Electric power spot market operation evaluation method based on multi-factor evaluation model

Abstract

The invention discloses an electric power spot market operation evaluation method based on a multi-factor evaluation model, and relates to the technical field of electric power spot market operation evaluation, and the method comprises the following steps: carrying out the initial market simulation based on the backtest of historical data, and obtaining a first state set; performing feature division on the first state set by adopting a clustering algorithm, taking a division result as a preset Bayesian network input signal, and constructing an operation evaluation decision model through a Bayesian network; according to the operation evaluation decision model, making a decision on a preset first action set to obtain a simulation alternating sequence; a preset adversarial network is adopted to carry out feature mapping on the simulation alternating sequence, and projection offset features are obtained; according to the projection offset features, the state transition probability of the preset Bayesian network is compensated, the operation evaluation decision model is optimized, and the problem of mode switching decision distortion caused by feature space mismatch in a traditional evaluation method is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electricity spot market operation, and more specifically, to an electricity spot market operation evaluation method based on a multi-factor evaluation model. Background Art

[0002] The electricity spot market is a market where eligible operators conduct day-ahead, intraday, and real-time electricity transactions. In this market, electricity prices are determined by a market-clearing mechanism, typically based on the intersection of the supply and demand curves. A key characteristic of the electricity spot market is its high price volatility, which fluctuates with supply and demand. Therefore, the stability and operational efficiency of the electricity spot market are crucial to the economic and security of the power system.

[0003] The electricity spot market should sequentially conduct simulation trials, settlement trials, and formal operations. This process requires a comprehensive assessment of multiple factors, including market mechanisms, price discovery, trading rules, and the behavior of market participants. Market operation involves complex resource allocation, market rule design, and supply-demand balance. Ensuring stable market operation is a significant challenge, particularly as the proportion of large-scale renewable energy (such as wind and solar) continues to increase. The randomness, intermittency, and volatility of renewable energy sources exacerbate price fluctuations in the electricity market and increase the uncertainty of market decision-making.

[0004] For example, the invention patent announcement with the announcement number CN113657705B discloses a method, device, and storage medium for evaluating the impact of parameters on the electricity spot market, which solves the problem of being able to specifically construct and analyze important factors affecting the electricity market in the prior art. The present invention includes the following steps: S1: establishing an evaluation model, setting economic evaluation indicators, electricity evaluation indicators, operation evaluation indicators, and other evaluation indicators within the evaluation model; S2: determining parameters that have a key impact on the electricity market; S3: determining the optimal evaluation indicator; S4: evaluating the impact of the parameters on the electricity market based on the number of optimal evaluation indicators, and determining the intensity of the parameter's impact based on the quantitative data of the optimal evaluation indicator. The present invention has the advantages of being able to accurately evaluate the operating status of the electricity market and having a wide range of applications.

[0005] For example, the invention patent announcement with announcement number: CN111754076B discloses a method and device for evaluating wholesale electricity market models, which provides a method for evaluating wholesale electricity market models, namely, performing simulated clearing of two or more market models for the target electricity market of the same regional power grid, and outputting comparison results based on comparison indicators according to the simulated clearing results of each market model; the comparison indicators include one or more of inter-provincial market connection, day-ahead optimization space, all-day power generation cost, and clean energy consumption. The present disclosure also provides an output device, a storage device, and a computing device using the above method. The present disclosure provides a specific electricity market model selection and evaluation method and device for the construction of regional electricity spot markets in my country.

[0006] The above disclosed technical solutions have at least the following technical problems: Existing technologies for evaluating electricity spot market operations, such as Monte Carlo simulation or agent-based simulation, require manual pre-set switching conditions when performing evaluation actions. These methods lack the ability to dynamically switch evaluation strategies and cannot adjust evaluation actions in a timely manner. Furthermore, due to the dynamic nature of the market, these strategies can result in the selection of mismatched evaluation actions when switching between them, leading to prediction failures or misjudgment of risks. The present invention addresses these issues by providing a solution. Summary of the Invention

[0007] To overcome the aforementioned shortcomings of the existing technology, embodiments of the present invention provide a method for evaluating electricity spot market operations based on a multi-factor evaluation model. This method constructs a market simulation model through historical data backtesting, combines it with a dynamic Bayesian network to quantify power supply uncertainty and market manipulation risk, and introduces a confidence-driven dynamic action switching mechanism to address the problem of mode switching decision distortion caused by feature space mismatch in traditional methods.

