, and finding the neuron closest to the sample as the winning neuron based on the self-organizing map neural network includes:

[0018] Based on Euclidean distance, find Zhongyu Minimum distance connection weight vector ,Right now:

[0019] ;

[0020] In the formula, The distance function is represented by Euclidean distance for datasets that connect distance attributes.

[0021] Furthermore, the step of determining the activation or inhibition effect of neurons near the winning neuron and dynamically adjusting it to generate a set of typical scenarios for new energy sources and loads includes the following steps:

[0022] The winning neuron is denoted as g. A neighborhood is defined to assess the activity of neurons near the winning neuron. This is based on the clustering mechanism of self-organizing map neural networks. When input sample data is fed into the network, each neuron calculates its similarity to the input sample. The neuron with the highest similarity and the best match for the input sample features is the winning neuron. Then, based on the spatial location of the winning neuron in the network topology, a local region with a specific range and shape is defined. This range dynamically shrinks during the training iteration process, thereby enabling the entire network to quickly converge towards the sample features. The formula is:

[0023] ;

[0024] ;

[0025] In the formula, The neighborhood function refers to the Gaussian function distribution model. Let be the distance between the i-th neuron and the j-th neuron; The initial neighborhood function value; The time constant of the neighborhood function; It is an exponentially decaying function, meaning that as the number of iterations and learning iterations increases, the bandwidth of the neighborhood function decreases, gradually reducing the influence of the winning neuron. After determining whether neurons surrounding the winning neuron g are activated or inhibited through the neighborhood function, the activation or inhibition behavior is further determined. The weight vector values ​​of the neurons are adjusted based on the input sample; activation brings the neuron closer to the input sample, while inhibition moves it away. This is achieved through the following adjustment formula:

[0026] ;

[0027] ;

[0028] In the formula: This represents the change in weight of the j-th neuron; It is an exponentially decaying function that controls the adjustment of neurons in the SOM, also known as the learning rate. This is the error term between the sample and the weights; The initial learning rate; It is the time constant of the learning rate;

[0029] Through the aforementioned activation or inhibition behaviors, different samples are continuously input, and the autonomous reinforcement learning process is repeated until the neuron structure is stable after training or the maximum number of iterations is reached. At this point, each input sample has a winning neuron, which outputs the clustering result.

[0030] Furthermore, the virtual power plant full-scenario benefit analysis method integrating entropy weight method and analytic hierarchy process includes:

[0031] Taking into account the total revenue of virtual power plants from various electricity commodities, the delayed revenue from grid investment, the cost savings from traditional reserves, the cost reduction from grid line losses, and the user-side revenue, an economic benefit analysis index for virtual power plants participating in various electricity commodity transactions is established.

[0032] Taking into account factors such as the improvement rate of grid frequency stability, the reduction rate of system peak-valley difference, and the rate of meeting ramp-up demand, we construct a grid operation benefit analysis index for virtual power plants participating in multiple types of electricity commodity transactions.

[0033] Taking into account carbon dioxide emissions and renewable energy absorption rate, we construct environmental benefit analysis indicators for virtual power plants participating in various types of electricity commodity transactions, integrate various benefit analysis indicators, and construct a benefit analysis indicator system for virtual power plants participating in various types of electricity commodity transactions.

[0034] Based on the generation of typical scenario sets through self-organizing map neural network clustering, a full-scenario cost analysis model covering all typical scenarios is constructed, including:

[0035] Constructing economic benefit indicators:

[0036] Total revenue from diversified services of virtual power plants:

[0037] ;

[0038] In the formula: Indicates peak shaving revenue; For frequency modulation revenue; Indicates reserve revenue; For the benefits of climbing, This represents the revenue from electrical energy, calculated according to the corresponding market rules, reflecting the overall economic return of the VPP operator;

[0039] Operating cost yield:

[0040] ;

[0041] In the formula: express Annual operating costs, including Dispatch fee, communication operation and maintenance fee, platform management fee, and user incentive fee; this formula is used to measure Profitability, of which This indicates that the operation is feasible;

[0042] Power grid investment delays returns:

[0043] ;

[0044] In the formula: Indicates none Investment in power grid expansion / upgrade (RMB); For access Post-grid investment; For the design life of the power grid; By reducing peak shaving and valley filling and providing backup power, the demand for grid capacity can be reduced, thus delaying investment.

[0045] Cost savings from traditional standby:

[0046] ;

[0047] In the formula: This represents the standby cost of traditional thermal power units, expressed in yuan / kWh. for Backup cost, in yuan / kWh; This represents the system's annual reserve capacity requirement, in kW; 8760 represents the number of hours per year. It reflects Cost advantages of replacing traditional backup resources;

[0048] Cost savings due to power grid line loss:

[0049] ;

[0050] In the formula: Indicates none Network line loss rate; Line loss rate after VPP connection; This indicates the annual power supply of the distribution network; Average market price of electricity; Reduce transmission losses by consuming the waste locally;

[0051] User resource participation benefits:

[0052] ;

[0053] In the formula: This indicates that the user is adjusting the power / capacity for the i-th type of service; Serving the i-th class Incentive unit price;

[0054] Constructing social benefit indicators:

[0055] System frequency stability improvement rate:

[0056] ;

[0057] In the formula: This represents the standard deviation of the system frequency without VPP. For access Post-frequency standard deviation; the smaller the frequency standard deviation, the more stable the system. Reduce frequency fluctuations by rapidly adjusting the frequency.

[0058] System peak-to-valley reduction rate:

[0059] ;

[0060] In the formula: This represents the maximum peak-to-valley difference of the system without VPP; The peak-valley difference after connecting to VPP; the smaller the peak-valley difference, the more balanced the power grid supply and demand.

[0061] Hill Climbing Requirement Satisfaction Rate:

[0062] ;

[0063] In the formula This indicates the ramping power provided by the VPP; This refers to the ramp-up demand of the system caused by sudden changes in renewable energy output / load fluctuations; a satisfaction rate of >90% indicates that wind and solar power curtailment or power supply gaps can be effectively avoided.

