An electricity price fluctuation prediction method and system based on artificial intelligence

Abstract

This application provides an artificial intelligence-based method and system for predicting electricity price fluctuations, relating to the field of electricity price management, and solves the technical problem of difficulty in assessing the impact of enterprise behavior on market electricity prices. The method includes: S1: constructing an aggregated entity and planning a prediction period, and obtaining the predicted user electricity consumption value of the aggregated entity within the prediction period; S2: obtaining the grid base price corresponding to the aggregated entity within the prediction period; S3: establishing a price influence quantification model that reflects the relationship between the aggregated entity's behavior and electricity prices, generating a price influence coefficient based on the predicted user electricity consumption value and the price influence quantification model; determining the fluctuating electricity price based on the price influence coefficient and the grid base price. This application can improve the accuracy of market electricity price prediction and the enterprise's own electricity price perception ability.

Description

Technical Field

[0001] This application relates to the field of electricity price management, and in particular to an artificial intelligence-based method and system for predicting electricity price fluctuations. Background Technology

[0002] The traditional power industry is largely vertically integrated, with electricity prices set by regulators and relatively stable. Since the 1990s, many countries worldwide have implemented power market reforms, the core of which is "separation of power generation and grid connection, and competitive bidding for grid connection." This reform introduced market competition into the power generation and sales stages. Electricity prices are no longer fixed values ​​but are determined by supply and demand, just like other commodities, resulting in significant, and sometimes even drastic, fluctuations. Furthermore, as a special commodity, electricity prices exhibit far more complex fluctuation patterns than ordinary commodities, posing a significant challenge to forecasting.

[0003] Currently, most AI-based electricity price fluctuation prediction methods struggle to shift from passively accepting market prices to assessing the impact of their own behavior on market electricity prices. For a specific "aggregate entity," such as a large industrial park, they overlook the fact that their own electricity consumption behavior will in turn affect electricity prices, leading to inaccurate predictions. Furthermore, they struggle to build a personalized electricity price impact prediction mechanism for aggregate entities that takes their own behavior into account, making them aware of how much more electricity they consume will increase the price, thus hindering production optimization.

[0004] Therefore, the present invention provides an artificial intelligence-based method and system for predicting electricity price fluctuations to solve the above problems. Summary of the Invention

[0005] This application provides an artificial intelligence-based method and system for predicting electricity price fluctuations, which solves the technical problem that existing technologies struggle to assess the impact of corporate behavior on market electricity prices when predicting them.

[0006] To achieve the above objectives, this application adopts the following technical solution: Firstly, an artificial intelligence-based method for predicting electricity price fluctuations is provided, including: S1: Construct aggregated entities and plan forecast periods, and obtain the predicted user electricity consumption values ​​of the aggregated entities within the forecast periods; wherein, aggregated entities include enterprises and factories, etc. S2: Obtain the basic electricity price of the grid corresponding to the aggregated entity within the prediction period; S3: Establish a price influence quantification model that can reflect the relationship between aggregate entity behavior and electricity price; generate price influence coefficients based on user electricity consumption forecasts and the price influence quantification model; determine fluctuating electricity prices based on price influence coefficients and grid base electricity prices.

[0007] In conjunction with the first aspect mentioned above, one possible implementation involves constructing an aggregate entity, including: Enterprises or factories participating in electricity price fluctuation forecasting are marked as target units, and these target units under unified management are manually marked as an aggregated entity; wherein, the unified management includes enterprises or factories belonging to the same company.

[0008] In conjunction with the first aspect mentioned above, one possible implementation method for planning and forecasting time periods includes: When the target time is reached, the historical electricity price of each time point in the region where the aggregated entity is located within the n days prior to the current target time is obtained, and the average value of the historical electricity price at each time point is marked as the reference electricity price at that time point; wherein, the target time and n are obtained manually, the target time is generally 0:01, and n is generally 30; When the reference electricity price is greater than the first electricity price threshold, the corresponding time point is marked as a peak time point; when the reference electricity price is not greater than the first electricity price threshold and not less than the second electricity price threshold, the corresponding time point is marked as a flat time point; when the reference electricity price is less than the second electricity price threshold, the corresponding time point is marked as a low time point. Here, the first and second electricity price thresholds are determined based on the reference electricity price. For example, if the reference electricity prices of the previous n days are sorted from largest to smallest, the first electricity price threshold is the reference electricity price at the 25th percentile, and the second electricity price threshold is the reference electricity price at the 70th percentile. Peak, off-peak, and low-peak time points are integrated to obtain peak, off-peak, and off-peak time domains. Weighting factors are determined based on the average electricity price within each of these time domains. These weighting factors include... Weighting factors and weighting factor ; The dynamic division duration for each time domain is obtained by multiplying the standard division duration by the weighting factor of each time domain and rounding it up; where rounding up means rounding up to the minute; the standard division duration is determined based on the manager's precision in electricity price forecasting. The higher the precision of electricity price forecasting, the smaller the standard division duration, and the lower the precision of electricity price forecasting, the larger the standard division duration. Based on the dynamic division duration of each time domain, the corresponding time domain is divided into several prediction periods.

