Electric power spot transaction data model construction method based on data analysis

Abstract

The invention relates to the technical field of data model construction, in particular to an electric power spot transaction data model construction method based on data analysis, and the method comprises the steps: collecting key data in real time; preprocessing data to generate a training set; constructing a model to be incremented; determining the stability of training data; identifying an abnormal conduction type; generating a deviation vector adjustment parameter; and incrementally training and updating the model. According to the method, in the clearing process of the high-fluctuation limited electric power spot market, multi-dimensional parameters such as the node marginal electricity price change rate, the main output unit climbing gradient, the real-time order book fractal dimension and the satellite remote sensing cloud layer moving speed are introduced at the same time for data analysis; the problems of risk early warning delay and poor market regulation and control effect caused by poor model adaptability and insufficient prediction accuracy due to lack of timely identification and response to market supply and demand fluctuation and abnormal behaviors are effectively solved.

Description

Technical Field

[0001] This invention relates to the field of data model construction technology, and in particular to a method for constructing a data model for electricity spot trading based on data analysis. Background Technology

[0002] With the large-scale integration of renewable energy, the impact of meteorological factors on the power system is becoming increasingly prominent. Weather phenomena such as cloud movement and wind speed changes directly affect the output fluctuations of new energy power generation, thereby causing changes in supply and demand and price fluctuations in the electricity market. How to effectively integrate real-time meteorological data with market transaction data to accurately reflect the impact of meteorological changes on market behavior has become an important direction for improving the risk monitoring and regulation capabilities of the electricity spot market. Faced with the spatiotemporal heterogeneity and dynamic complexity of multi-source data, achieving deep integration and accurate modeling of meteorological information and market data remains a major technical challenge.

[0003] Chinese Patent Application Publication No. CN116433261A discloses a modeling method for a power spot trading data model. The method includes: S110, constructing the basic architecture of the power spot trading data model, referencing the CIM modeling standard, and setting specific hierarchical content; S120, collecting power spot trading data, including data from the trading center, dispatching system, and other data to support trading decision analysis; and S130, constructing data tables and assigning indicator attributes, specifically based on the collected and organized power spot trading data, following the constructed data model framework. The process involves: S110, constructing data tables and assigning indicator attributes to electricity spot market transaction data; S140, constructing a full-process data business domain model, specifically by sorting out the full-cycle business process of electricity spot market transaction data based on the basic transaction rules, the full-cycle business process of the transaction, and the categories of market participants, and constructing a full-process data business domain model; S150, forming a complete data model for electricity spot transactions, specifically by integrating the data physical model content from steps S110 to S140, establishing a data logical model of inter-table relationships, and forming a complete data model for electricity spot transactions.

[0004] Therefore, the aforementioned electricity spot market data modeling method has the following problems: This method focuses on static data model construction and business process streamlining, lacking modeling and analysis of time-series characteristics such as real-time volatility, dynamic changes in supply and demand response, and price fluctuations in the electricity spot market, making it difficult to cope with the real-time control needs of a highly volatile market; This method is mainly based on data from the trading center, dispatching system, and auxiliary decision-making, without involving the integration of multi-dimensional and multi-source heterogeneous data such as real-time meteorological information, remote sensing data, and fractal characteristics of market behavior, which limits the model's sensitivity and ability to identify external environment and abnormal market behavior. Summary of the Invention

[0005] To address this, the present invention provides a data analysis-based method for constructing a power spot trading data model. This method overcomes the problems in existing technologies, such as poor model adaptability and insufficient prediction accuracy due to the lack of timely identification and response to market supply and demand fluctuations and abnormal behaviors, which in turn lead to delayed risk warnings and poor market regulation effects.

[0006] To achieve the above objectives, this invention provides a method for constructing a data model for electricity spot trading based on data analysis, comprising: The system collects data in real time on the marginal electricity price change rate at nodes, the ramp-up gradient of major generating units, the fractal dimension of the real-time order book, and the cloud movement speed in satellite remote sensing images during the clearing process of the highly volatile and restricted electricity spot market. The collected data is then preprocessed within a preset training period to obtain an initial training set. Construct the incremental model based on the initial training set; Based on the model to be incremented, a steady-state determination is made according to the node marginal electricity price change rate, the ramp gradient and the preset steady-state threshold to obtain a first determination result or a second determination result. Based on the first determination result, the abnormal transmission path type is determined according to the cloud layer movement speed and the fractal dimension, resulting in either a supply and demand disturbance type or a price manipulation type. Based on the supply and demand disturbance type, a structural deviation vector is generated according to the marginal electricity price change rate at the node and the ramp gradient; based on the price manipulation type, a behavioral deviation vector is generated according to the fractal dimension and the cloud movement speed. The preset steady-state threshold and the preset training period are adjusted based on the structural deviation vector and the behavioral deviation vector; Based on the second determination result, the incremental model is trained according to the collection results in the next preset training period to obtain the target model.

[0007] Further, the process of determining the steady state based on the nodal marginal electricity price change rate, the ramp gradient, and the preset steady-state threshold, and obtaining the first and second determination results, includes: Several price fluctuation values ​​are determined based on the marginal electricity price change rate from the initial time to each time within the preset steady-state determination period; Several supply and demand response accelerations are determined based on the ramp gradient within the preset steady-state determination period; The steady-state index at the end of the preset steady-state determination period is determined based on all the price fluctuation values ​​and all the supply and demand response accelerations. A steady-state determination is made based on the comparison between the steady-state index and the preset steady-state threshold, resulting in either the first determination result or the second determination result.

[0008] Furthermore, the process of determining the steady-state index at the end of the preset steady-state determination period based on all the aforementioned price fluctuation values ​​and all the aforementioned supply and demand response accelerations includes: The steady-state index is calculated based on all the price fluctuation values, the preset price weights, all the supply and demand response accelerations, and the preset response weights.