[0064] Backup response compliance rate:

[0065] ;

[0066] In the formula: This indicates the number of times the VPP has completed a standby call according to the instructions; This represents the total number of times the power grid backup system has been called up; a compliance rate of 95% indicates that the VPP can serve as a reliable backup resource, improving the system's ability to respond to faults.

[0067] Load peak-to-valley fluctuation:

[0068] ;

[0069] In the formula, These represent the maximum and minimum loads of the virtual power plant's energy supply area, respectively.

[0070] Constructing environmental benefit indicators:

[0071] Carbon dioxide emissions:

[0072] ;

[0073] In the formula: The carbon dioxide emission intensity of electricity purchased from gas turbines and the main power grid, respectively;

[0074] New energy consumption rate:

[0075] ;

[0076] In the formula, For the renewable energy consumption rate; These are photovoltaic power generation, wind turbine power generation, and gas turbine power generation, respectively.

[0077] Furthermore, the virtual power plant full-scenario benefit analysis method that integrates entropy weight method and analytic hierarchy process also includes:

[0078] Based on a given operating scenario, the indicators are first positively normalized, that is, the dimensions of the indicators are removed, and the entropy and weight values ​​of the indicators are calculated.

[0079] The subjective weights of the indicators are calculated using the analytic hierarchy process (AHP). An indicator importance judgment matrix is ​​constructed based on expert experience. The consistency index and consistency ratio are obtained by solving the matrix based on the largest eigenvalue. Then, the geometric mean of each row of elements is calculated and normalized to obtain the subjective weight vector.

[0080] By combining the objective weights obtained by the entropy weight method with the subjective weights obtained by the analytic hierarchy process, a comprehensive weight for a given scenario can be obtained.

[0081] Considering various schemes for virtual power plants to provide electricity trading and trading of multiple types of electricity commodities, calculate the comprehensive score of the virtual power plant trading scheme for a specific scenario;

[0082] Finally, based on the probability distribution of typical virtual power plant scenarios, considering the probability of different scenarios occurring, the benefit evaluation results of all scenarios are comprehensively evaluated by using probability weighted summation to obtain the full-scenario benefit analysis results.

[0083] Given a specific operational scenario t, the objective weights are determined using the entropy weight method. First, the indicators are positively normalized, i.e., the dimensions of the indicators are removed.

[0084] For cost-related indicators, the standardization is as follows:

[0085] ;

[0086] For efficiency-related indicators, the standardization is as follows:

[0087] ;

[0088] In the formula, To normalize decision matrix B t elements, Let j be the index value of the j-th index in the i-th scheme. and These are the maximum and minimum values ​​of the j-th indicator in the i-th scheme, respectively;

[0089] Then, the entropy and weight values ​​of the indicators are calculated:

[0090] ;

[0091] ;

[0092] ;

[0093] In the formula, Let j be the information entropy of the t-th scenario. To obtain the objective weight of index j, For process decision variables;

[0094] Secondly, the subjective weights of the indicators are calculated using the analytic hierarchy process (AHP), and an indicator importance judgment matrix is ​​constructed based on expert experience.

[0095] ;

[0096] In the formula, To determine the (i, j)th element in the matrix;

[0097] Based on the solution of the largest eigenvalue of the matrix, the consistency index and consistency ratio are obtained as follows:

[0098] ;

[0099] ;

[0100] In the formula, Let CI be the largest eigenvalue, CR be the consistency index, and RI be the average random consistency index; when When <0.1, the matrix is ​​considered to be If the degree of inconsistency is within the acceptable range, then the original judgment matrix A needs to be adjusted;

[0101] Calculate the geometric mean of each row's elements and normalize it to obtain the subjective weight vector:

[0102] ;

[0103] ;

[0104] In the formula, For subjective weight vectors, The j-th element of the subjective weight vector Let be the set average of the j-th row of matrix A;

[0105] By combining the objective weights obtained by the entropy weight method with the subjective weights obtained by the analytic hierarchy process, we obtain the comprehensive weight of t in a certain scenario:

[0106] ;

[0107] In the formula, The overall weight of scenario t, Let the information entropy of index j be the information entropy of the t-th scenario;

[0108] Considering various options such as virtual power plants providing electricity trading and trading in multiple types of electricity commodities, for a specific scenario t, calculate the comprehensive score of virtual power plant trading scheme i.

[0109] ;

[0110] In the formula, The overall score of transaction i in scenario t;

[0111] Based on the probability distribution of typical virtual power plant scenarios, considering the probability of different scenarios occurring, a comprehensive evaluation of the benefit assessment results of all scenarios is conducted using probability-weighted summation to obtain the full-scenario benefit analysis results:

[0112] ;

[0113] In the formula, This represents the overall score of transaction solution i across all scenarios. Let be the probability of scenario t occurring.

[0114] This invention also includes a full-scenario cost analysis system for virtual power plants participating in electricity commodity trading, using the method described above, the system comprising:

[0115] The feature distance calculation unit is used to collect the renewable energy output curve and load demand power curve from the historical operation data of the virtual power plant, and construct the scene feature distance matrix based on the spatiotemporal distribution characteristics of renewable energy output and load demand.

[0116] The scene generation unit is used to take the distance matrix as input, take one of its elements as a sample, find the neuron closest to the sample as the winning neuron based on the self-organizing map neural network, and judge the activation or inhibition effect of the neurons near the winning neuron, make dynamic adjustments, and generate a set of typical new energy and load scenarios.

[0117] The analysis index construction unit is used to consider the various types of electricity commodities that virtual power plants provide to the power grid operation, such as electrical energy, peak shaving, and frequency regulation, and to construct a benefit analysis index system for various types of electricity commodities of virtual power plants.