[0009] In conjunction with the first aspect mentioned above, one possible implementation involves determining the weighting factors for the peak, off-peak, and flat time domains based on the average electricity price within the peak, off-peak, and flat time domains, including: Obtain the average electricity price during peak time. Average electricity price during off-peak hours Average electricity price over a flat period of time Based on average electricity price Average electricity price and average electricity price The weight partitioning factor for the corresponding time domain is determined by formula (1). ;in, For the time domain numbering, when The time period represents the peak time range, when The time period represents the trough time domain, when The time period represents the average time domain; The calculation formula (1) is: .

[0010] In conjunction with the first aspect above, in one possible implementation, obtaining the predicted user electricity consumption value of the aggregation entity within the prediction period includes: Extract the product types and production quantities of each type planned by the aggregated entity in each forecast period. Obtain the electricity consumption of each product type produced by the aggregated entity in the n days prior to the current time. Multiply the electricity consumption of each product type by the corresponding production quantity to determine the total electricity consumption of each type of product in each forecast period. Add the total electricity consumption of different types of products in the same forecast period to obtain the predicted user electricity consumption value in the forecast period.

[0011] In conjunction with the first aspect above, one possible implementation involves obtaining the grid base price corresponding to the aggregated entity within the prediction period, including: Obtain the predicted user electricity consumption values ​​of other aggregated entities within the same area as the aggregated entity during the prediction period, and add several of these predicted user electricity consumption values ​​to obtain a reference electricity consumption prediction value; obtain the electricity consumption in the historical data of the area within the same duration as the prediction period, where the deviation from the reference electricity consumption prediction value does not exceed a deviation threshold, and integrate the electricity prices corresponding to these electricity consumption values ​​into an electricity price group; wherein, the deviation threshold is set based on the magnitude of the reference electricity consumption prediction value, and can be set to 3% of the reference electricity consumption prediction value; Obtain the variance of the electricity price group and determine whether the variance is greater than the electricity price variance threshold. If yes, obtain the average electricity price in the electricity price group, remove the electricity price with the largest absolute value of the difference from the average electricity price, re-calculate the variance of the electricity price group and re-determine the variance until the variance of the electricity price group is not greater than the electricity price variance threshold. Then, calculate the average value of the electricity prices retained in the electricity price group to obtain the grid base price. If no, calculate the average value of the electricity prices retained in the electricity price group to obtain the grid base price. The electricity price variance threshold is obtained through empirical setting.

[0012] In conjunction with the first aspect mentioned above, one possible implementation involves establishing a price influence quantification model that reflects the relationship between aggregate entity behavior and electricity prices, including: Extract the user electricity consumption forecast and price impact coefficient of each aggregated entity from the historical reference data; where the historical reference data includes the user electricity consumption forecast and the price impact coefficient set by experts based on the user electricity consumption forecast and the corresponding electricity price; The predicted user electricity consumption and price impact coefficient are integrated into several sets of training data and test data. The training data is used to train the artificial intelligence model, and the test data is used to test the trained artificial intelligence model. The artificial intelligence model is adjusted according to the test results. Finally, a price influence quantification model is obtained with the predicted user electricity consumption as input and the price impact coefficient as output. The artificial intelligence model is a nonlinear regression model based on neural networks, which is implemented using a BP neural network structure and / or an RBF neural network structure.

[0013] In conjunction with the first aspect above, in one possible implementation, a price influence coefficient is generated based on the user's electricity consumption forecast and the price influence quantification model, including: By inputting the current user electricity consumption forecast of the aggregated entity into the price influence quantification model, a price influence coefficient that can quantify the current aggregated entity's impact on electricity prices is obtained.

[0014] In conjunction with the first aspect above, one possible implementation involves determining the fluctuating electricity price based on the price impact coefficient and the grid base price, including: A1: Multiply the price impact coefficient by the grid base price to obtain the preliminary price, and determine whether the preliminary price is greater than the maximum value of the standard price range; if yes, use the maximum value of the standard price range as the fluctuating price of the aggregated entity in the corresponding forecast period; if no, jump to A2; A2: Determine whether the preliminary electricity price is less than the minimum value of the standard electricity price range; if yes, use the minimum value of the standard electricity price range as the fluctuating electricity price of the aggregated entity in the corresponding forecast period; if no, use the preliminary electricity price as the fluctuating electricity price of the aggregated entity in the corresponding forecast period.

[0015] Secondly, an artificial intelligence-based electricity price fluctuation prediction system is provided, including: an aggregation prediction module and an electricity price prediction module; The aggregation prediction module is used to construct aggregation entities and plan prediction periods, and to obtain the predicted user electricity consumption values ​​of the aggregation entities within the prediction periods; wherein, the aggregation entities include enterprises and factory areas; The electricity price prediction module is used to: obtain the grid base price corresponding to the aggregated entity during the prediction period; establish a price influence quantification model that can reflect the relationship between the behavior of the aggregated entity and the electricity price; generate a price influence coefficient based on the user electricity consumption prediction value and the price influence quantification model; and determine the fluctuating electricity price based on the price influence coefficient and the grid base price.