[0009] Furthermore, based on the first determination result, the process of determining the abnormal transmission path type according to the cloud layer movement speed and the fractal dimension, and obtaining the supply and demand disturbance type or price manipulation type, includes: The directional consistency coefficient of the wind and cloud vector field in the satellite remote sensing image is determined based on the cloud movement speed. The periodicity index of the hanging order structure is determined based on the fractal dimension. The weather market impact index is determined based on the directional consistency coefficient and the periodic index. Based on the weather market impact index, the abnormal transmission path type is determined to be either the supply and demand disturbance type or the price manipulation type.

[0010] Furthermore, the process of determining the weather market impact index based on the directional consistency coefficient and the periodicity index includes: The weather market impact index is determined based on the directional consistency coefficient and the periodicity index within the predetermined period, the predetermined consistency weight, and the predetermined periodicity weight.

[0011] Furthermore, the process of determining whether the abnormal transmission path type is the supply-demand disturbance type or the price manipulation type based on the weather market impact index includes: Based on the comparison between the weather market impact index and the preset impact index threshold, the abnormal transmission path type is determined to be either the supply and demand disturbance type or the price manipulation type.

[0012] Furthermore, based on the aforementioned supply and demand disturbance type, the process of generating a structural deviation vector according to the nodal marginal electricity price change rate and the ramp gradient includes: The electricity price fluctuation amplitude is determined based on the marginal electricity price change rate of the node within the preset time window, and the gradient acceleration is determined based on the ramp gradient within the preset time window. The electricity price fluctuation amplitude and the ramp gradient acceleration are synchronized according to a preset time window to form a two-dimensional deviation feature sequence. The mean deviation and extreme deviation within each preset time window are calculated based on the two-dimensional deviation feature sequence, and the mean deviation and extreme deviation are combined into several structural deviation components. By connecting all the structural deviation components in chronological order, the structural deviation vector is obtained.

[0013] Furthermore, based on the aforementioned price manipulation type, the process of generating a behavioral deviation vector according to the fractal dimension and the cloud movement speed includes: The fractal dimension change rate is determined based on the fractal dimension within a preset time window, and the directional change rate of the cloud movement speed within the preset time window is calculated. The fractal dimension change rate and the cloud layer movement speed direction change rate are synchronously paired according to the preset time window to obtain a two-dimensional behavioral feature sequence. Based on the two-dimensional behavioral feature sequence, calculate the relative change deviation and correlation coefficient deviation within each preset time window to determine several behavioral deviation components; The behavior deviation vector is obtained by connecting all the behavior deviation components in chronological order.

[0014] Furthermore, the process of adjusting the preset steady-state threshold and the preset training period based on the structural bias vector and the behavioral bias vector includes: The deviation correlation is determined based on all the structural deviation vectors and all the behavioral deviation vectors from the time the first determination result is obtained to the time of the next steady-state determination; When the absolute value of the deviation correlation is greater than the preset deviation correlation threshold, the preset steady-state threshold and the preset training period are adjusted according to the deviation correlation and the preset deviation correlation threshold.

[0015] Furthermore, the process of adjusting the preset steady-state threshold and the preset training period based on the deviation correlation and the preset deviation correlation threshold includes: The preset steady-state threshold is increased based on the absolute value of the deviation correlation degree, the relative deviation of the preset deviation correlation threshold, and the preset first adjustment coefficient. The preset training period is reduced based on the absolute value of the deviation correlation, the relative deviation of the preset deviation correlation threshold, the preset second adjustment coefficient, and the lower limit of the threshold period.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: by simultaneously introducing multi-dimensional parameters such as the marginal electricity price change rate at nodes, the ramp-up gradient of major generating units, the fractal dimension of the real-time order book, and the cloud movement speed of satellite remote sensing during the clearing process of a highly volatile and constrained electricity spot market, the characteristics of price fluctuations, supply and demand dynamics, market microstructure, and external meteorological disturbances can be synergistically reflected in a unified model. In the processes of steady-state determination, abnormal transmission path identification, and deviation vector generation, a progressive mapping relationship is established between the parameters, from price changes to supply and demand responses, and then to market behavior and external disturbances. This achieves accurate characterization of market conditions and adaptive adjustment of dynamic thresholds and training cycles, ensuring that the model still has high generalization ability and prediction accuracy when facing different types of disturbances and changing scenarios. Furthermore, it can continuously optimize decision-making performance in subsequent incremental training, effectively solving the problems of poor model adaptability and insufficient prediction accuracy caused by the lack of timely identification and response to market supply and demand fluctuations and abnormal behaviors, which in turn leads to delayed risk warnings and poor market regulation effects.

[0017] Furthermore, by quantifying the standard deviation of the marginal electricity price change rate at nodes and the first difference of the unit ramp-up gradient into price fluctuation values ​​and supply-demand response acceleration, respectively, and determining the steady-state index, a precise characterization of the market operation state is achieved. Price fluctuations reflect the market's sensitivity to supply and demand, trading activity, and external disturbances, while the supply-demand response acceleration reflects the dynamic matching level between unit regulation capacity and load changes. The combination of the two can comprehensively reveal the dynamic coupling relationship between market fluctuation amplitude and regulation rate. By comparing the steady-state index with a preset threshold, it is possible to distinguish whether the data is in a stable operation or a state of violent fluctuation, providing a reliable basis for training set selection, abnormal path identification, and dynamic parameter adjustment during subsequent incremental model training, thereby improving the model's adaptability and accuracy in dealing with different market scenarios.