[0008] To achieve the above object, the present invention provides the following technical solutions: The invention discloses an operation evaluation method for an electricity spot market based on a multi-factor evaluation model, comprising the following steps: performing an initial market simulation based on backtesting of historical data and obtaining a first state set; performing feature division on the first state set using a clustering algorithm, and using the division result as a preset Bayesian network input signal, and constructing an operation evaluation decision model through the Bayesian network; making a decision on a preset first action set according to the operation evaluation decision model, and obtaining a simulation alternating sequence; performing feature mapping on the simulation alternating sequence using a preset adversarial network, and obtaining a projection offset feature; and supplementing the state transition probability of the preset Bayesian network according to the projection offset feature, and optimizing the operation evaluation decision model.

[0009] In a preferred embodiment, the initial market simulation is performed based on backtesting of historical data, specifically: historical transaction data of the electricity spot market is obtained, and a first model is constructed; market clearing price fluctuations are obtained according to the first model, an initial market simulation data set is obtained, and the simulation data is verified by a backtesting method; the simulation data verification includes calculation of error rates of market simulation data and historical transaction data, and the validity of the market simulation data is judged by the error rate.

[0010] In a preferred embodiment, the first state set includes new energy state characteristics, which are specifically obtained as follows: obtaining the first data of new energy and performing discretization processing to obtain a state variable set; performing causal relationship learning on the state variable set to construct a Bayesian network model; based on the Bayesian network model, dividing the state variable set into several fuzzy intervals to obtain a fuzzy rule base; based on the fuzzy rule base, calculating the conditional probabilities between the nodes of the Bayesian network model to obtain an initial conditional probability table; using the initial conditional probability table as the initial parameters of the Bayesian network model and performing parameter optimization to obtain a probability distribution matrix; based on the probability distribution matrix, calculating the probability value that the electricity price exceeds a preset first price threshold within a preset time window, and using it as the new energy state characteristic.

[0011] In a preferred embodiment, the first state set also includes operational risk characteristics, which are specifically obtained as follows: the electricity spot market is divided into several sub-markets, and first data is obtained, the first data including historical bidding capacity share data, bidding price data and bidding behavior data of each sub-market; the historical bidding capacity share data is calculated by Herfindahl-Hirschman index to obtain a market concentration assessment value for each sub-market; based on the bidding price data, the price volatility of each sub-market is calculated; based on the bidding behavior data, the behavior switching rate and behavior fluctuation frequency of each sub-market are calculated; preset weights are assigned to the behavior switching rate and the behavior fluctuation frequency to obtain a behavior stability score value; the behavior stability score value is feature-fused with the market concentration assessment value to obtain an operational risk characteristic.

[0012] In a preferred embodiment, the first state set also includes game characteristics, which are specifically obtained as follows: constructing a game tree model and screening the game tree paths according to the behavior stability score value; calculating the price deviation on the remaining paths and the cumulative behavior cost value of the paths to obtain the path evaluation value; screening the remaining paths based on the path evaluation value to obtain a high-risk path set; performing an abnormal evaluation on the high-risk path set to obtain abnormal bidding behavior characteristics; fusing the abnormal bidding behavior characteristics and the price volatility of the sub-market according to preset weights to obtain game characteristics.

[0013] In a preferred embodiment, the clustering algorithm is used to perform feature division on the first state set, and the division result is used as a preset Bayesian network input signal, and an operation evaluation decision model is constructed through the Bayesian network, specifically: feature extraction is performed on the first state set to obtain a first state vector, and the state transition probability of the first state vector is calculated using a preset Bayesian network; the first state vector is compressed using a dimensionality reduction algorithm, and feature fusion is performed with the state transition probability to obtain a first feature space; the first feature space is clustered using a clustering algorithm to obtain a first area and a second area, and a Monte Carlo simulation model library is constructed in the first area, and an Agent-based market simulation model library is constructed in the second area; the spatial distance between the center of mass of the first area and the second area is calculated, and the model library is activated according to the spatial distance to obtain an operation evaluation decision model.

[0014] In a preferred embodiment, the operation evaluation decision model is used to make a decision on the preset first action set to obtain a simulation alternation sequence, where the first action set includes Monte Carlo simulation and Agent-based market simulation. Specifically, the preset first action set is dynamically divided into states according to the operation evaluation decision model. If the current state is divided into the first area, the preset optimal parameter set of the Monte Carlo simulation model library is loaded and the simulation is started; if the current state is divided into the second area, the preset behavior rule library of the Agent-based model library is activated and the interaction frequency of the intelligent agents is dynamically adjusted; the switching time points and sequence of the simulation are obtained to obtain a simulation alternation sequence.

[0015] In a preferred embodiment, the use of a preset adversarial network to perform feature mapping on the simulated alternating sequence to obtain projection offset features is specifically as follows: constructing an adversarial network, the adversarial network including a generator and a discriminator; performing feature extraction on the simulated alternating sequence to obtain Monte Carlo feature vectors and Agent-based feature vectors; mapping the Monte Carlo feature vectors and Agent-based feature vectors to a first feature space according to the generator to obtain Monte Carlo projection vectors and Agent-based projection vectors; performing error judgment according to the discriminator to obtain projection offset features of the Monte Carlo projection vectors and Agent-based projection vectors.