[0016] This application provides a method and system for predicting electricity price fluctuations based on artificial intelligence, with the following advantages: 1. This invention constructs aggregated entities and plans forecast periods to obtain the predicted user electricity consumption of the aggregated entities within the forecast period; obtains the grid base price corresponding to the aggregated entities within the forecast period; establishes a price influence quantification model that reflects the relationship between the behavior of aggregated entities and electricity prices; generates a price influence coefficient based on the predicted user electricity consumption and the price influence quantification model; and determines the fluctuating electricity price based on the price influence coefficient and the grid base price. This invention solves the technical problem of difficulty in assessing the impact of one's own behavior on market electricity prices when forecasting electricity prices. This invention can improve the accuracy of market electricity price forecasting and enhance enterprises' ability to perceive electricity prices.

[0017] 2. This invention constructs a dynamic closed-loop and highly intelligent electricity price formation and prediction mechanism, which fundamentally overcomes the limitations of traditional static pricing or single-model prediction, and realizes accurate prediction and proactive guidance of electricity price fluctuations, thereby bringing progress to the safe and stable operation of the power system, the optimization of economic benefits for market participants, and the improvement of overall social energy efficiency.

[0018] 3. This invention constructs a data-driven and highly flexible dynamic time period division mechanism, abandoning the rigid mode of traditional fixed peak and valley time period division. Through in-depth mining and intelligent analysis of historical electricity price data, it achieves accurate alignment between the predicted time period and the actual fluctuation characteristics of the market, thus laying a solid foundation for the scientificity and accuracy of subsequent electricity price prediction, and bringing significant technical and economic value.

[0019] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0020] Figure 1 A schematic diagram illustrating the steps of an artificial intelligence-based electricity price fluctuation prediction method provided in this application embodiment; Figure 2 A schematic diagram illustrating the steps of planning and forecasting time periods provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of an artificial intelligence-based electricity price fluctuation prediction system provided in an embodiment of this application. Detailed Implementation

[0021] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0022] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0023] like Figure 1 As shown in the embodiment of this application, an artificial intelligence-based method for predicting electricity price fluctuations includes: S1: Construct aggregated entities and plan forecast periods, and obtain the predicted user electricity consumption values ​​of the aggregated entities within the forecast periods; wherein, aggregated entities include enterprises and factories, etc. S2: Obtain the basic electricity price of the grid corresponding to the aggregated entity within the prediction period; S3: Establish a price influence quantification model that can reflect the relationship between aggregate entity behavior and electricity price; generate price influence coefficients based on user electricity consumption forecasts and the price influence quantification model; determine fluctuating electricity prices based on price influence coefficients and grid base electricity prices.

[0024] It is worth noting that the core of the artificial intelligence-based electricity price fluctuation prediction method proposed in this invention lies in the construction of a dynamic, closed-loop, and highly intelligent electricity price formation and prediction mechanism. This fundamentally overcomes the limitations of traditional static pricing or single-model prediction, and achieves accurate prediction and proactive guidance of electricity price fluctuations. This brings revolutionary progress to the safe and stable operation of the power system, the optimization of economic benefits for market participants, and the improvement of overall social energy efficiency. Specifically, this method first defines enterprises and factories as "aggregated entities" and obtains their predicted user electricity consumption values ​​during the forecast period. This allows the forecasting model to delve from the macro-system level to micro-economic units with specific behavioral patterns, greatly enriching the data dimensions and granularity, and laying a solid foundation for accurate analysis. Second, it establishes a "price influence quantification model" that reflects the relationship between aggregated entity behavior and electricity prices. This model no longer simply regards electricity load as a passive result of electricity prices, but uses artificial intelligence technology to deeply mine and quantify the reaction force of aggregated entities as rational economic agents on electricity prices through their production planning, scheduling strategies, cost control, and other behaviors, thereby generating a forward-looking "price influence coefficient." Finally, by combining this coefficient with the grid base price, the determined "fluctuating electricity price" is no longer an isolated technical parameter, but a comprehensive signal that integrates supply and demand fundamentals, market expectations, and the game of participant behavior. The synergistic effect of this series of designs offers several advantages. For grid operators, this method provides a more refined tool for load forecasting and management, enabling early warning of potential supply-demand imbalances and facilitating the development of more flexible and economical dispatch strategies. This effectively ensures grid security and reduces operational risks. For aggregate entities themselves, the model allows them to predict the impact of their behavior on electricity prices and even their own electricity costs, thereby incentivizing them to optimize production scheduling and participate in demand-side response. This transforms them from mere electricity consumers into active market participants and value creators, achieving cost reduction and efficiency improvement. For the entire electricity market, this price signal more accurately and quickly reflects the scarcity of resources, guiding social capital towards efficient energy investment and consumption, promoting the absorption of renewable energy, and driving the construction of a cleaner, more efficient, and resilient new power system. Therefore, this invention is not only a technological breakthrough but also a profound reshaping of the future electricity market structure and business model. Through artificial intelligence, it realizes the transformation of electricity prices from "passive acceptance" to "active shaping," providing core algorithmic support for building a smart energy ecosystem. It possesses extremely high technological application value and broad market prospects.