[0018] Furthermore, by normalizing the price fluctuation value and the acceleration of supply and demand response respectively, the market fluctuation characteristics of different dimensions and value ranges can be quantitatively compared under the same evaluation system. By combining the weighted calculation of preset price weight and preset response weight, a steady-state index that can comprehensively reflect the magnitude of market price changes and the unit's adjustment capability is obtained. Thus, at the end of the preset steady-state judgment period, the market operation status can be quantitatively assessed, which helps to accurately identify the superimposed effect of short-term market fluctuations and supply and demand imbalances, and improves the sensitivity and accuracy of steady-state judgment.

[0019] Furthermore, by utilizing cloud movement speed to reflect the dynamic impact of weather systems on regional meteorological conditions, and extracting the wind and cloud motion vector field and calculating the directional consistency coefficient through optical flow algorithms, the stability and concentration of meteorological disturbances in space are measured. Simultaneously, the complexity of the real-time order book structure in the market is analyzed through fractal dimension, and the main periodic components are extracted through Fourier transform to calculate the periodic index, thereby characterizing the regularity of market trading behavior in the time dimension. Combining meteorological disturbance characteristics with market structure characteristics forms a weather market impact index, enabling the transmission effect of meteorological changes on power supply and demand fluctuations and price anomalies to be quantified and distinguished. This effectively identifies two types of abnormal paths: supply and demand disturbance type and price manipulation type, providing accurate basis for risk warning and trading strategy optimization in the power spot market.

[0020] Furthermore, by normalizing the directional consistency coefficient and the periodicity index within a preset type determination period, the comparability of parameters with different dimensions and numerical scales is achieved, ensuring that meteorological disturbance characteristics and market order structure characteristics play a role in a unified evaluation framework. The directional normalization coefficient reflects the stability of the spatial distribution of meteorological disturbances, while the periodic normalization index characterizes the time periodic changes in the market order structure. Under the influence of weighted parameters, the two form a weather market impact index, which can quantify the transmission effect of meteorological changes on market prices and supply and demand relationships. It can achieve the synergistic fusion of multi-source heterogeneous parameters without distorting the original information, improve the accuracy and sensitivity of abnormal transmission path identification, and provide more targeted decision-making basis for early warning and strategy adjustment in the electricity spot market.

[0021] Furthermore, by comparing the weather market impact index with a preset impact index threshold, a quantitative distinction is achieved between meteorological disturbance characteristics and market structure characteristics: a high index value indicates that changes in meteorological conditions have a significant impact on the supply and demand side, thus triggering the determination of a supply and demand disturbance-type path; a low index value indicates that market price fluctuations are more due to abnormalities in trading structure or behavioral patterns, pointing to a price manipulation-type path. This determination mechanism organically combines meteorological dynamics, market order characteristics, and price change trends, enabling the rapid differentiation of different abnormal causes.

[0022] Furthermore, by combining the dynamic information of the marginal electricity price change rate at nodes with the ramp gradient of generator units, comprehensive features reflecting changes in the supply and demand structure are extracted. First, the price fluctuation amplitude is calculated using the electricity price change rate to characterize the intensity of fluctuations at the market price level. Then, the gradient acceleration is calculated using the ramp gradient to reflect the speed and magnitude of changes in power generation output. Subsequently, the two types of parameters are synchronized within a unified time window to ensure the comparability of price and supply and demand response changes. Based on the synchronized two-dimensional feature sequence, the mean deviation and extreme deviation of each window are calculated to capture both stable and sudden shifts. Finally, these deviation components are connected in chronological order to form a structural deviation vector, realizing a continuous quantitative description of the dynamic shift pattern of the market supply and demand structure, and providing an accurate basis for identifying structural anomalies.

[0023] Furthermore, by simultaneously constructing behavioral feature sequences based on the fractal dimension change rate and the cloud layer movement speed direction change rate, the interaction between the complex dynamics of the market order structure and changes in the external environment is effectively revealed. By calculating the relative change deviation and correlation coefficient deviation, the temporal characteristics of abnormal behavior can be accurately characterized, thus forming a behavioral deviation vector with temporal coherence. This method fully reflects the intrinsic connection between market internal structural fluctuations and environmental disturbances, improves the ability to identify price manipulation and the model's response sensitivity, thereby providing scientific and dynamic support for risk monitoring and regulation in the electricity spot market.

[0024] Furthermore, by calculating the correlation between structural deviation vectors and behavioral deviation vectors, the intrinsic link between market supply and demand dynamics and abnormal behavior is effectively captured. When the deviation correlation exceeds a preset threshold, the threshold for steady-state determination and the training period can be dynamically adjusted to improve the model's sensitivity and adaptability to market fluctuations, thereby more accurately reflecting the real-time status and potential risks of the electricity spot market. This dynamic adjustment mechanism rationally utilizes the synergistic changes among multi-dimensional parameters to achieve a fine characterization and effective response to complex market environments, enhancing the accuracy and reliability of risk monitoring and early warning.

[0025] Furthermore, by comparing the deviation correlation with a preset deviation correlation threshold, and adjusting the steady-state threshold and training period based on their relative deviations, the model can automatically raise the steady-state judgment threshold and shorten the model update cycle when structural and behavioral deviations are highly correlated, thereby enhancing its sensitivity and response speed to sudden market disturbances. When the correlation is low, a longer training period and a lower steady-state threshold are maintained to preserve the model's stability and generalization ability. This mechanism utilizes the dynamic coupling relationship between parameters to make the threshold adjustment magnitude proportional to the correlation strength, while introducing a lower limit on the period to prevent over-adjustment, balancing real-time performance and robustness, and effectively improving the accuracy and reliability of risk identification and control in the electricity spot market. Attached Figure Description

[0026] Figure 1 This is a flowchart of the data model construction method for electricity spot trading based on data analysis in this embodiment; Figure 2 This is the logic diagram for steady-state determination in this embodiment; Figure 3 This is the logic diagram for determining the type of abnormal propagation path in this embodiment; Figure 4 The logic diagram for adjusting the preset steady-state threshold and preset training period in this embodiment is shown. Detailed Implementation