[0016] In a preferred embodiment, the state transition probability of the preset Bayesian network is supplemented according to the projection offset feature to optimize the operation evaluation decision model, specifically: the amplitude and direction of the projection offset feature are obtained, and the projection offset feature is discretized to obtain the offset level; the state transition probability of the preset Bayesian network is obtained, the state transition path is dynamically adjusted according to the offset level, and the state transition probability is recalculated to obtain the supplemented state transition probability; the supplemented state transition probability is dynamically applied to the operation evaluation decision model to obtain the optimized operation evaluation decision model.

[0017] The technical effects and advantages of the electricity spot market operation evaluation method based on the multi-factor evaluation model of the present invention are as follows: 1. The present invention ensures the accuracy of the initial state set by performing market simulation based on backtesting of historical data, and uses a clustering algorithm to divide the feature space and optimize the rationality of state classification, thereby reducing the risk of feature space mismatch. Secondly, the Bayesian network is used to construct an operation evaluation decision model, and through a systematic analysis of new energy state characteristics, operational risk characteristics and game characteristics, multi-dimensional feature fusion is achieved, so that the model can accurately perceive the market status. In addition, in the decision-making process, the method alternates between the dual modes of Monte Carlo simulation and Agent-based market simulation, and uses spatial distance calculation to activate the adaptive mode library, making the simulation mode switching more targeted and reducing erroneous switching caused by feature mismatch. Furthermore, the method introduces an adversarial network to perform feature mapping on the simulation alternating sequence, uses the projection offset feature to correct the state transition probability, and realizes the dynamic complementation of the Bayesian network, thereby optimizing the evaluation decision model so that it can still maintain high adaptability and accuracy in the face of changes in the market environment. Compared with traditional methods, this method can not only effectively avoid the mismatch problem of feature space and improve the accuracy of mode switching, but also perform adaptive optimization during the state transition process, improve the stability and reliability of the evaluation, and provide more accurate and intelligent decision-making support for the formal operation of the electricity spot market. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of the electricity spot market operation evaluation method based on the multi-factor evaluation model of the present invention. DETAILED DESCRIPTION

[0019] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0020] Example 1, Figure 1 The present invention provides an electricity spot market operation evaluation method based on a multi-factor evaluation model, which includes the following steps: S1, perform initial market simulation based on backtesting of historical data and obtain the first state set; In this example, the initial market simulation is performed based on the backtest of historical data, specifically: Obtain historical transaction data of the electricity spot market and build the first model; The market clearing price fluctuations are obtained according to the first model, an initial market simulation data set is obtained, and the simulation data is verified by a backtesting method; the simulation data verification includes calculating the error rate of the market simulation data and the historical transaction data, and judging the validity of the market simulation data by the error rate.

[0021] It should be noted that the initial market simulation based on historical data backtesting requires first obtaining historical transaction data from the electricity spot market and constructing a power output probabilistic model and a load demand response model. Power output is affected by factors such as weather, fuel supply, and equipment failures, and can be modeled using Markov chains or kernel density estimation methods. The load demand response model, on the other hand, incorporates user electricity usage habits and price elasticity, using regression analysis or machine learning to simulate electricity consumption trends under different price conditions. The significance of this process lies in building a dynamic supply and demand model based on real market behavior, thereby improving the reliability and adaptability of the simulation. Based on this, the constructed power output probabilistic model and load demand response model are used to simulate market clearing prices, forming the initial market simulation dataset. Market clearing refers to the calculation of the market equilibrium price during the process of matching supply and demand. For example, when wind power output increases, electricity prices may fall, while the temporary withdrawal of thermal power units may cause prices to rise. Such market dynamics can be modeled using Monte Carlo simulation or agent-based simulation methods. By simulating market clearing price fluctuations under different scenarios, we can have a more comprehensive understanding of the market operation mechanism, predict market price risks, and provide support for market management and investment decisions.

[0022] In addition, backtesting verifies the error rate between simulated data and historical trading data, and uses statistical metrics such as mean squared error (MSE) and mean absolute percentage error (MAPE) to assess the accuracy of simulation results. For example, if the error between simulated and historical market prices is within 5%, the model effectively reflects market trends. However, if the error is larger, the modeling methodology needs to be optimized or input assumptions adjusted. The significance of error assessment is to ensure the reliability of simulation results, enhance the credibility of the model, and make it more suitable for market analysis and forecasting. Finally, the effectiveness of the initial market simulation is determined based on the error rate. When the backtest error meets a preset threshold (e.g., an error below 10%), the simulation model is considered valid and can be used for subsequent market evaluation. If the error exceeds the threshold, input parameters need to be optimized, market behavior assumptions adjusted, or the simulation methodology needs to be refined to improve the overall accuracy of the simulation system.