[0025] It should be noted that when the fluctuating electricity price of the aggregate entity is known during the forecast period, the changes in electricity price caused by the aggregate entity's own production can be understood. This allows for adjustments to production before the forecast period, thereby bringing the electricity cost of production to an ideal state.

[0026] In one possible implementation of this application embodiment, the above-mentioned S1 can be implemented by the following S101, S102, S103 and S104, which are described in detail below: S101: Construct the aggregate entity, including: Enterprises or factories participating in electricity price fluctuation forecasting are marked as target units, and these target units under unified management are manually marked as an aggregated entity; wherein, the unified management includes enterprises or factories belonging to the same company.

[0027] like Figure 2 As shown, S102: Planning and forecasting period, including: When the target time is reached, the historical electricity price of each time point in the region where the aggregated entity is located within the n days prior to the current target time is obtained, and the average value of the historical electricity price at each time point is marked as the reference electricity price at that time point; wherein, the target time and n are obtained manually, the target time is generally 0:01, and n is generally 30; When the reference electricity price is greater than the first electricity price threshold, the corresponding time point is marked as a peak time point; when the reference electricity price is not greater than the first electricity price threshold and not less than the second electricity price threshold, the corresponding time point is marked as a flat time point; when the reference electricity price is less than the second electricity price threshold, the corresponding time point is marked as a low time point. Here, the first and second electricity price thresholds are determined based on the reference electricity price. For example, if the reference electricity prices of the previous n days are sorted from largest to smallest, the first electricity price threshold is the reference electricity price at the 25th percentile, and the second electricity price threshold is the reference electricity price at the 70th percentile. Peak, off-peak, and low-peak time points are integrated to obtain peak, off-peak, and off-peak time domains. Weighting factors are determined based on the average electricity price within each of these time domains. These weighting factors include... Weighting factors and weighting factor ; The dynamic division duration for each time domain is obtained by multiplying the standard division duration by the weighting factor of each time domain and rounding it up; where rounding up means rounding up to the minute; the standard division duration is determined based on the manager's precision in electricity price forecasting. The higher the precision of electricity price forecasting, the smaller the standard division duration, and the lower the precision of electricity price forecasting, the larger the standard division duration. Based on the dynamic division duration of each time domain, the corresponding time domain is divided into several prediction periods.

[0028] It is worth noting that the planning and forecasting method proposed in this step constructs a data-driven, adaptive, and highly flexible dynamic time-segmentation mechanism. It abandons the rigid model of traditional fixed peak-valley time-segmentation and, through in-depth mining and intelligent analysis of historical electricity price data, achieves precise alignment between the forecasted time period and the actual market fluctuation characteristics. This lays a solid foundation for the scientific accuracy of subsequent electricity price forecasts and brings significant technical and economic value. Specifically, this method automatically captures and calculates the historical average electricity price at the same time point in the previous n days as a reference price, ensuring the objectivity and timeliness of the time-segmentation basis. This smooths out short-term random fluctuations and captures stable, cyclical electricity consumption patterns. Furthermore, by automatically determining the electricity price threshold using the dynamic percentile method, it intelligently identifies peak, flat, and low-valley time points. This design gives the system strong adaptability; regardless of whether the overall market electricity price level rises or falls, the time-segmentation always accurately reflects the current relative price distribution, avoiding the problem of manually set fixed thresholds becoming ineffective due to market changes and ensuring the continuous effectiveness of the segmentation results. On this basis, this method does not simply convert discrete time periods into fixed peak-valley periods. Instead of simply merging time points, this design innovatively introduces the concepts of "weighting factor" and "dynamic partitioning duration." The weight of each time period is determined by calculating the average electricity price, and then combined with a "standard partitioning duration" set by the administrator based on the required forecast granularity. This ultimately generates a "dynamic partitioning duration" for each time period. The brilliance of this design lies in its intelligent adjustment of the granularity of time period partitioning. Peak periods with volatile electricity prices and stronger value signals are automatically assigned smaller dynamic partitioning durations, resulting in more granular segmentation. This allows the forecasting model to capture more key fluctuation details. Conversely, during off-peak periods with relatively stable electricity prices, the partitioning granularity is widened, avoiding waste of computational resources and achieving an optimal balance between forecast accuracy and computational efficiency. Therefore, this step not only enables the division of forecast periods to adapt to market changes, ensuring high quality and high relevance of the input data for the electricity price forecasting model, but also, by transforming the management objective of forecast precision into quantifiable parameters, endows the entire system with excellent configurability and scalability, meeting the application needs of different levels and scenarios. Ultimately, this dynamic, intelligent, and precise time-segment planning method enables the subsequently generated fluctuating electricity price signals to have higher resolution and stronger guiding significance in the time dimension, accurately reflecting the true value of electricity resources at different times. This more effectively incentivizes aggregated entities to adjust their electricity consumption behavior, guides load peak shaving and valley filling, and improves the economy and stability of grid operation, providing key technical support for building a more intelligent, efficient, and market-oriented integrated power ecosystem.