[0027] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0028] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0029] Please see Figure 1 The diagram shown is a flowchart of the data model construction method for electricity spot trading based on data analysis in this embodiment. This embodiment provides a data model construction method for electricity spot trading based on data analysis, including: The system collects data in real time on the marginal electricity price change rate at nodes, the ramp-up gradient of major generating units, the fractal dimension of the real-time order book, and the cloud movement speed in satellite remote sensing images during the clearing process of the highly volatile and restricted electricity spot market. The collected data is then preprocessed within a preset training period to obtain an initial training set. Construct the incremental model based on the initial training set; Based on the model to be incremented, a steady-state determination is made according to the node marginal electricity price change rate, the ramp gradient and the preset steady-state threshold to obtain a first determination result or a second determination result. Based on the first determination result, the abnormal transmission path type is determined according to the cloud layer movement speed and the fractal dimension, resulting in either a supply and demand disturbance type or a price manipulation type. Based on the supply and demand disturbance type, a structural deviation vector is generated according to the marginal electricity price change rate at the node and the ramp gradient; based on the price manipulation type, a behavioral deviation vector is generated according to the fractal dimension and the cloud movement speed. The preset steady-state threshold and the preset training period are adjusted based on the structural deviation vector and the behavioral deviation vector; Based on the second determination result, the incremental model is trained according to the collection results in the next preset training period to obtain the target model.

[0030] In this embodiment, the high-volatility restricted electricity spot market refers to an electricity spot trading market where the marginal price of electricity at each node fluctuates significantly in a short period of time, but is constrained by operational constraints such as unit ramp-up capacity, inter-regional transmission channel capacity, frequency regulation reserves, and safety margins, thus limiting the price fluctuation range to a certain range. This type of market is common in regions with a high proportion of renewable energy and rapid load changes. Its clearing process refers to the entire process by which the market operator, in each clearing cycle, calculates the marginal price (LMP) of each trading node and the cleared electricity volume using a clearing algorithm based on the bidding information submitted by the generation and consumption sides, combined with grid power flow calculations, operational constraints, transmission capacity limitations, and dispatch strategies, and generates a price-electricity allocation scheme. During the above clearing process, in order to simultaneously depict the internal supply and demand status of the market and external weather conditions... The comprehensive impact of disturbances was assessed by collecting four types of parameters in real time: the marginal electricity price change rate at nodes (the percentage change in LMP between two adjacent clearing cycles, obtained through the price data interface of the market operator and calculated using time series difference), the ramp gradient of the main power generating units (the change rate of active power output per unit time, obtained and calculated in real time through the SCADA or EMS system of the dispatch center), the fractal dimension of the real-time order book (characterizing the complexity of the order distribution, calculated using snapshot data of the order book on the trading platform and the multi-scale bin counting method), and the cloud movement speed in satellite remote sensing images introduced based on the characteristic that new energy output is significantly affected by cloud cover (obtained by receiving continuous cloud image data from geostationary or low-orbit meteorological satellites and extracting cloud displacement vectors using optical flow or feature point matching algorithms).

[0031] In this embodiment, the incremental model constructed based on the initial training set is trained by features such as the rate of change of marginal electricity price at nodes, the ramp gradient, the fractal dimension of the real-time order book, and the cloud movement speed. The model can capture the operating rules and potential abnormal patterns of the electricity spot market under different weather and load conditions.

[0032] The preset training period is the time window for statistical and analytical data before incremental training of the model. It depends on the frequency of fluctuations in the electricity spot market price, the response characteristics of the generating units, and the cycle of external disturbances. It is usually set between 1 hour and 24 hours. In this embodiment, it is set to 6 hours, which can capture the short-cycle fluctuation characteristics of the market and reduce invalid data interference while taking into account the representativeness of the data and the real-time performance of the calculation.

[0033] The preset steady-state threshold is a critical value of the steady-state index used to determine whether the market is stable. It depends on the statistical distribution characteristics of historical operating data, the setting of different parameter weights, and the tolerance for abnormal fluctuations. It is usually set between 0.3 and 0.7. In this embodiment, it is set to 0.5, which can effectively distinguish whether the market is in a stable adjustment phase or a disturbed fluctuation phase, and provide a clear standard for the selection of model training data.

[0034] By simultaneously introducing multi-dimensional parameters such as the marginal rate of change of nodal prices, the ramp-up gradient of major generating units, the fractal dimension of the real-time order book, and the cloud movement speed from satellite remote sensing during the clearing process of the highly volatile and constrained electricity spot market, the model enables the coordinated representation of price fluctuation characteristics, supply and demand dynamics, market microstructure characteristics, and external meteorological disturbances within a unified model. In the processes of steady-state determination, anomaly transmission path identification, and deviation vector generation, a progressive mapping relationship is established between the parameters, from price changes to supply and demand responses, and then to market behavior and external disturbances. This achieves accurate characterization of market conditions and adaptive adjustment of dynamic thresholds and training cycles, ensuring that the model maintains high generalization ability and prediction accuracy when facing different types of disturbances and changing scenarios. Furthermore, it allows for continuous optimization of decision-making performance in subsequent incremental training. This effectively solves the problem of poor model adaptability and insufficient prediction accuracy caused by a lack of timely identification and response to market supply and demand fluctuations and anomalies, leading to delayed risk warnings and ineffective market regulation.

[0035] Please see Figure 2 The diagram shown illustrates the logic for steady-state determination in this embodiment. In this embodiment, the process of determining steady-state based on the marginal electricity price change rate at each node, the ramp gradient, and a preset steady-state threshold, to obtain a first determination result and a second determination result, includes: calculating the standard deviation of the marginal electricity price change rate from the initial time to each time point within the preset steady-state determination period to determine several price fluctuation values; calculating the first difference of the ramp gradient between any two adjacent times within the preset steady-state determination period to determine several supply and demand response accelerations; determining the steady-state index at the end of the preset steady-state determination period based on all price fluctuation values ​​and all supply and demand response accelerations; determining the incremental training data as unstable when the steady-state index is greater than the preset steady-state threshold, thus obtaining a first determination result; and determining the incremental training data as stable when the steady-state index is less than or equal to the preset steady-state threshold, thus obtaining a second determination result.