[0023] Furthermore, the uncertainty state characteristics of renewable energy, operational risk characteristics, and game characteristics were selected as the first state set for the initial market simulation to evaluate the stability and operation of the electricity spot market. The uncertainty state characteristics of renewable energy reflect the output volatility of renewable energy sources such as wind power and photovoltaics due to meteorological factors. This uncertainty can cause significant market price fluctuations and affect the balance of electricity supply and demand. Therefore, probabilistic modeling is needed to quantify the impact of renewable energy on the market. The operational risk characteristics measure the stability of market structure and market behavior. Factors such as market concentration, bidding behavior fluctuations, and price volatility are directly related to the degree of market competition and bidding strategies. Excessive market concentration or abnormal bidding behavior fluctuations can lead to market manipulation and price distortion, thus affecting market fairness and efficiency. The game characteristics focus on the interactive relationships between market participants. By analyzing the bidding behavior, price deviation, and market strategy evolution of market participants through a game tree, abnormal bidding behavior, high-risk paths, and the possibility of market manipulation can be identified, thereby improving the accuracy of market supervision. Therefore, the comprehensive consideration of these three characteristics can not only comprehensively reflect the core uncertainties and risk sources of market operation, but also provide more accurate decision-making basis for market assessment, thereby ensuring the stability, fairness, and sustainability of market operation.

[0024] In this example, the first state set includes new energy state characteristics, which are obtained as follows: Obtaining the first data of the new energy and performing discretization processing to obtain a state variable set; Conduct causal relationship learning on the state variable set and build a Bayesian network model; Based on the Bayesian network model, the state variable set is divided into several fuzzy intervals to obtain the fuzzy rule base; Based on the fuzzy rule base, the conditional probabilities between the nodes of the Bayesian network model are calculated to obtain the initial conditional probability table; The initial condition probability table is used as the initial parameters of the Bayesian network model and the parameters are optimized to obtain the probability distribution matrix; Based on the probability distribution matrix, the probability value of the electricity price exceeding the preset first price threshold within the preset time window is calculated and used as the new energy state feature.

[0025] It should be noted that the first data on renewable energy sources, including key variables such as historical renewable energy power forecast errors, renewable energy penetration rates, and electricity price volatility, is used to construct a state variable set. Causal relationship learning is then performed using a Bayesian network to accurately characterize the uncertainty of renewable energy output and its impact on market prices. By fuzzy partitioning the state variables and establishing a fuzzy rule base, the model can more flexibly handle complex market dynamics, avoid misjudgments caused by data noise or sudden changes, and improve the adaptability of the model. The conditional probability of the Bayesian network derived from the fuzzy rule base enables the model to effectively predict the market impact of renewable energy under uncertain conditions. Parameter optimization is then used to obtain a probability distribution matrix that better reflects actual market conditions, further enhancing forecast accuracy. Furthermore, this probability distribution matrix is ​​used to calculate the probability of electricity prices exceeding a threshold within a preset time window, providing a quantitative risk assessment indicator for renewable energy uncertainty. This helps market managers develop more robust scheduling strategies, mitigate the impact of renewable energy fluctuations on market prices, and improve the reliability of market simulations, making the assessment process more scientific and data-driven, thereby providing strong support for the stable operation of the electricity spot market.

[0026] In this example, the first state set also includes operational risk characteristics, which are obtained as follows: Divide the electricity spot market into a plurality of sub-markets and obtain first data, wherein the first data includes historical bid capacity share data, bid price data, and bid behavior data of each sub-market; The Herfindahl-Hirschman Index is calculated based on the historical bidding capacity share data to obtain the market concentration assessment value of each sub-market; Calculate the price volatility of each sub-market based on the bid price data; Based on bidding behavior data, calculate the behavior switching rate and behavior fluctuation frequency of each sub-market; Assign preset weights to the behavior switching rate and behavior fluctuation frequency to obtain a behavior stability score; The behavioral stability score value and the market concentration assessment value are integrated to obtain the operational risk characteristics.

[0027] In this example, the price volatility of each sub-market is calculated as follows;

[0028] in, is the price volatility, is the number of bidding data, is the i-th bidding price, The preset base price.