[0029] It should be noted that the peak time point, average time point, and trough time point are integrated to obtain the peak time domain, trough time domain, and average time domain. The specific steps are as follows: Peak time points are integrated into a peak time domain, and average time points are integrated into an average time domain, while low time points are integrated into a low time domain. If a time domain with a duration shorter than a length threshold exists within a peak time domain, a trough time domain, or a flat time domain, that time domain will be merged into an adjacent time domain. The merging rules are as follows: if the time domains before and after the time domain are the same type of time domain, then the time domain will be randomly merged into either the preceding or following time domain, connecting the preceding and following time domains into one; if the time domains before and after the time domain are different time domains, then the flat time domain will be merged first, and if there is no flat time domain, then the trough time domain will be merged first. The length threshold is set manually and can be 30 minutes.

[0030] It should be noted that, based on the dynamic division duration of each time domain, the corresponding time domain is divided into several prediction periods. Specifically, the time domain is divided sequentially from front to back based on the size of the dynamic division duration. If the last part of the time after division has a length shorter than the dynamic division duration, it is further determined whether the length of the last part of the time is less than half of the dynamic division duration. If so, the last part of the time is incorporated into the previous prediction period; otherwise, the last part of the time is planned as a prediction period.

[0031] S103: Determine the weighting factors for the peak, off-peak, and flat time domains based on the average electricity price within the peak, off-peak, and flat time domains, including: Obtain the average electricity price during peak time. Average electricity price during off-peak hours Average electricity price over a flat period of time Based on average electricity price Average electricity price and average electricity price The weight partitioning factor for the corresponding time domain is determined by formula (1). ;in, For the time domain numbering, when The time period represents the peak time range, when The time period represents the trough time domain, when The time period represents the average time domain; The calculation formula (1) is: .

[0032] It should be noted that when hour As the weighting factor for the peak time domain, when hour As the weighting factor for the trough time domain, when hour This is the weighting factor for the flat time domain.

[0033] S104: Obtain the predicted user electricity consumption value of the aggregation entity within the prediction period, including: Extract the product types and production quantities of each type planned by the aggregated entity in each forecast period. Obtain the electricity consumption of each product type produced by the aggregated entity in the n days prior to the current time. Multiply the electricity consumption of each product type by the corresponding production quantity to determine the total electricity consumption of each type of product in each forecast period. Add the total electricity consumption of different types of products in the same forecast period to obtain the predicted user electricity consumption value in the forecast period.

[0034] It should be noted that the specific steps for obtaining the electricity consumption of individual products of each type of product produced by the aggregated entity in the n days prior to the current time are as follows: obtain the electricity consumption of several individual products of type one produced by the aggregated entity in the n days prior to the current time, and mark the average value of the electricity consumption as the electricity consumption of individual products.

[0035] In one possible implementation of this application embodiment, the above-mentioned S2 can be implemented by the following S201, which will be described in detail below: S201: Obtain the grid base electricity price corresponding to the aggregated entity within the forecast period, including: Obtain the predicted user electricity consumption values ​​of other aggregated entities within the same area as the aggregated entity during the prediction period, and add several of these predicted user electricity consumption values ​​to obtain a reference electricity consumption prediction value; obtain the electricity consumption in the historical data of the area within the same duration as the prediction period, where the deviation from the reference electricity consumption prediction value does not exceed a deviation threshold, and integrate the electricity prices corresponding to these electricity consumption values ​​into an electricity price group; wherein, the deviation threshold is set based on the magnitude of the reference electricity consumption prediction value, and can be set to 3% of the reference electricity consumption prediction value; Obtain the variance of the electricity price group and determine whether the variance is greater than the electricity price variance threshold. If yes, obtain the average electricity price in the electricity price group, remove the electricity price with the largest absolute value of the difference from the average electricity price, re-calculate the variance of the electricity price group and re-determine the variance until the variance of the electricity price group is not greater than the electricity price variance threshold. Then, calculate the average value of the electricity prices retained in the electricity price group to obtain the grid base price. If no, calculate the average value of the electricity prices retained in the electricity price group to obtain the grid base price. The electricity price variance threshold is obtained through empirical setting.