[0036] The preset steady-state determination period is a continuous time window selected when calculating the steady-state index. It depends on the dynamic change rate of electricity price fluctuations and unit response, as well as the market clearing frequency. It is usually set between 5 minutes and 60 minutes. In this embodiment, it is set to 30 minutes, which can accurately reflect the transient characteristics of the market while balancing short-term sensitivity and fluctuation smoothness.

[0037] By quantifying the standard deviation of the marginal electricity price change rate at nodes and the first difference of the unit ramp-up gradient into price volatility and supply-demand response acceleration, respectively, and determining the steady-state index, a precise characterization of the market's operating state is achieved. Price volatility reflects the market's sensitivity to supply and demand, trading activity, and external disturbances, while supply-demand response acceleration reflects the dynamic matching level between unit regulation capacity and load changes. Combining these two metrics comprehensively reveals the dynamic coupling relationship between market volatility amplitude and regulation rate. By comparing the steady-state index with a preset threshold, it is possible to distinguish between stable operation and volatile conditions, providing a reliable basis for subsequent incremental model training, including training set selection, anomaly path identification, and dynamic parameter adjustment, thereby improving the model's adaptability and accuracy in dealing with different market scenarios.

[0038] Specifically, the process of determining the steady-state index at the end of the preset steady-state determination period based on all the price fluctuation values ​​and all the supply and demand response accelerations includes: normalizing the price fluctuation values ​​to obtain a price normalization value, normalizing the supply and demand response accelerations to obtain a response normalization value; and calculating the steady-state index at the end of the preset steady-state determination period based on the price normalization value, the response normalization value, the preset price weight, and the preset response weight, Q = a × A + b × B, where Q is the steady-state index, a is the preset price weight, A is the price normalization value, b is the preset response weight, and B is the response normalization value.

[0039] The preset price weight is the proportion of influence assigned to the normalized price value in the steady-state index calculation. It depends on the importance of price fluctuations in the market stability assessment and the historical fluctuation pattern. It is usually set between 0.4 and 0.7. In this embodiment, it is set to 0.6, which can highlight the contribution of price changes to the steady-state determination.

[0040] The preset response weight is the proportion of influence assigned to the normalized value of supply and demand response in the steady-state index calculation. It depends on the importance of the unit ramp-up characteristics and the speed of supply and demand adjustment, and is usually set between 0.3 and 0.6. In this embodiment, it is set to 0.4, which can take into account the auxiliary judgment role of supply and demand response on market stability.

[0041] By normalizing the price fluctuation value and the acceleration of supply and demand response respectively, the market fluctuation characteristics of different dimensions and value ranges can be quantitatively compared under the same evaluation system. By combining the weighted calculation of preset price weight and preset response weight, a steady-state index that can comprehensively reflect the magnitude of market price changes and the unit's adjustment capability is obtained. Thus, at the end of the preset steady-state judgment period, the market operation status can be quantitatively assessed, which helps to accurately identify the superimposed effect of short-term market fluctuations and supply and demand imbalances, and improves the sensitivity and accuracy of steady-state judgment.

[0042] Specifically, based on the first determination result, the process of determining the type of abnormal transmission path according to the cloud layer movement speed and the fractal dimension, and obtaining the supply and demand disturbance type or price manipulation type, includes: An optical flow algorithm is used to calculate the displacement vectors of cloud pixel blocks in two consecutive satellite remote sensing images, thereby obtaining the wind and cloud motion vector field for each pixel. The magnitude of the vector represents the cloud's movement speed, and the direction represents the cloud's azimuth. The direction angles of all effective vectors in the vector field are converted into unit direction vectors, and the magnitude of their vector sum is normalized to obtain a direction consistency coefficient. A Fourier transform is performed on the fractal dimension sequence to extract the main periodic components, and the periodicity index is calculated. The weather market impact index is determined based on the direction consistency coefficient and the periodicity index. Based on the weather market impact index, the abnormal transmission path type is determined to be either the supply and demand disturbance type or the price manipulation type.

[0043] By utilizing cloud movement speed to reflect the dynamic impact of weather systems on regional meteorological conditions, and extracting the wind and cloud motion vector field and calculating the directional consistency coefficient through optical flow algorithms, the stability and concentration of meteorological disturbances in space are measured. Simultaneously, the complexity of the real-time order book structure in the market is analyzed through fractal dimension, and the main periodic components are extracted through Fourier transform to calculate the periodic index, thereby characterizing the regularity of market trading behavior in the time dimension. Combining meteorological disturbance characteristics with market structure characteristics forms a weather market impact index, enabling the quantification and differentiation of the transmission effects of meteorological changes on power supply and demand fluctuations and price anomalies. This effectively identifies two types of abnormal paths: supply and demand disturbances and price manipulation, providing precise basis for risk warning and trading strategy optimization in the power spot market.

[0044] Specifically, the process of determining the weather market impact index based on the directional consistency coefficient and the periodic index includes: Based on the consistency coefficients of all directions within a predetermined period, the consistency coefficients of the directions at the end of the predetermined period are normalized to obtain the direction normalization coefficient. Based on the consistency coefficients of all directions within a predetermined period, the periodicity index at the end of the predetermined period is normalized to obtain the period normalization index. Based on the predetermined periodicity weight, predetermined consistency weight, direction normalization coefficient, and period normalization index, the weather market impact index is determined, U = i × I + j × J, where U is the weather market impact index, i is the predetermined consistency weight, I is the direction normalization coefficient, j is the predetermined periodicity weight, and J is the period normalization index.