[0029] It should be noted that in the electricity spot market, the construction of operational risk characteristics requires a refined analysis based on the characteristics of the sub-markets. Based on the proximity of transaction time and delivery time and the differences in market functions, the electricity spot market is usually divided into three sub-markets: the day-ahead market, the intraday market, and the real-time market. The trading mechanisms and risk characteristics of each market are significantly different, specifically: Day-ahead market: Electricity trading conducted one day in advance, with power generation plans and prices determined based on the 24-hour load forecast for the next day. Intraday market: transactions conducted after the day-ahead market closes and before the real-time market opens (usually within the same day), used to adjust power generation plans; Real-time market: Transactions conducted before the actual delivery of electricity (usually 15 minutes to 1 hour) to balance real-time supply and demand deviations.

[0030] Furthermore, by dividing the electricity spot market into multiple submarkets and calculating the Herfindahl-Hirschman Index to assess market concentration, it is possible to determine whether market competition is sufficient and avoid operational risks caused by an irrational market structure. The standard deviation of bid prices measures market price volatility, helping to identify drastic price fluctuations and thus providing early warning of anomalies that may affect market stability. Furthermore, by analyzing bidding behavior data, the adaptability and stability of market participants under different market conditions are assessed by calculating the switching rate and frequency of fluctuations. A behavioral stability score is then calculated using weights, resulting in a more accurate assessment of bidding behavior. Finally, the integration of market concentration and behavioral stability scores comprehensively characterizes operational risks in market operations, allowing the assessment to not only focus on the rationality of the overall market structure but also delve into changes in the bidding patterns of individual market participants. This approach can provide a scientific basis for the formal operation of the spot market, ensuring the stability of the market mechanism under different operating conditions. It also provides managers with reliable data support, enabling timely intervention measures when market anomalies are detected, thereby mitigating systemic risks that may arise during the formal operation of the market.

[0031] In this example, the first state set of the initial market simulation also includes game characteristics, which are obtained as follows: Construct a game tree model and filter the game tree paths based on the behavior stability score; Calculate the price deviation on the remaining path and the cumulative behavior cost of the path to obtain the path evaluation value; The remaining paths are screened based on the path evaluation value to obtain a set of high-risk paths; Anomaly evaluation is performed on the high-risk path set to obtain abnormal bidding behavior characteristics; The abnormal bidding behavior characteristics and the price volatility of the sub-market are integrated according to the preset weights to obtain the game characteristics.

[0032] In this example, the price deviation is calculated as follows:

[0033] in, is the price deviation on the i-th residual path, is the price of the i-th remaining path at time t, is the preset market benchmark price at time t, is the market quotation standard at time t, is the number of time nodes included in the remaining path.

[0034] In this example, the calculation formula for the cumulative behavior cost of the path is as follows:

[0035] in, The cumulative behavior cost of the path on the i-th remaining path, is the price of the i-th remaining path at time (t-1), is the behavior switching frequency of the i-th remaining path at time t.

[0036] It is important to note that by constructing a game tree model to simulate the bidding decision paths of market participants under different market conditions and screening the game tree paths based on behavioral stability scores, this helps eliminate highly random or unrepresentative paths, thereby improving analysis accuracy and computational efficiency. Next, the price deviation and cumulative behavioral cost of the remaining paths are calculated to quantify the rationality of the market participants' strategies under specific circumstances. This avoids relying solely on historical data and instead conducts a dynamic assessment based on the actual behavior of market participants. Based on these calculation results, the paths are screened to extract a set of high-risk paths that could cause significant market price fluctuations or affect market stability, thereby identifying market participants or trading patterns that may exhibit abnormal bidding behavior. Subsequently, this set of high-risk paths is further evaluated for abnormalities, extracting characteristics of abnormal bidding behavior, such as extreme fluctuations in bid prices and frequent adjustments to bidding strategies. This allows identification of behavioral patterns that may affect market stability and ensures that the bidding strategies of market participants remain predictable and rational during the official operation. Finally, the abnormal bidding behavior characteristics and the price volatility of the sub-market are integrated according to the preset weights, so that the game characteristics can fully reflect the correlation between the decision-making patterns of market entities and the overall market fluctuations, providing a more intuitive reference for market supervision and operation evaluation.

[0037] The advantage of this approach lies in its focus not only on overall market performance but also on in-depth analysis of the specific decision-making behaviors of market participants, thereby enhancing the comprehensiveness and accuracy of the assessment. Compared to traditional market performance assessment methods, this approach, based on a game theory approach, dynamically tracks the strategic adjustments of market participants under different market conditions. Furthermore, it incorporates price volatility and abnormal bidding behavior to form a three-dimensional assessment of market operational risks. During the operation of the electricity spot market, this approach can help managers proactively identify potential market anomalies, enabling more timely and effective regulatory measures. Furthermore, through multi-level path screening and feature fusion, this approach avoids the potential bias associated with single-factor judgments, making the assessment results more scientific and reliable. Ultimately, this approach not only provides objective data support for the official market operation but also serves as an important basis for future market optimization and adjustments, ensuring the stability, security, and efficiency of the electricity spot market after its official operation.