[0118] The evaluation unit is used to combine a set of typical scenarios with a benefit analysis index system, consider the probability distribution of each scenario, and propose a virtual power plant full-scenario benefit analysis method that integrates the entropy weight method and the analytic hierarchy process.

[0119] The present invention also includes a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described above.

[0120] The present invention also includes a storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described above.

[0121] The beneficial effects of this invention are as follows: by integrating Euclidean distance and self-organizing map (SOM) neural network clustering for full-scenario benefit analysis, and by using an autonomous reinforcement learning mechanism to achieve intelligent dimensionality reduction and topology preservation in multi-commodity trading scenarios, accurate clustering of virtual power plant multi-type electricity commodity trading scenarios is achieved. For virtual power plants participating in multiple types of electricity commodity trading such as energy, peak shaving, frequency regulation, and reserve, a full-scenario benefit analysis model for virtual power plants participating in multiple types of electricity commodity trading is constructed, providing quantitative decision-making basis for virtual power plants participating in multi-market trading collaborative operation, and achieving effective guidance for virtual power plants participating in multiple types of electricity commodity trading. Attached Figure Description

[0122] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0123] Figure 1 This is a flowchart of the method in Example 1;

[0124] Figure 2 This is a schematic diagram of the system structure in Example 1;

[0125] Figure 3 This is a flowchart of the method in Example 2;

[0126] Figure 4 This is a schematic diagram of the structure of the computer device of the present invention. Detailed Implementation

[0127] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0128] Example 1:

[0129] like Figure 1 As shown: A full-scenario cost analysis method for virtual power plants participating in electricity commodity trading, including the following steps:

[0130] Collect the renewable energy output curves and load demand power curves from the historical operation data of the virtual power plant, and construct a scenario feature distance matrix for each typical scenario based on the spatiotemporal distribution characteristics of renewable energy output and load requirements;

[0131] Using the distance matrix as input and one of its elements as a sample, the system finds the neuron closest to the sample based on the self-organizing map neural network as the winning neuron, and judges the activation or inhibition effect of neurons near the winning neuron, making dynamic adjustments to generate a set of typical scenarios for new energy and load.

[0132] Considering that virtual power plants provide various types of electricity products such as electrical energy, peak shaving, and frequency regulation for grid operation, a benefit analysis index system for various types of electricity products of virtual power plants is constructed.

[0133] By combining a set of typical scenarios with a benefit analysis index system and considering the probability distribution of each scenario, a virtual power plant full-scenario benefit analysis method integrating entropy weight method and analytic hierarchy process is proposed.

[0134] By integrating Euclidean distance and self-organizing map (SOM) neural network clustering for full-scenario benefit analysis, and through an autonomous reinforcement learning mechanism to achieve intelligent dimensionality reduction and topology preservation in multi-commodity trading scenarios, accurate clustering of virtual power plants in various electricity commodity trading scenarios is achieved. For virtual power plants participating in various electricity commodity trading such as energy, peak shaving, frequency regulation, and reserve, a full-scenario benefit analysis model for virtual power plants participating in various electricity commodity trading is constructed, providing quantitative decision-making basis for virtual power plants participating in multi-market trading collaborative operation and achieving effective guidance for virtual power plants participating in various electricity commodity trading.

[0135] In this embodiment, the spatiotemporal distribution characteristics of new energy output and load requirements are used to construct a scene feature distance matrix, including:

[0136] The data on renewable energy generation capacity, user electricity load, and their changing trends are divided by day, with T time segments defined by a preset duration, such as 15 minutes. Each day serves as a sample feature, and vector data are used to analyze these features. It means that among them It is a feature value of the sample after the input data has been standardized;

[0137] In a self-organizing map neural network, each neuron is a vector composed of features in different proportions. , The weight of a feature is represented by an arbitrary neuron vector as input.

[0138] The Euclidean distance d between the neuron and the sample is calculated using the following formula:

[0139]

[0140] The process involves using a distance matrix as input and one of its elements as a sample. A self-organizing map neural network is used to find the neuron closest to the sample as the winning neuron. This includes:

[0141] Based on Euclidean distance, find Zhongyu Minimum distance connection weight vector ,Right now:

[0142]

[0143] In the formula, The distance function is represented by Euclidean distance for datasets that connect distance attributes.

[0144] The activation or inhibition effect of neurons near the winning neuron is determined, and dynamic adjustments are made to generate a set of typical scenarios for new energy sources and loads. This includes the following steps:

[0145] Let the winning neuron be g. We determine a neighborhood to assess the activity of neurons near the winning neuron, using the following formula:

[0146]

[0147]

[0148] In the formula, The neighborhood function refers to the Gaussian function distribution model. Let be the distance between the i-th neuron and the j-th neuron; The initial neighborhood function value; The time constant of the neighborhood function; It is an exponentially decaying function, meaning that as the number of iterations and learning iterations increases, the bandwidth of the neighborhood function decreases, gradually reducing the influence of the winning neuron. After determining whether neurons surrounding the winning neuron g are activated or inhibited through the neighborhood function, the activation or inhibition behavior is further determined. The weight vector values ​​of the neurons are adjusted based on the input sample; activation brings the neuron closer to the input sample, while inhibition moves it away. This is achieved through the following adjustment formula:

[0149]

[0150]

[0151] In the formula: This represents the change in weight of the j-th neuron; It is an exponentially decaying function that controls the adjustment of neurons in the SOM, also known as the learning rate. This is the error term between the sample and the weights; The initial learning rate; It is the time constant of the learning rate;

[0152] Through the aforementioned activation or inhibition behaviors, different samples are continuously input, and the autonomous reinforcement learning process is repeated until the neuron structure is stable after training or the maximum number of iterations is reached. At this point, each input sample has a winning neuron, which outputs the clustering result.

[0153] A set of typical scenarios is generated by self-organizing map neural network clustering, and a full-scenario cost analysis model covering all typical scenarios is constructed. The full-scenario cost analysis model includes:

[0154] Taking into account the total revenue of virtual power plants from various electricity commodities, the delayed revenue from grid investment, the cost savings from traditional reserves, the cost reduction from grid line losses, and the user-side revenue, an economic benefit analysis index for virtual power plants participating in various electricity commodity transactions is established.