[0045] The preset type determination period is the time window used in the abnormal path type determination to statistically and normally normalize the directional consistency coefficient and periodicity index. It depends on the response delay characteristics of the target electricity spot market price and supply and demand fluctuations, and is usually set between 5 minutes and 60 minutes. In this embodiment, it is set to 30 minutes, which can take into account both the propagation timeliness of meteorological disturbances and the stability of changes in market structure characteristics.

[0046] The preset consistency weight is the weight coefficient of the directional consistency coefficient when calculating the weather market impact index. It depends on the sensitivity of meteorological disturbances to price fluctuations and is usually set between 0.3 and 0.7. In this embodiment, it is set to 0.5, which can balance the contribution of meteorological factors and market order structure factors in the index calculation.

[0047] By normalizing the directional consistency coefficient and periodicity index within a preset type determination period, the comparability of parameters with different dimensions and numerical scales is achieved, ensuring that meteorological disturbance characteristics and market order structure characteristics play a role in a unified evaluation framework. The directional normalization coefficient reflects the stability of the spatial distribution of meteorological disturbances, while the periodic normalization index characterizes the time periodic changes in the market order structure. Under the influence of weighted parameters, the two form a weather market impact index, which can quantify the transmission effect of meteorological changes on market prices and supply and demand. It can achieve the synergistic fusion of multi-source heterogeneous parameters without distorting the original information, improve the accuracy and sensitivity of abnormal transmission path identification, and provide more targeted decision-making basis for early warning and strategy adjustment in the electricity spot market.

[0048] Please see Figure 3 As shown, this is the logic diagram for determining the type of abnormal transmission path in this embodiment. In this embodiment, the process of determining the type of abnormal transmission path as either the supply and demand disturbance type or the price manipulation type based on the weather market impact index includes: determining the type of abnormal transmission path as the supply and demand disturbance type when the weather market impact index is greater than or equal to a preset impact index threshold; and determining the type of abnormal transmission path as the price manipulation type when the weather market impact index is less than the preset impact index threshold.

[0049] The preset impact index threshold is the boundary value that distinguishes between supply and demand disturbances and price manipulation. It depends on the distribution characteristics of the weather market impact index under the two types of abnormal situations in historical data. It is usually set between 0.4 and 0.6. In this embodiment, it is set to 0.5, which can achieve accurate identification of abnormal path types.

[0050] By comparing the weather market impact index with a preset impact index threshold, a quantitative distinction is achieved between meteorological disturbance characteristics and market structure characteristics: a high index value indicates that changes in meteorological conditions have a significant impact on the supply and demand side, thus triggering the determination of a supply and demand disturbance-type path; a low index value indicates that market price fluctuations are more due to abnormalities in trading structure or behavioral patterns, pointing to a price manipulation-type path. This determination mechanism organically combines meteorological dynamics, market order characteristics, and price change trends, enabling the rapid differentiation of different abnormal causes.

[0051] Specifically, based on the supply and demand disturbance type, the process of generating a structural deviation vector according to the nodal marginal electricity price change rate and the ramp gradient includes: determining the electricity price fluctuation amplitude according to the nodal marginal electricity price change rate within a preset time window, and determining the gradient acceleration according to the ramp gradient within the preset time window; synchronizing the electricity price fluctuation amplitude and the ramp gradient acceleration according to the preset time window to form a two-dimensional deviation feature sequence; calculating the mean deviation and extreme deviation within each preset time window according to the two-dimensional deviation feature sequence, and combining the mean deviation and extreme deviation into several structural deviation components; connecting all the structural deviation components in chronological order to obtain the structural deviation vector.

[0052] The preset time window is used to divide the time interval for data collection and analysis. It depends on the fluctuation frequency of the electricity spot market and the monitoring accuracy requirements. It is usually set between 5 minutes and 6 hours. In this embodiment, it is set to 15 minutes, which can effectively capture the short-term change characteristics of electricity price and ramp gradient, and ensure the timeliness and accuracy of structural deviation vector.

[0053] By combining the dynamic information of the marginal electricity price change rate at nodes with the ramp gradient of generator units, a comprehensive feature reflecting changes in the supply and demand structure is extracted. First, the price fluctuation amplitude is calculated using the electricity price change rate to characterize the intensity of fluctuations at the market price level. Then, the gradient acceleration is calculated using the ramp gradient to reflect the speed and magnitude of changes in power generation output. Subsequently, the two types of parameters are synchronized within a unified time window to ensure the comparability of price and supply and demand response changes. Based on the synchronized two-dimensional feature sequence, the mean deviation and extreme value deviation of each window are calculated to capture both stable and sudden shifts. Finally, these deviation components are connected in chronological order to form a structural deviation vector, achieving a continuous quantitative description of the dynamic shift pattern of the market supply and demand structure, and providing an accurate basis for identifying structural anomalies.

[0054] Specifically, based on the price manipulation model, the process of generating a behavioral deviation vector according to the fractal dimension and the cloud movement speed includes: determining the fractal dimension change rate according to the fractal dimension within a preset time window, and calculating the directional change rate of the cloud movement speed within the preset time window; synchronously pairing the fractal dimension change rate and the directional change rate of the cloud movement speed according to the preset time window to obtain a two-dimensional behavioral feature sequence; calculating the relative change deviation and correlation coefficient deviation within each preset time window based on the two-dimensional behavioral feature sequence to determine several behavioral deviation components; and connecting all the behavioral deviation components in chronological order to obtain the behavioral deviation vector.