[0038] S2, using a clustering algorithm to perform feature segmentation on the first state set, and using the segmentation results as a preset Bayesian network input signal to construct an operation evaluation decision model through the Bayesian network; In this example, a clustering algorithm is used to perform feature segmentation on the first state set, and the segmentation results are used as the input signal of the preset Bayesian network. The Bayesian network is used to construct an operation evaluation decision model, specifically: Extract features from the first state set to obtain a first state vector, and calculate the state transition probability of the first state vector using a preset Bayesian network; The first state vector is compressed using a dimensionality reduction algorithm, and feature fusion is performed with the state transition probability to obtain the first feature space; Clustering the first feature space using a clustering algorithm to obtain a first region and a second region, and constructing a Monte Carlo simulation model library in the first region and an Agent-based market simulation model library in the second region; The spatial distance between the centroids of the first region and the second region is calculated, and the pattern library is activated according to the spatial distance to obtain an operation evaluation decision model.

[0039] It should be noted that the above method provides a more intelligent and adaptive mode switching decision-making solution for evaluating electricity spot market operations. First, a clustering algorithm is used to characterize the first state set, enabling precise differentiation of market states and avoiding inappropriate evaluations caused by state ambiguity. Second, a Bayesian network is used to calculate state transition probabilities, enabling a probabilistic description of the market state evolution process, improving the robustness of the model. Furthermore, a dimensionality reduction algorithm is introduced to compress the state vector and incorporate state transition probabilities, reducing computational complexity while preserving core information about market state changes and ensuring the effectiveness of the evaluation. Subsequently, based on the clustering results, different states are divided into two regions: a Monte Carlo simulation model library and an agent-based market simulation model library. This allows the evaluation model to flexibly select the optimal simulation method based on the different characteristics of the market state, improving evaluation accuracy. Finally, by calculating the spatial distance between the centroids of the two regions, the corresponding model library is dynamically activated, achieving intelligent switching between simulation methods. The advantages of this method are: it avoids the limitation of traditional evaluation methods that require manual pre-set switching conditions and realizes data-driven mode switching; it effectively solves the decision distortion problem caused by feature space mismatch and improves the adaptability of evaluation; through Bayesian network modeling, it makes the evaluation more in line with the randomness and uncertainty of the market; it improves the stability and computational efficiency of the evaluation model, making the simulation and prediction of the power market more accurate.

[0040] S3, making a decision on the preset first action set according to the operation evaluation decision model to obtain a simulation alternating sequence; In this example, based on the running evaluation decision model, a decision is made on a preset first action set to obtain a simulation alternating sequence. The first action set includes Monte Carlo simulation and Agent-based market simulation, specifically: Perform dynamic state division on the preset first action set according to the operation evaluation decision model. If the current state is divided into the first area, load the preset optimal parameter set of the Monte Carlo simulation model library and start the simulation; If the current state is classified into the second area, activate the preset behavior rule library of the Agent-based pattern library and dynamically adjust the agent interaction frequency; The switching time points and order of the simulation are obtained to obtain the simulation alternation sequence.

[0041] S4, using a preset adversarial network to perform feature mapping on the simulated alternating sequence to obtain the projection offset feature; In this example, a preset adversarial network is used to perform feature mapping on the simulated alternating sequence to obtain the projection offset feature, specifically: Constructing an adversarial network, wherein the adversarial network includes a generator and a discriminator; Perform feature extraction on the simulation alternating sequence to obtain Monte Carlo feature vectors and Agent-based feature vectors; According to the generator, the Monte Carlo feature vector and the Agent-based feature vector are mapped to the first feature space to obtain the Monte Carlo projection vector and the Agent-based projection vector; The discriminator performs error judgment and obtains the projection offset features of the Monte Carlo projection vector and the Agent-based projection vector.

[0042] It should be noted that the generator in the adversarial network consists of a fully connected layer and a multi-head attention mechanism, which is used to map the Monte Carlo feature vector and the agent-based feature vector to a shared latent space; the discriminator uses a convolutional neural network structure to discriminate the simulation type of the source of the latent space features. The specific steps of adversarial training are as follows: (1) Inputting the Monte Carlo simulation data into a generator to generate a Monte Carlo projection vector; (2) Inputting the Agent-based simulation data into the generator to generate an Agent-based projection vector; (3) Calculating the source discrimination error of the two types of projection features through the discriminator; (4) Jointly optimizing the loss function of the generator and the discriminator, wherein the loss function includes adversarial loss, feature reconstruction loss, and spatial orthogonality constraint.