[0155] Taking into account factors such as the improvement rate of grid frequency stability, the reduction rate of system peak-valley difference, and the rate of meeting ramp-up demand, we construct a grid operation benefit analysis index for virtual power plants participating in multiple types of electricity commodity transactions.

[0156] Taking into account carbon dioxide emissions and renewable energy absorption rate, we construct environmental benefit analysis indicators for virtual power plants participating in various types of electricity commodity transactions, integrate various benefit analysis indicators, and construct a benefit analysis indicator system for virtual power plants participating in various types of electricity commodity transactions.

[0157] (1) Economic benefit indicators

[0158] Total revenue from diversified services of virtual power plants:

[0159]

[0160] In the formula: Indicates peak shaving revenue; For frequency modulation revenue; Indicates reserve revenue; For the benefits of climbing, This represents the revenue from electrical energy, calculated according to the corresponding market rules (e.g., frequency regulation is calculated as "capacity + electricity consumption", and standby is calculated as "committed capacity x duration"), reflecting the overall economic return of the VPP operator.

[0161] Operating cost yield:

[0162]

[0163] In the formula: express Annual operating costs (RMB), including Dispatch fees, communication operation and maintenance fees, platform management fees, and user incentive fees; this formula is used to measure the profitability efficiency of VPPs. This indicates that the operation is feasible.

[0164] Power grid investment delays returns:

[0165]

[0166] In the formula: This represents the investment (in yuan) for grid expansion / upgrade without VPP. Investment in the power grid after VPP integration (in RMB); The power grid's design life (in years, typically 20-30 years); Peak shaving and valley filling, along with providing backup power, reduce grid capacity demand and delay investment.

[0167] Cost savings from traditional standby:

[0168]

[0169] In the formula: This represents the standby cost of traditional thermal power units (yuan / (kWh)). VPP backup cost (RMB / (kWh)); This represents the system's annual reserve capacity requirement (kW); 8760 represents the number of hours per year. This reflects the cost advantage of VPP replacing traditional backup resources.

[0170] Cost savings due to power grid line loss:

[0171]

[0172] In the formula: This indicates the distribution network line loss rate (%) without VPP. Line loss rate (%) after VPP connection; This represents the annual power supply of the distribution network (kWh). Average market price of electricity (RMB / kWh); VPP reduces transmission losses through "local consumption".

[0173] User resource participation benefits:

[0174]

[0175] In the formula: This indicates the amount of electricity / capacity (kWh / kW) that a user adjusts when participating in service type i. The incentive unit price for VPP services of type i (RMB / kW·h or RMB / (kW·month)); such as capacity incentives for user energy storage to participate in frequency regulation, and electricity incentives for load to participate in peak shaving.

[0176] (2) Social benefit indicators

[0177] System frequency stability improvement rate:

[0178]

[0179] In the formula: This represents the standard deviation of the system frequency (Hz) without VPP. The standard deviation of the frequency (Hz) after connecting to VPP; the smaller the standard deviation of the frequency, the more stable the system; VPP reduces frequency fluctuations through fast frequency tuning.

[0180] System peak-to-valley reduction rate:

[0181]

[0182] In the formula: This represents the maximum peak-to-valley difference (kW) of the system without VPP. The peak-to-valley difference (kW) after connecting to VPP; the smaller the peak-to-valley difference, the more balanced the power grid supply and demand, avoiding "power shortage during peak hours and equipment idleness during valley hours".

[0183] Hill Climbing Requirement Satisfaction Rate:

[0184]

[0185] In the formula This indicates the ramping power (kW) provided by the VPP; The ramp-up demand (kW) caused by sudden changes in renewable energy output / load fluctuations in the system; a satisfaction rate of >90% indicates that "wind and solar curtailment" or "power supply gap" can be effectively avoided.

[0186] Backup response compliance rate:

[0187]

[0188] In the formula: This indicates the number of times the VPP has completed a standby call according to the instructions; This represents the total number of times the power grid backup system has been called up; a compliance rate of 95% indicates that the VPP can serve as a reliable backup resource, improving the system's ability to respond to faults.

[0189] Load peak-to-valley fluctuation:

[0190] Virtual power plants can issue charging and discharging commands to electric vehicles, which can alleviate peak load pressure to a certain extent, increase load during off-peak hours, smooth the load curve, and stabilize the local power grid load. Therefore, the load peak-valley difference fluctuation rate can be used as a social benefit indicator.

[0191]

[0192] In the formula, These represent the maximum and minimum loads of the virtual power plant's energy supply area, respectively.

[0193] (3) Environmental benefit indicators

[0194] Carbon dioxide emissions:

[0195] Traditional power sources consume large amounts of fossil fuels during operation, resulting in greenhouse gases. Virtual power plants, acting as distributed energy aggregators, optimize unit output based on dispatch targets, absorbing clean energy and reducing carbon emissions. Carbon dioxide emissions are used as the primary indicator of their environmental benefits. The formula is:

[0196]

[0197] In the formula: The carbon dioxide emission intensity is calculated separately for gas turbines and electricity purchased from the main power grid.

[0198] New energy consumption rate:

[0199] The renewable energy consumption rate is the ratio of wind and solar power generation to the total power generation of the virtual power plant. The power generation is the actual output of the virtual power plant units, without considering energy storage or grid purchases. The formula is:

[0200]

[0201] In the formula, For the renewable energy consumption rate; These are photovoltaic power generation, wind turbine power generation, and gas turbine power generation, respectively.

[0202] Combining a set of typical scenarios and a benefit analysis index system, and considering the probability distribution of each scenario, a virtual power plant full-scenario benefit analysis method integrating the entropy weight method and the analytic hierarchy process is proposed, including:

[0203] Based on a given operating scenario, the indicators are first positively normalized, that is, the dimensions of the indicators are removed, and the entropy and weight values ​​of the indicators are calculated.