[0055] By simultaneously constructing behavioral feature sequences based on the fractal dimension change rate and the cloud movement speed and direction change rate, this method effectively reveals the interaction between the complex dynamics of market order structure and changes in the external environment. By calculating relative change deviations and correlation coefficient deviations, the temporal characteristics of abnormal behavior can be accurately characterized, thus forming a behavioral deviation vector with temporal coherence. This method fully reflects the intrinsic connection between internal market structural fluctuations and environmental disturbances, improving the ability to identify price manipulation and the model's response sensitivity, thereby providing scientific and dynamic support for risk monitoring and regulation in the electricity spot market.

[0056] Please see Figure 4 As shown, this is a logic diagram for adjusting the preset steady-state threshold and the preset training period in this embodiment. In this embodiment, the process of adjusting the preset steady-state threshold and the preset training period according to the structural deviation vector and the behavioral deviation vector includes: calculating the Pearson correlation coefficient of all the structural deviation vectors and behavioral deviation vectors from the time of obtaining the first determination result to the time of the next steady-state determination, and obtaining the deviation correlation degree; when the absolute value of the deviation correlation degree is greater than the preset deviation correlation threshold, adjusting the preset steady-state threshold and the preset training period according to the deviation correlation degree and the preset deviation correlation threshold.

[0057] The preset deviation correlation threshold is a threshold used to determine the degree of correlation between structural deviation vectors and behavioral deviation vectors. It depends on the statistical characteristics of historical market data and the model's need for anomaly sensitivity, and is usually set between 0.6 and 0.9. In this embodiment, it is set to 0.75, which can effectively distinguish significant deviation correlation from normal fluctuations, thereby supporting dynamic adjustment of steady-state threshold and training cycle, and improving the flexibility and accuracy of the model.

[0058] By calculating the correlation between structural deviation vectors and behavioral deviation vectors, the intrinsic link between market supply and demand dynamics and abnormal behavior is effectively captured. When the deviation correlation exceeds a preset threshold, the steady-state determination threshold and training period can be dynamically adjusted to improve the model's sensitivity and adaptability to market fluctuations, thereby more accurately reflecting the real-time status and potential risks of the electricity spot market. This dynamic adjustment mechanism rationally utilizes the synergistic changes among multi-dimensional parameters to achieve a fine characterization and effective response to complex market environments, enhancing the accuracy and reliability of risk monitoring and early warning.

[0059] Specifically, the process of adjusting the preset steady-state threshold and the preset training period based on the deviation correlation and the preset deviation correlation threshold includes: increasing the preset steady-state threshold based on the absolute value of the deviation correlation and the relative deviation of the preset deviation correlation threshold, and a preset first adjustment coefficient, H'=H×[1+k1×(R-R0) / R0], where H' is the increased preset steady-state threshold, H is the original preset steady-state threshold, k1 is the preset first adjustment coefficient, R is the absolute value of the deviation correlation, and R0 is the preset deviation correlation threshold; decreasing the preset training period based on the absolute value of the deviation correlation and the relative deviation of the preset deviation correlation threshold, a preset second adjustment coefficient, and a lower limit of the threshold period, T'=T×max[1-k2×(R-R0) / R0), P], where T' is the decreased preset training period, T is the original preset training period, k2 is the preset second adjustment coefficient, and P is the lower limit of the threshold period.

[0060] The preset deviation correlation threshold is a reference value used to determine whether the correlation between structural deviation vectors and behavioral deviation vectors has reached a significant level. It depends on the correlation distribution characteristics of the two types of deviation vectors in historical market data and the sensitivity requirements for risk identification. It is usually set between 0.3 and 0.7. In this embodiment, it is set to 0.5, which can accurately identify market disturbances with high deviation correlation while taking into account the sensitivity of anomaly detection and the false alarm rate.

[0061] The preset first adjustment coefficient is a proportional factor used to dynamically increase the steady-state threshold based on the correlation between the deviation and the relative deviation magnitude. It depends on the sensitivity requirements and fault tolerance range of the steady-state determination under different market fluctuation levels, and is usually set between 0.1 and 0.5. In this embodiment, it is set to 0.2, which can moderately increase the steady-state threshold when the correlation increases significantly, thereby avoiding over-triggering of model training caused by short-term abnormal correlation.

[0062] The preset second adjustment coefficient is a proportional factor used to dynamically shorten the training cycle based on the relative deviation magnitude of the deviation correlation. It depends on the balance between the model update speed and the stability of the training data, and is usually set between 0.1 and 0.4. In this embodiment, it is set to 0.15, which can accelerate the model update rhythm in the case of high correlation, introduce the latest market features in a timely manner to improve the risk response speed, and at the same time prevent excessively frequent adjustments by combining the lower limit of the cycle.

[0063] By comparing the correlation between deviations with a preset correlation threshold and adjusting the steady-state threshold and training period based on their relative deviations, the model can automatically raise the steady-state judgment threshold and shorten the model update cycle when structural and behavioral deviations are highly correlated, thereby enhancing its sensitivity and response speed to sudden market disturbances. Conversely, when the correlation is low, a longer training period and a lower steady-state threshold are maintained to preserve the model's stability and generalization ability. This mechanism utilizes the dynamic coupling relationship between parameters to make the threshold adjustment magnitude proportional to the correlation strength, while introducing a lower limit on the period to prevent over-adjustment, balancing real-time performance and robustness, and effectively improving the accuracy and reliability of risk identification and control in the electricity spot market.