[0043] It should be noted that feature mapping of simulation alternation sequences using a pre-set adversarial network offers significant advantages in intelligent adaptation and feature alignment. First, by introducing a generator and a discriminator to construct an adversarial network, the features of two different simulation modes—Monte Carlo simulation and agent-based market simulation—are mapped into a unified feature space, thereby reducing evaluation bias caused by differences in feature spaces. Second, by extracting features from the simulation alternation sequence, Monte Carlo feature vectors and agent-based feature vectors are generated, respectively. The generator then performs the mapping, ensuring consistent representation of the two simulation modes within the same feature space and improving the comparability and reliability of the simulation results. Furthermore, the discriminator performs error analysis on the mapped feature vectors and calculates projection offset features. This allows the system to quantify the deviations between different simulation methods and further optimize the model's state transition process, mitigating decision distortion caused by feature mismatch during mode switching. This approach not only improves the validity of simulation data and enhances the accuracy of market evaluation under different simulation modes, but also enables dynamic feature correction during decision optimization through adversarial training, enhancing the adaptability and robustness of the market evaluation model.

[0044] S5, based on the projection offset characteristics, the state transition probability of the preset Bayesian network is supplemented to optimize the operation evaluation decision model; In this example, the state transition probability of the preset Bayesian network is supplemented according to the projection offset characteristics to optimize the operation evaluation decision model. Specifically: Obtain the amplitude and direction of the projection offset feature, and discretize the projection offset feature to obtain the offset level; Obtain the state transition probability of the preset Bayesian network, dynamically adjust the state transition path according to the offset level, and recalculate the state transition probability to obtain the state transition probability after correction; The state transition probability after filling in the truth is dynamically applied to the operation evaluation decision model to obtain the optimized operation evaluation decision model.

[0045] It should be noted that the amplitude and direction of the projection offset feature are obtained, and the projection offset feature is discretized to obtain the offset level. The specific steps are as follows: When the offset exceeds the first threshold, it is marked as a high-risk offset, triggering online fine-tuning of the generator parameters; When the offset is between the second threshold and the first threshold, it is marked as a medium-risk offset and an abnormality log is recorded; When the offset is lower than the second threshold, it is marked as normal fluctuation and the current mapping parameters are maintained.

[0046] Furthermore, the offset level is embedded into the Bayesian network as a new observation node, and a three-dimensional conditional probability relationship including state, action and offset characteristics is constructed. The original state transition probability is directional adjusted according to the offset level: when the deviation is high, the probability weight of the unconventional transition path (such as steady state → transition state) is enhanced; when the deviation is low, the original probability distribution is maintained or fine-tuned. Combining the corrected transition probability and real-time risk indicators, an operation strategy integrating anti-offset capability is generated, giving priority to simulated alternating paths with low switching costs and significant risk suppression effects, and feeding back the execution effect of the true-to-false decision (such as switching success rate and risk control rate) to the feature mapping module. The projection parameters of the adversarial network and the probability adjustment weights of the Bayesian network are dynamically calibrated to form an adaptive decision optimization closed loop.

[0047] It should be noted that by calculating the amplitude and direction of the projected offset features and discretizing them to form offset levels, the system can quantify simulation errors and provide a basis for targeted corrections. Secondly, based on the offset levels, the state transition paths of the Bayesian network are dynamically adjusted, thereby reducing state misjudgments caused by feature space offsets and improving the accuracy of the market assessment model. The recalculated state transition probabilities after correction make the market state transition process more consistent with real market laws, helping to reduce bias caused by inherent model assumptions and improve the rationality of assessment decisions. Furthermore, the optimized state transition probabilities are dynamically applied to the operational assessment decision model, enabling it to continuously adjust to changes in the market environment, enhancing the robustness and intelligence of market assessments. Compared to traditional Bayesian networks with fixed parameters, this approach can perceive changes in market characteristics in real time and dynamically optimize state transition rules, making assessment results more timely and accurate.

[0048] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0049] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0050] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0051] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0052] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0053] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model, characterized in that: The following steps are involved: Perform initial market simulation based on backtesting of historical data and obtain the first state set; A clustering algorithm is used to perform feature division on the first state set, and the division results are used as the input signal of the preset Bayesian network to construct an operation evaluation decision model through the Bayesian network; According to the operation evaluation decision model, a decision is made on the preset first action set to obtain a simulation alternating sequence; A preset adversarial network is used to perform feature mapping on the simulated alternating sequence to obtain projection offset features; The state transition probability of the preset Bayesian network is supplemented according to the projection offset characteristics to optimize the operation evaluation decision model.

2. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 1, characterized in that: The initial market simulation based on the backtest of historical data is specifically as follows: Obtain historical transaction data of the electricity spot market and build the first model; The market clearing price fluctuations are obtained according to the first model, an initial market simulation data set is obtained, and the simulation data is verified by a backtesting method; the simulation data verification includes calculating the error rate of the market simulation data and the historical transaction data, and judging the validity of the market simulation data by the error rate.

3. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 2, characterized in that: The first state set includes new energy state characteristics, which are obtained in the following manner: Obtaining the first data of the new energy and performing discretization processing to obtain a state variable set; Conduct causal relationship learning on the state variable set and build a Bayesian network model; Based on the Bayesian network model, the state variable set is divided into several fuzzy intervals to obtain the fuzzy rule base; Based on the fuzzy rule base, the conditional probabilities between the nodes of the Bayesian network model are calculated to obtain the initial conditional probability table; The initial condition probability table is used as the initial parameters of the Bayesian network model and the parameters are optimized to obtain the probability distribution matrix; Based on the probability distribution matrix, the probability value of the electricity price exceeding the preset first price threshold within the preset time window is calculated and used as the new energy state feature.

4. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 3, characterized in that: The first status set also includes operational risk characteristics, which are obtained in the following manner: Divide the electricity spot market into a plurality of sub-markets and obtain first data, wherein the first data includes historical bid capacity share data, bid price data, and bid behavior data of each sub-market; The Herfindahl-Hirschman Index is calculated based on the historical bidding capacity share data to obtain the market concentration assessment value of each sub-market; Calculate the price volatility of each sub-market based on the bid price data; Based on bidding behavior data, calculate the behavior switching rate and behavior fluctuation frequency of each sub-market; Assign preset weights to the behavior switching rate and behavior fluctuation frequency to obtain a behavior stability score; The behavioral stability score value and the market concentration assessment value are integrated to obtain the operational risk characteristics.

5. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 4, characterized in that: The first state set also includes game characteristics, which are obtained in the following way: Construct a game tree model and filter the game tree paths based on the behavior stability score; Calculate the price deviation on the remaining path and the cumulative behavior cost of the path to obtain the path evaluation value; The remaining paths are screened based on the path evaluation value to obtain a set of high-risk paths; Anomaly evaluation is performed on the high-risk path set to obtain abnormal bidding behavior characteristics; The abnormal bidding behavior characteristics and the price volatility of the sub-market are integrated according to the preset weights to obtain the game characteristics.

6. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 5, characterized in that: The clustering algorithm is used to perform feature division on the first state set, and the division result is used as a preset Bayesian network input signal to construct an operation evaluation decision model through the Bayesian network, specifically: Extract features from the first state set to obtain a first state vector, and calculate the state transition probability of the first state vector using a preset Bayesian network; The first state vector is compressed using a dimensionality reduction algorithm, and feature fusion is performed with the state transition probability to obtain the first feature space; Clustering the first feature space using a clustering algorithm to obtain a first region and a second region, and constructing a Monte Carlo simulation model library in the first region and an Agent-based market simulation model library in the second region; The spatial distance between the centroids of the first region and the second region is calculated, and the pattern library is activated according to the spatial distance to obtain an operation evaluation decision model.

7. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 6, characterized in that: According to the operation evaluation decision model, a decision is made on a preset first action set to obtain a simulation alternating sequence, wherein the first action set includes Monte Carlo simulation and Agent-based market simulation, specifically: Perform dynamic state division on the preset first action set according to the operation evaluation decision model. If the current state is divided into the first area, load the preset optimal parameter set of the Monte Carlo simulation model library and start the simulation; If the current state is classified into the second area, activate the preset behavior rule library of the Agent-based pattern library and dynamically adjust the agent interaction frequency; The switching time points and order of the simulation are obtained to obtain the simulation alternation sequence.

8. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 7, characterized in that: The preset adversarial network is used to perform feature mapping on the simulated alternating sequence to obtain the projection offset feature, which is specifically: Constructing an adversarial network, wherein the adversarial network includes a generator and a discriminator; Perform feature extraction on the simulation alternating sequence to obtain Monte Carlo feature vectors and Agent-based feature vectors; According to the generator, the Monte Carlo feature vector and the Agent-based feature vector are mapped to the first feature space to obtain the Monte Carlo projection vector and the Agent-based projection vector; The discriminator performs error judgment and obtains the projection offset features of the Monte Carlo projection vector and the Agent-based projection vector.

9. The method for evaluating the operation of the electricity spot market based on a multi-factor evaluation model according to claim 8, characterized in that: The state transition probability of the preset Bayesian network is supplemented according to the projection offset feature to optimize the operation evaluation decision model, specifically: Obtain the amplitude and direction of the projection offset feature, and discretize the projection offset feature to obtain the offset level; Obtain the state transition probability of the preset Bayesian network, dynamically adjust the state transition path according to the offset level, and recalculate the state transition probability to obtain the state transition probability after correction; The state transition probability after filling in the truth is dynamically applied to the operation evaluation decision model to obtain the optimized operation evaluation decision model.

Images