[0204] The subjective weights of the indicators are calculated using the analytic hierarchy process (AHP). An indicator importance judgment matrix is ​​constructed based on expert experience. The consistency index and consistency ratio are obtained by solving the matrix based on the largest eigenvalue. Then, the geometric mean of each row of elements is calculated and normalized to obtain the subjective weight vector.

[0205] By combining the objective weights obtained by the entropy weight method with the subjective weights obtained by the analytic hierarchy process, a comprehensive weight for a given scenario can be obtained.

[0206] Considering various schemes for virtual power plants to provide electricity trading and trading of multiple types of electricity commodities, calculate the comprehensive score of the virtual power plant trading scheme for a specific scenario;

[0207] Finally, based on the probability distribution of typical virtual power plant scenarios, considering the probability of different scenarios occurring, the benefit evaluation results of all scenarios are comprehensively evaluated by using probability weighted summation to obtain the full-scenario benefit analysis results.

[0208] Given a specific operational scenario t, the objective weights are determined using the entropy weight method. First, the indicators are positively normalized, i.e., the dimensions of the indicators are removed.

[0209] For cost-related indicators, the standardization is as follows:

[0210] (20)

[0211] For efficiency-related indicators, the standardization is as follows:

[0212] (twenty one)

[0213] In the formula, To normalize decision matrix B t elements, Let j be the index value of the j-th index in the i-th scheme. and These are the maximum and minimum values ​​of the j-th indicator in the i-th scheme, respectively.

[0214] Then, the entropy and weight values ​​of the indicators are calculated:

[0215] (twenty two)

[0216] (twenty three)

[0217] (twenty four)

[0218] In the formula, Let j be the information entropy of index j in the t-th scenario. To obtain the objective weight of index j, These are process decision variables.

[0219] Secondly, the subjective weights of the indicators are calculated using the analytic hierarchy process (AHP), and an indicator importance judgment matrix is ​​constructed based on expert experience.

[0220] (25)

[0221] In the formula, To determine the (i, j)th element in the matrix.

[0222] Based on the solution of the largest eigenvalue of the matrix, the consistency index and consistency ratio are obtained as follows:

[0223] (26)

[0224] (27)

[0225] In the formula, Let CI be the largest eigenvalue, CR be the consistency index, and RI be the average random consistency index. When <0.1, the matrix is ​​considered to be The degree of inconsistency is within acceptable limits.

[0226] Calculate the geometric mean of each row's elements and normalize it to obtain the subjective weight vector:

[0227] (28)

[0228] (29)

[0229] In the formula, For subjective weight vectors, The j-th element of the subjective weight vector Let be the set of tie values ​​for the j-th row of matrix A.

[0230] By combining the objective weights obtained by the entropy weight method with the subjective weights obtained by the analytic hierarchy process, we obtain the comprehensive weight of t in a certain scenario:

[0231] (30)

[0232] In the formula, The overall weight of scenario t, Let be the information entropy of index j in the t-th scenario.

[0233] For a specific scenario t, calculate the comprehensive score of virtual power plant trading scheme i.

[0234] (31)

[0235] In the formula, Let be the overall score of transaction i in scenario t. Considering the probabilities of different scenarios, the benefits of all scenarios are comprehensively evaluated to obtain the overall scenario benefit analysis results:

[0236] (32)

[0237] In the formula, This represents the overall score of transaction solution i across all scenarios. Let be the probability of scenario t occurring.

[0238] like Figure 2 As shown, this implementation example also includes a full-scenario cost analysis system for virtual power plants participating in electricity commodity trading. Using the method described above, the system includes:

[0239] The feature distance calculation unit is used to collect the renewable energy output curve and load demand power curve from the historical operation data of the virtual power plant, and construct the scene feature distance matrix based on the spatiotemporal distribution characteristics of renewable energy output and load demand.

[0240] The scene generation unit takes a distance matrix as input, one of its elements as a sample, and finds the neuron closest to the sample based on a self-organizing map neural network as the winning neuron. It then judges the activation or inhibition effect of neurons near the winning neuron and makes dynamic adjustments to generate a set of typical new energy and load scenarios.

[0241] The analysis index construction unit is used to consider the various types of electricity commodities that virtual power plants provide to the power grid operation, such as electrical energy, peak shaving, and frequency regulation, and to construct a benefit analysis index system for various types of electricity commodities of virtual power plants.

[0242] The evaluation unit is used to combine a set of typical scenarios with a benefit analysis index system, consider the probability distribution of each scenario, and propose a virtual power plant full-scenario benefit analysis method that integrates the entropy weight method and the analytic hierarchy process.

[0243] Example 2:

[0244] like Figure 3 As shown, this embodiment includes a full-scenario cost analysis method for virtual power plants participating in electricity commodity trading, comprising the following steps:

[0245] Step (1): Collect historical renewable energy output data of the virtual power plant, standardize it to form a renewable energy output curve and load demand power curve on a daily basis; divide the time into 96 time segments with 15-minute intervals, and use each day as a sample feature, using vector... It means that among them It refers to a feature value of a sample after the input data has been standardized. In a self-organizing map neural network, each neuron is a vector composed of features in different proportions. , The weight of a feature is represented by an arbitrary neuron vector. The Euclidean distance between the neuron and the sample is calculated. For each typical scenario i, the Euclidean distance is calculated based on the spatiotemporal distribution characteristics of new energy output and load requirements.

[0246]

[0247] In equation (1), It is a feature value of a sample after the input data has been standardized. The weight of a feature is represented by the square of the Euclidean distance. For simplicity, the square of the Euclidean distance is calculated, and the neuron g that is closest to the sample is found based on the obtained Euclidean distance matrix.