[0036] It is worth noting that this step constructs a highly valuable benchmark electricity price generation mechanism. Through in-depth mining and intelligent filtering of historical big data, it can remove random disturbances and extreme events in the market, accurately restore the "pure market" electricity price level that is highly matched with the current prediction scenario, and thus provide an extremely solid and fair benchmark for the accurate calculation of subsequent fluctuating electricity prices, significantly improving the scientificity, stability and reliability of the entire prediction method. This method first breaks through the limitations of using the published electricity price in a single, static way. Instead, it innovatively aggregates the electricity consumption forecasts of other aggregated entities within the region to form a "reference electricity consumption forecast" that can macroscopically reflect the trend of aggregate demand in the region. Using this as an anchor, it performs context matching in a vast amount of historical data to find the most similar historical electricity consumption scenarios. This similarity-based retrieval logic ensures that the selected historical electricity price samples are highly correlated and comparable with the current market fundamentals. Furthermore, by setting a deviation threshold that adapts to the magnitude of the reference electricity consumption forecast, the matching process is dynamically quantified. This ensures both the sufficiency of the sample library and avoids the introduction of invalid samples with excessively large discrepancies, thus guaranteeing the effectiveness of the base electricity price from the source. The most crucial breakthrough lies in the introduction of an iterative data cleaning and purification mechanism. This mechanism calculates the variance of the sample "electricity price group" and compares it with a "price variance threshold," actively identifying and eliminating extreme outliers in the sample. These outliers include price spikes or troughs caused by sudden failures, market manipulation, or irrational speculation. This iterative process continues until the price fluctuations of the remaining samples converge to a reasonable range, at which point the base grid price is finally calculated using the average. This design acts like an intelligent "signal filter," effectively filtering out "noise" from historical data and retaining "effective signals" reflecting the fundamental supply and demand relationship. This ensures that the final base grid price is not simply an arithmetic average of all historical data, but a fair price that has undergone rigorous quality control and represents a normal market condition under similar load levels. Therefore, this step greatly enhances the stability and anti-interference capability of the benchmark electricity price, avoiding benchmark distortion caused by a few extreme historical data points, providing a clean input for subsequent price influence quantification models, and ensuring the robustness of the prediction results. Secondly, this dynamic benchmark generation method based on scenario matching and outlier removal enables the grid base price to better adapt to dynamic market changes, always maintaining a level that best reflects the current actual supply and demand relationship, thus improving the accuracy of prediction. Finally, a fair and stable grid base price is the cornerstone of the entire electricity price prediction model. It ensures that the final generated fluctuating electricity price can more realistically reflect the marginal impact of aggregate entity behavior, thereby providing a more reliable decision-making basis for electricity market participants, providing key technical support for the refined scheduling and market-oriented operation of the power grid, and powerfully promoting the overall improvement of power resource allocation efficiency.

[0037] It should be noted that the region where the aggregated entity is located is a region where electricity prices can be managed uniformly.

[0038] It should be noted that the method of obtaining the electricity consumption in the historical data of the region that deviates from the reference electricity consumption forecast value by no more than the deviation threshold within the same time period as the forecast period is to traverse the historical time based on the length of the forecast period and determine several time lengths that can be compared and analyzed.

[0039] It should be noted that, among the electricity prices in the aforementioned price group with the largest absolute value of the difference from the average electricity price, if the electricity price in the price group with the largest absolute value of the difference from the average electricity price has both a maximum price and a minimum price, then the minimum price will be removed first.

[0040] It should be noted that if, after removing 90% of the electricity price, the variance of the remaining electricity price is still not less than the defined value for electricity price variance, then the average value of the original electricity price of the electricity price group will be used as the basic electricity price of the power grid.

[0041] In one possible implementation of the embodiments of this application, the above-mentioned S3 can be implemented by the following S301, S302 and S303, which are described in detail below: S301: Establish a quantitative model of price influence that can reflect the relationship between aggregated entity behavior and electricity prices, including: Extract the user electricity consumption forecast and price impact coefficient of each aggregated entity from the historical reference data; where the historical reference data includes the user electricity consumption forecast and the price impact coefficient set by experts based on the user electricity consumption forecast and the corresponding electricity price; The predicted user electricity consumption and price impact coefficient are integrated into several sets of training data and test data. The training data is used to train the artificial intelligence model, and the test data is used to test the trained artificial intelligence model. The artificial intelligence model is adjusted according to the test results. Finally, a price influence quantification model is obtained with the predicted user electricity consumption as input and the price impact coefficient as output. The artificial intelligence model is a nonlinear regression model based on neural networks, which is implemented using a BP neural network structure and / or an RBF neural network structure.

[0042] Specifically, the steps for testing the trained AI model using validation data and adjusting the AI ​​model based on the validation results are as follows: The predicted user electricity consumption from the test data is input into the trained AI model to obtain the corresponding price impact coefficient. The price impact coefficient is compared with the corresponding price impact coefficient in the test data. If the difference between the two is within a threshold (obtained empirically), no parameter adjustment is needed, and the next set of test data is tested. If it is not within the threshold, the corresponding parameters are adjusted until the price impact coefficient of the corresponding test data is within the threshold, and then the next set of test data is tested. When the number of test data where the difference in price impact coefficients obtained from all test data is within the threshold accounts for 90% or more of the total test data, a prediction model with user electricity consumption prediction as input and price impact coefficient as output is obtained.

[0043] S302: Generate price influence coefficients based on user electricity consumption forecasts and the aforementioned price influence quantification model, including: By inputting the current user electricity consumption forecast of the aggregated entity into the price influence quantification model, a price influence coefficient that can quantify the current aggregated entity's impact on electricity prices is obtained.