[0064] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing a data model for electricity spot trading based on data analysis, characterized in that, include: The system collects data in real time on the marginal electricity price change rate at nodes, the ramp-up gradient of major generating units, the fractal dimension of the real-time order book, and the cloud movement speed in satellite remote sensing images during the clearing process of the highly volatile and restricted electricity spot market. The collected data is then preprocessed within a preset training period to obtain an initial training set. Construct the incremental model based on the initial training set; Based on the model to be incremented, a steady-state determination is made according to the node marginal electricity price change rate, the ramp gradient and the preset steady-state threshold to obtain a first determination result or a second determination result. Based on the first determination result, the abnormal transmission path type is determined according to the cloud layer movement speed and the fractal dimension, resulting in either a supply and demand disturbance type or a price manipulation type. Based on the supply and demand disturbance type, a structural deviation vector is generated according to the marginal electricity price change rate at the node and the ramp gradient; based on the price manipulation type, a behavioral deviation vector is generated according to the fractal dimension and the cloud movement speed. The preset steady-state threshold and the preset training period are adjusted based on the structural deviation vector and the behavioral deviation vector; Based on the second determination result, the incremental model is trained according to the collection results in the next preset training period to obtain the target model.

2. The method for constructing a data model for electricity spot trading based on data analysis according to claim 1, characterized in that, The process of determining steady-state conditions based on the nodal marginal electricity price change rate, the ramp gradient, and a preset steady-state threshold, and obtaining the first and second determination results, includes: Several price fluctuation values ​​are determined based on the marginal electricity price change rate from the initial time to each time within the preset steady-state determination period; Several supply and demand response accelerations are determined based on the ramp gradient within the preset steady-state determination period; The steady-state index at the end of the preset steady-state determination period is determined based on all the price fluctuation values ​​and all the supply and demand response accelerations. A steady-state determination is made based on the comparison between the steady-state index and the preset steady-state threshold, resulting in either the first determination result or the second determination result.

3. The method for constructing a data model for electricity spot trading based on data analysis according to claim 2, characterized in that, The process of determining the steady-state index at the end of the preset steady-state determination period based on all the aforementioned price fluctuation values ​​and all the aforementioned supply and demand response accelerations includes: The steady-state index is calculated based on all the price fluctuation values, the preset price weights, all the supply and demand response accelerations, and the preset response weights.

4. The method for constructing a data model for electricity spot trading based on data analysis according to claim 3, characterized in that, Based on the first determination result, the process of determining the abnormal transmission path type according to the cloud layer movement speed and the fractal dimension, and obtaining the supply and demand disturbance type or the price manipulation type, includes: The directional consistency coefficient of the wind and cloud vector field in the satellite remote sensing image is determined based on the cloud movement speed. The periodicity index of the hanging order structure is determined based on the fractal dimension. The weather market impact index is determined based on the directional consistency coefficient and the periodic index. Based on the weather market impact index, the abnormal transmission path type is determined to be either the supply and demand disturbance type or the price manipulation type.

5. The method for constructing a data model for electricity spot trading based on data analysis according to claim 4, characterized in that, The process of determining the weather market impact index based on the directional consistency coefficient and the periodic index includes: The weather market impact index is determined based on the directional consistency coefficient and the periodicity index within the predetermined period, the predetermined consistency weight, and the predetermined periodicity weight.

6. The method for constructing a data model for electricity spot trading based on data analysis according to claim 5, characterized in that, The process of determining whether an abnormal transmission path type is the supply-demand disturbance type or the price manipulation type based on the weather market impact index includes: Based on the comparison between the weather market impact index and the preset impact index threshold, the abnormal transmission path type is determined to be either the supply and demand disturbance type or the price manipulation type.

7. The method for constructing a data model for electricity spot trading based on data analysis according to claim 6, characterized in that, Based on the aforementioned supply and demand disturbance type, the process of generating a structural deviation vector according to the nodal marginal electricity price change rate and the ramp gradient includes: The electricity price fluctuation amplitude is determined based on the marginal electricity price change rate of the node within the preset time window, and the gradient acceleration is determined based on the ramp gradient within the preset time window. The electricity price fluctuation amplitude and the climbing gradient acceleration are synchronized according to a preset time window to form a two-dimensional deviation feature sequence. The mean deviation and extreme deviation within each preset time window are calculated based on the two-dimensional deviation feature sequence, and the mean deviation and extreme deviation are combined into several structural deviation components. By connecting all the structural deviation components in chronological order, the structural deviation vector is obtained.

8. The method for constructing a data model for electricity spot trading based on data analysis according to claim 7, characterized in that, Based on the aforementioned price manipulation type, the process of generating a behavioral deviation vector according to the fractal dimension and the cloud movement speed includes: The fractal dimension change rate is determined based on the fractal dimension within a preset time window, and the directional change rate of the cloud movement speed within the preset time window is calculated. The fractal dimension change rate and the cloud layer movement speed direction change rate are synchronously paired according to the preset time window to obtain a two-dimensional behavioral feature sequence. Based on the two-dimensional behavioral feature sequence, calculate the relative change deviation and correlation coefficient deviation within each preset time window to determine several behavioral deviation components; The behavior deviation vector is obtained by connecting all the behavior deviation components in chronological order.

9. The method for constructing a data model for electricity spot trading based on data analysis according to claim 8, characterized in that, The process of adjusting the preset steady-state threshold and the preset training period based on the structural bias vector and the behavioral bias vector includes: The deviation correlation is determined based on all the structural deviation vectors and all the behavioral deviation vectors from the time the first determination result is obtained to the time of the next steady-state determination; When the absolute value of the deviation correlation is greater than the preset deviation correlation threshold, the preset steady-state threshold and the preset training period are adjusted according to the deviation correlation and the preset deviation correlation threshold.

10. The method for constructing a data model for electricity spot trading based on data analysis according to claim 9, characterized in that, The process of adjusting the preset steady-state threshold and the preset training period based on the deviation correlation and the preset deviation correlation threshold includes: The preset steady-state threshold is increased based on the absolute value of the deviation correlation degree, the relative deviation of the preset deviation correlation threshold, and the preset first adjustment coefficient. The preset training period is reduced based on the absolute value of the deviation correlation, the relative deviation of the preset deviation correlation threshold, the preset second adjustment coefficient, and the lower limit of the threshold period.

Images