[0248] Step (3): Find the sample vector corresponding to the shortest distance. Let g be the winning neuron. For each neuron, besides the winning neuron being activated, other neurons are actually activated or inhibited to varying degrees. Therefore, a neighborhood is determined to judge the activity of neurons surrounding the winning neuron. The formula is:

[0249]

[0250]

[0251] In formula (2): The neighborhood function can refer to any distribution model: the Mexican hat model, the Bubble model, the Gaussian distribution model, and the top hat function model. Here, we choose the Gaussian function distribution model. It is an exponentially decaying function, which represents the control bandwidth. That is, as the number of iterations and learning times increases, the bandwidth of the neighborhood function decreases, the influence of the winning neuron gradually decreases, and the entire neural network tends to change smoothly.

[0252] In formula (3): It is a constant. After determining whether neurons surrounding the winning neuron g are activated or inhibited through the neighborhood function, the activation or inhibition behavior is further determined. In SOM, activation and inhibition behavior are achieved by adjusting the neuron's weight vector values ​​based on the input sample. Activation brings the neuron closer to the input sample, while inhibition moves it away from the input sample. This is achieved through the adjustment formula:

[0253]

[0254]

[0255] In equation (4): The weight vector of neuron j The amount of update in the current iteration; Let x be the topological neighborhood function, representing the strength of the neighborhood relationship between neuron j and the winning neuron i, ensuring that the winning neuron and all neurons in its neighborhood are adjusted toward the input vector x. It is an exponentially decaying function that decreases with the number of iterations t, controlling the adjustment of neurons in the SOM, and is also called the learning rate function;

[0256] Through the aforementioned activation or inhibition behaviors, different samples are continuously input, and the autonomous reinforcement learning process is repeated until the neuron structure is stable after training or the maximum number of iterations is reached. At this point, each input sample has a winning neuron, which outputs the clustering result.

[0257] Step (4): Analyze the full-scenario benefits of virtual power plants participating in electricity commodity trading, and construct a benefit evaluation index system, including economic, environmental, and social aspects:

[0258] Establish a benefit evaluation index system for multi-type electricity commodity transactions involving virtual power plants. The evaluation indexes include:

[0259] (1) Economic benefit indicators

[0260] Total revenue from diversified services of virtual power plants:

[0261]

[0262] In the formula: Indicates peak shaving revenue; For frequency modulation revenue; Indicates reserve revenue; For the benefits of climbing, This represents the revenue from electrical energy, calculated according to the corresponding market rules, reflecting the overall economic return of the VPP operator.

[0263] Operating cost yield:

[0264]

[0265] In the formula: express Annual operating costs (RMB), including Dispatch fees, communication operation and maintenance fees, platform management fees, and user incentive fees; this formula is used to measure the profitability efficiency of VPPs. This indicates that the operation is feasible;

[0266] Power grid investment delays returns:

[0267]

[0268] In the formula: This represents the investment (in yuan) for grid expansion / upgrade without VPP. Investment in the power grid after VPP integration (in RMB); For the design life of the power grid; Peak shaving and valley filling, along with providing backup power, reduce grid capacity demand and delay investment.

[0269] Cost savings from traditional standby:

[0270]

[0271] In the formula: This represents the standby cost of traditional thermal power units (yuan / (kWh)). VPP backup cost (RMB / (kWh)); This represents the system's annual reserve capacity requirement (kW); 8760 represents the number of hours per year. This reflects the cost advantage of VPP replacing traditional backup resources;

[0272] Cost savings due to power grid line loss:

[0273]

[0274] In the formula: This indicates the distribution network line loss rate (%) without VPP. Line loss rate (%) after VPP connection; This represents the annual power supply of the distribution network (kWh). Average market price of electricity (RMB / kWh); VPP reduces transmission losses through "local consumption".

[0275] User resource participation benefits:

[0276]

[0277] In the formula: This indicates the amount of electricity / capacity (kWh / kW) that a user adjusts when participating in service type i. The incentive unit price for VPP services of type i (RMB / kW·h or RMB / (kW·month)); such as capacity incentives for user energy storage to participate in frequency regulation, and electricity incentives for load to participate in peak shaving.

[0278] (2) Social benefit indicators

[0279] System frequency stability improvement rate:

[0280]

[0281] In the formula: This represents the standard deviation of the system frequency (Hz) without VPP. The standard deviation of the frequency (Hz) after connecting to VPP; the smaller the standard deviation of the frequency, the more stable the system; VPP reduces frequency fluctuations through fast frequency tuning.

[0282] System peak-to-valley reduction rate:

[0283]

[0284] In the formula: This represents the maximum peak-to-valley difference (kW) of the system without VPP. The peak-to-valley difference (kW) after connecting to VPP; the smaller the peak-to-valley difference, the more balanced the power grid supply and demand, avoiding "power shortage during peak hours and equipment idleness during valley hours".

[0285] Hill Climbing Requirement Satisfaction Rate:

[0286]

[0287] In the formula This indicates the ramping power (kW) provided by the VPP; The ramp-up demand (kW) caused by sudden changes in renewable energy output / load fluctuations in the system; a satisfaction rate of >90% indicates that "wind and solar curtailment" or "power supply gap" can be effectively avoided.

[0288] Backup response compliance rate:

[0289]

[0290] In the formula: This indicates the number of times the VPP has completed a standby call according to the instructions; This represents the total number of times the power grid backup system has been called up; a compliance rate of 95% indicates that the VPP can serve as a reliable backup resource, improving the system's ability to respond to faults.

[0291] Load peak-valley fluctuation

[0292] Virtual power plants can issue charging and discharging commands to electric vehicles, which can alleviate peak load pressure to a certain extent, increase load during off-peak hours, smooth the load curve, and stabilize the local power grid load. Therefore, the load peak-valley difference fluctuation rate can be used as a social benefit indicator.

[0293]

[0294] In the formula, These represent the maximum and minimum loads of the virtual power plant's energy supply area, respectively.