[0044] S303: Determining fluctuating electricity prices based on price impact coefficients and grid base electricity prices, including: A1: Multiply the price impact coefficient by the grid base price to obtain the preliminary price, and determine whether the preliminary price is greater than the maximum value of the standard price range; if yes, use the maximum value of the standard price range as the fluctuating price of the aggregated entity in the corresponding forecast period; if no, jump to A2; A2: Determine whether the preliminary electricity price is less than the minimum value of the standard electricity price range; if yes, use the minimum value of the standard electricity price range as the fluctuating electricity price of the aggregated entity in the corresponding forecast period; if no, use the preliminary electricity price as the fluctuating electricity price of the aggregated entity in the corresponding forecast period.

[0045] The foregoing primarily describes the solutions of the embodiments of this application from the perspective of device implementation. It is understood that each device, for example, an artificial intelligence-based electricity price fluctuation prediction system, includes at least one of the hardware structures and software modules corresponding to the execution of each function in order to achieve the above-mentioned functions. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in a hardware-driven or software-driven manner depends on the specific application and design constraints of the technical solution. Those skilled in the art 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.

[0046] This application embodiment can divide an artificial intelligence-based electricity price fluctuation prediction system into functional units based on the above method example. For example, each function can be divided into separate functional units, or two or more functions can be integrated into one electricity price prediction module. The integrated units can be implemented in hardware or as software functional units. It should be noted that the unit division in this application embodiment is illustrative and only represents a logical functional division; other division methods may be used in actual implementation.

[0047] When using integrated units, Figure 3 The diagram shows a possible structure of an artificial intelligence-based electricity price fluctuation prediction system (referred to as prediction system 30) involved in the above embodiments. The prediction system 30 includes an electricity price prediction module 301 and an aggregate prediction module 302, and may also include a storage medium (referred to as storage unit 303). Figure 3 The schematic diagram shown can be used to illustrate the structure of an artificial intelligence-based electricity price fluctuation prediction system involved in the above embodiments.

[0048] when Figure 3 The schematic diagram shown illustrates the structure of an artificial intelligence-based electricity price fluctuation prediction system involved in the above embodiments. The electricity price prediction module 301 is used to control and manage the operation of the artificial intelligence-based electricity price fluctuation prediction system, the aggregation prediction module 302 is used to control and manage the time period and aggregation entities of the artificial intelligence-based electricity price fluctuation prediction system, and the storage unit 303 is used to store the program code and data of the artificial intelligence-based electricity price fluctuation prediction system.

[0049] For example, the aggregation prediction module 302 is used to construct aggregation entities and plan prediction periods, and obtain the predicted user electricity consumption values ​​of the aggregation entities within the prediction periods; wherein, the aggregation entities include enterprises and factories; Electricity price forecasting module 301: used to obtain the grid base electricity price corresponding to the aggregated entity during the forecast period; establish a price influence quantification model that can reflect the relationship between the behavior of the aggregated entity and the electricity price; generate a price influence coefficient based on the user electricity consumption forecast value and the price influence quantification model; and determine the fluctuating electricity price based on the price influence coefficient and the grid base electricity price.

[0050] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0051] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely illustrative descriptions of the application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

[0052] Some of the data in the above calculation formula are obtained by removing dimensions and taking their numerical values. The calculation formula is a calculation formula that is closest to the real situation, obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the calculation formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

Claims

1. A method for predicting price fluctuation based on artificial intelligence, characterized by, include: S1: Construct aggregated entities and plan forecast periods, and obtain the predicted user electricity consumption values ​​of the aggregated entities within the forecast periods; wherein, aggregated entities include enterprises and factory areas; S2: Obtain the basic electricity price of the grid corresponding to the aggregated entity within the prediction period; S3: Establish a price influence quantification model that can reflect the relationship between aggregate entity behavior and electricity price, and generate price influence coefficients based on user electricity consumption forecasts and the price influence quantification model; Fluctuating electricity prices are determined based on price impact coefficients and the grid base price.

2. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The planning and forecasting period includes: When the target time is reached, obtain the historical electricity price of each time point in the region where the aggregated entity is located within the n days prior to the current target time, and mark the average of the historical electricity price of the time point as the reference electricity price of the time point. When the reference electricity price is greater than the first electricity price threshold, the corresponding time point is marked as a peak time point; when the reference electricity price is not greater than the first electricity price threshold and not less than the second electricity price threshold, the corresponding time point is marked as a flat time point; when the reference electricity price is less than the second electricity price threshold, the corresponding time point is marked as a low time point; wherein, the first and second electricity price thresholds are determined based on the reference electricity price. Peak, off-peak, and low-peak time points are integrated to obtain peak, off-peak, and off-peak time domains. Weighting factors are determined based on the average electricity price within each of these time domains. These weighting factors include... Weighting factors and weighting factor ; The dynamic division duration for each time domain is obtained by multiplying the standard division duration by the weighting factor of each time domain and rounding up; where rounding up means rounding up to the minute; the standard division duration is determined based on the manager's precision in electricity price forecasting; Based on the dynamic division duration of each time domain, the corresponding time domain is divided into several prediction periods.

3. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 2, characterized in that, The method for determining the weighting factors corresponding to the peak, off-peak, and flat time domains based on the average electricity price within the peak, off-peak, and flat time domains includes: Obtain the average electricity price during peak time. Average electricity price during off-peak hours Average electricity price over a flat period of time Based on average electricity price Average electricity price and average electricity price The weight partitioning factor for the corresponding time domain is determined by formula (1). ;in, For the time domain numbering, when The time period represents the peak time range, when The time period represents the trough time domain, when The time period represents the average time domain; The calculation formula (1) is: 。 4. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The step of obtaining the predicted user electricity consumption value of the aggregated entity within the prediction period includes: Extract the product types and production quantities of each type planned by the aggregated entity in each forecast period. Obtain the electricity consumption of each product type produced by the aggregated entity in the n days prior to the current time. Multiply the electricity consumption of each product type by the corresponding production quantity to determine the total electricity consumption of each type of product in each forecast period. Add the total electricity consumption of different types of products in the same forecast period to obtain the predicted user electricity consumption value in the forecast period.

5. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The acquisition of the grid base electricity price corresponding to the aggregated entity within the prediction period includes: Obtain the user electricity consumption forecast values ​​of other aggregated entities within the area where the aggregated entity is located during the forecast period, and add several user electricity consumption forecast values ​​to obtain a reference electricity consumption forecast value; obtain the electricity consumption in the historical data of the area within the same duration as the forecast period, where the deviation from the reference electricity consumption forecast value does not exceed a deviation threshold, and integrate the electricity prices corresponding to the electricity consumption into an electricity price group; wherein, the deviation threshold is set according to the magnitude of the reference electricity consumption forecast value; Obtain the variance of the electricity price group and determine whether the variance is greater than the electricity price variance threshold. If yes, obtain the average electricity price in the electricity price group, remove the electricity price with the largest absolute value of the difference from the average electricity price in the electricity price group, re-calculate the variance of the electricity price group and re-determine the variance until the variance of the electricity price group is not greater than the electricity price variance threshold. Then, calculate the average value of the electricity prices retained in the electricity price group to obtain the grid base price. If no, calculate the average value of the electricity prices retained in the electricity price group to obtain the grid base price.

6. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The establishment of a price influence quantification model that reflects the relationship between aggregate entity behavior and electricity prices includes: Extract the user electricity consumption forecast and price impact coefficient of each aggregated entity from the historical reference data; where the historical reference data includes the user electricity consumption forecast and the price impact coefficient set by experts based on the user electricity consumption forecast and the corresponding electricity price; The predicted user electricity consumption and price impact coefficient are integrated into several sets of training data and test data. The training data is used to train the artificial intelligence model, and the test data is used to test the trained artificial intelligence model. The artificial intelligence model is adjusted according to the test results. Finally, a price influence quantification model is obtained with the predicted user electricity consumption as input and the price impact coefficient as output. The artificial intelligence model is a nonlinear regression model based on neural networks, which is implemented using a BP neural network structure and / or an RBF neural network structure.

7. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The generation of price influence coefficients based on user electricity consumption forecasts and the price influence quantification model includes: By inputting the current user electricity consumption forecast of the aggregated entity into the price influence quantification model, a price influence coefficient that can quantify the current aggregated entity's impact on electricity prices is obtained.

8. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The determination of fluctuating electricity prices based on price impact coefficients and the grid base price includes: A1: Multiply the price impact coefficient by the grid base price to obtain the preliminary price, and determine whether the preliminary price is greater than the maximum value of the standard price range; if yes, use the maximum value of the standard price range as the fluctuating price of the aggregated entity in the corresponding forecast period; if no, jump to A2; A2: Determine whether the preliminary electricity price is less than the minimum value of the standard electricity price range; if yes, use the minimum value of the standard electricity price range as the fluctuating electricity price of the aggregated entity in the corresponding forecast period; if no, use the preliminary electricity price as the fluctuating electricity price of the aggregated entity in the corresponding forecast period.

9. The method for predicting electricity price fluctuations based on artificial intelligence according to claim 1, characterized in that, The construction of the aggregate entity includes: Enterprises or factories participating in electricity price fluctuation forecasting are marked as target units, and these target units under unified management are manually marked as an aggregated entity; wherein, the unified management includes enterprises or factories belonging to the same company.

10. An artificial intelligence-based electricity price fluctuation prediction system, used to run the artificial intelligence-based electricity price fluctuation prediction method according to any one of claims 1 to 9, characterized in that, include: Aggregated forecasting module and electricity price forecasting module; The aggregation prediction module: constructs aggregation entities and plans prediction periods, and obtains the predicted user electricity consumption values ​​of the aggregation entities within the prediction periods; wherein, the aggregation entities include enterprises and factory areas; The electricity price prediction module: obtains the basic electricity price of the power grid corresponding to the aggregated entity during the prediction period; establishes a price influence quantification model that can reflect the relationship between the behavior of the aggregated entity and the electricity price; generates a price influence coefficient based on the user's electricity consumption prediction value and the price influence quantification model; and determines the fluctuating electricity price based on the price influence coefficient and the basic electricity price of the power grid.

Images