[0295] (3) Environmental benefit indicators

[0296] carbon dioxide emissions

[0297] Traditional power sources consume large amounts of fossil fuels during operation, resulting in greenhouse gases. Virtual power plants, acting as distributed energy aggregators, optimize unit output based on dispatch targets, absorbing clean energy and reducing carbon emissions. Carbon dioxide emissions are used as the primary indicator of their environmental benefits. The formula is:

[0298]

[0299] In the formula: The carbon dioxide emission intensity is calculated separately for gas turbines and electricity purchased from the main power grid.

[0300] (2) New energy consumption rate

[0301] The renewable energy consumption rate is the ratio of wind and solar power generation to the total power generation of the virtual power plant. The power generation is the actual output of the virtual power plant units, without considering energy storage or grid purchases. The formula is:

[0302]

[0303] In the formula, For the renewable energy consumption rate; These are photovoltaic power generation, wind turbine power generation, and gas turbine power generation, respectively.

[0304] Step (5): Combining the typical scenario set and benefit analysis index system, and considering the probability distribution of each scenario, a virtual power plant full-scenario benefit analysis method integrating entropy weight method and analytic hierarchy process is proposed, including:

[0305] Given a specific operational scenario t, the objective weights are determined using the entropy weight method. First, the indicators are positively normalized, i.e., the dimensions of the indicators are removed.

[0306] For cost-related indicators, the standardization is as follows:

[0307] (20)

[0308] For efficiency-related indicators, the standardization is as follows:

[0309] (twenty one)

[0310] In the formula, To normalize decision matrix B t elements, Let j be the index value of the j-th index in the i-th scheme. and These are the maximum and minimum values ​​of the j-th indicator in the i-th scheme, respectively.

[0311] Then, the entropy and weight values ​​of the indicators are calculated:

[0312] (twenty two)

[0313] (twenty three)

[0314] (twenty four)

[0315] In the formula, Let j be the information entropy of the t-th scenario. To obtain the objective weight of index j, These are process decision variables.

[0316] Secondly, the subjective weights of the indicators are calculated using the analytic hierarchy process (AHP), and an indicator importance judgment matrix is ​​constructed based on expert experience.

[0317] (25)

[0318] In the formula, To determine the (i, j)th element in the matrix.

[0319] Based on the solution of the largest eigenvalue of the matrix, the consistency index and consistency ratio are obtained as follows:

[0320] (26)

[0321] (27)

[0322] In the formula, Let CI be the largest eigenvalue, CR be the consistency index, and RI be the average random consistency index. When <0.1, the matrix is ​​considered to be The degree of inconsistency is within acceptable limits.

[0323] Calculate the geometric mean of each row's elements and normalize it to obtain the subjective weight vector:

[0324] (28)

[0325] (29)

[0326] In the formula, For subjective weight vectors, The j-th element of the subjective weight vector Let be the set of tie values ​​for the j-th row of matrix A.

[0327] By combining the objective weights obtained by the entropy weight method with the subjective weights obtained by the analytic hierarchy process, we obtain the comprehensive weight of t in a certain scenario:

[0328] (30)

[0329] In the formula, The overall weight of scenario t, Let be the information entropy of index j in the t-th scenario.

[0330] For a specific scenario t, calculate the comprehensive score of virtual power plant trading scheme i.

[0331] (31)

[0332] In the formula, Let be the overall score of transaction i in scenario t. Considering the probabilities of different scenarios, the benefits of all scenarios are comprehensively evaluated to obtain the overall scenario benefit analysis results:

[0333] (32)

[0334] In the formula, This represents the overall score of transaction solution i across all scenarios. Let be the probability of scenario t occurring.

[0335] Please see Figure 4 The diagram shows a structural schematic of a computer device provided in an embodiment of this application. An embodiment of this application provides a computer device 400, including a processor 410 and a memory 420. The memory 420 stores a computer program executable by the processor 410. When the computer program is executed by the processor 410, it performs the method described above.

[0336] This application embodiment also provides a storage medium 430, on which a computer program is stored, and the computer program is executed by a processor 410 to perform the above method.

[0337] The storage medium 430 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0338] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. "A plurality of" means two or more, unless otherwise explicitly specified.

[0339] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0340] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0341] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0342] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0343] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0344] Those skilled in the art will understand that all or part of the steps of the methods described in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it includes one or a combination of the steps of the method embodiments.

[0345] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for full-scenario cost analysis of virtual power plants participating in electricity commodity trading, characterized in that, Includes the following steps: Collect the renewable energy output curve and load demand power curve from the historical operation data of the virtual power plant, and construct the scene feature distance matrix based on the spatiotemporal distribution characteristics of renewable energy output and load demand; Using the distance matrix as input and one of its elements as a sample, a neuron that is closest to the sample is found as the winning neuron based on a self-organizing map neural network. The activation or inhibition effect of neurons near the winning neuron is judged and dynamically adjusted to generate a set of typical scenarios of new energy and load. Considering that virtual power plants provide various types of electricity products for grid operation, such as electrical energy, peak shaving, and frequency regulation, a benefit analysis index system for various types of electricity products of virtual power plants is constructed. By combining a set of typical scenarios with a benefit analysis index system and considering the probability distribution of each scenario, a virtual power plant full-scenario benefit analysis method integrating entropy weight method and analytic hierarchy process is proposed.

2. The full-scenario cost analysis method for virtual power plants participating in electricity commodity trading according to claim 1, characterized in that, The scenario feature distance matrix is ​​constructed based on the spatiotemporal distribution characteristics of new energy output and load demand, including: The values ​​and trends of new energy power generation and user electricity load in different time periods are divided by day. T time segments are defined with a preset duration as each time node. Each day is used as a sample feature, and vector data are used to... It means that, among them It is the nth feature value of the sample after the input data has been standardized; In a self-organizing map neural network, each neuron is a vector composed of features in different proportions. , The weight of the nth feature is represented by any neuron vector as input. The Euclidean distance d between the neuron and the sample is calculated using the following formula: ； In the formula, x i Belongs to X, W i It belongs to W.

3. The full-scenario cost analysis method for virtual power plants participating in electricity commodity trading according to claim 2, characterized in that, The step of using the distance matrix as input, taking one of its elements as a sample, and finding the neuron closest to the sample as the winning neuron based on a self-organizing map neural network includes: Based on Euclidean distance, find Zhongyu Minimum distance connection weight vector ,Right now: ； In the formula, The distance function is represented by Euclidean distance for datasets that connect distance attributes.

4. The full-scenario cost analysis method for virtual power plants participating in electricity commodity trading according to claim 3, characterized in that, The process of determining the activation or inhibition effect of neurons near the winning neuron, making dynamic adjustments, and generating a set of typical scenarios for new energy sources and loads includes the following steps: The winning neuron is denoted as g. A neighborhood is defined to assess the activity of neurons near the winning neuron. This is based on the clustering mechanism of self-organizing map neural networks. When input sample data is fed into the network, each neuron calculates its similarity to the input sample. The neuron with the highest similarity and the best match for the input sample features is the winning neuron. Then, based on the spatial location of the winning neuron in the network topology, a local region with a specific range and shape is defined. This range dynamically shrinks during the training iteration process, thereby enabling the entire network to quickly converge towards the sample features. The formula is: ； ； In the formula, The neighborhood function refers to the Gaussian function distribution model. Let be the distance between the i-th neuron and the j-th neuron; The initial neighborhood function value; The time constant of the neighborhood function; It is an exponentially decaying function. As the number of iterations and learning iterations increases, the bandwidth of the neighborhood function decreases, gradually reducing the influence of the winning neuron. After determining whether neurons surrounding the winning neuron g are activated or inhibited through the neighborhood function, the activation or inhibition behavior is further determined. The weight vector values ​​of the neurons are adjusted according to the input sample. Activation moves the neuron closer to the input sample, while inhibition moves it away from the input sample. This is achieved through the following adjustment formula: ； ； In the formula: This represents the change in weight of the j-th neuron; It is an exponentially decaying function that controls the adjustment of neurons in the SOM, also known as the learning rate. This is the error term between the sample and the weights; The initial learning rate; It is the time constant of the learning rate; Through the aforementioned activation or inhibition behaviors, different samples are continuously input, and the autonomous reinforcement learning process is repeated until the neuron structure is stable after training or the maximum number of iterations is reached. At this point, each input sample has a winning neuron, which outputs the clustering result.

5. The full-scenario cost analysis method for virtual power plants participating in electricity commodity trading according to claim 1, characterized in that, The virtual power plant full-scenario benefit analysis method that integrates entropy weight method and analytic hierarchy process includes: Taking into account the total revenue of virtual power plants from various electricity commodities, the delayed revenue from grid investment, the cost savings from traditional reserves, the cost reduction from grid line losses, and the user-side revenue, an economic benefit analysis index for virtual power plants participating in various electricity commodity transactions is established. Taking into account factors such as the improvement rate of grid frequency stability, the reduction rate of system peak-valley difference, and the rate of meeting ramp-up demand, we construct a grid operation benefit analysis index for virtual power plants participating in multiple types of electricity commodity transactions. Taking into account carbon dioxide emissions and renewable energy absorption rate, we construct environmental benefit analysis indicators for virtual power plants participating in various types of electricity commodity transactions, integrate various benefit analysis indicators, and construct a benefit analysis indicator system for virtual power plants participating in various types of electricity commodity transactions.

6. The method for full-scenario benefit analysis of virtual power plants participating in electricity commodity trading according to claim 5, characterized in that, The virtual power plant full-scenario benefit analysis method that integrates entropy weight method and analytic hierarchy process also includes: Based on a given operating scenario, the indicators are first positively normalized, that is, the dimensions of the indicators are removed, and the entropy and weight values ​​of the indicators are calculated. The subjective weights of the indicators are calculated using the analytic hierarchy process (AHP). An indicator importance judgment matrix is ​​constructed based on expert experience. The consistency index and consistency ratio are obtained by solving the matrix based on the largest eigenvalue. Then, the geometric mean of each row of elements is calculated and normalized to obtain the subjective weight vector. By combining the objective weights obtained by the entropy weight method with the subjective weights obtained by the analytic hierarchy process, a comprehensive weight for a given scenario can be obtained. Considering various schemes for virtual power plants to provide electricity trading and trading of multiple types of electricity commodities, calculate the comprehensive score of the virtual power plant trading scheme for a specific scenario; Finally, based on the probability distribution of typical virtual power plant scenarios, considering the probability of different scenarios occurring, the benefit evaluation results of all scenarios are comprehensively evaluated by using probability weighted summation to obtain the full-scenario benefit analysis results.

7. A full-scenario cost analysis system for virtual power plants participating in electricity commodity trading, characterized in that, Using the method of any one of claims 1 to 6, the system comprises: The feature distance calculation unit is used to collect the renewable energy output curve and load demand power curve from the historical operation data of the virtual power plant, and construct the scene feature distance matrix based on the spatiotemporal distribution characteristics of renewable energy output and load demand. The scene generation unit is used to take the distance matrix as input, take one of its elements as a sample, find the neuron closest to the sample as the winning neuron based on the self-organizing map neural network, and judge the activation or inhibition effect of the neurons near the winning neuron, make dynamic adjustments, and generate a set of typical new energy and load scenarios. The analysis index construction unit is used to consider the various types of electricity products that virtual power plants provide for grid operation, such as electricity energy, peak shaving, and frequency regulation, and to construct a benefit analysis index system for various types of electricity products of virtual power plants. The evaluation unit is used to combine a set of typical scenarios with a benefit analysis index system, consider the probability distribution of each scenario, and propose a virtual power plant full-scenario benefit analysis method that integrates entropy weight method and analytic hierarchy process.

8. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1-6.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-6.

Images