Market boundary adaptive correction method and related device

Abstract

The application discloses a market boundary self-adaptive correction method and a related device, and relates to the technical field of new energy sources; the method comprises the following steps: acquiring a new energy available output original prediction curve of a plurality of time periods in the future of new energy sources and an operation parameter of a current time; identifying a current market scene according to the operation parameter; calculating a prediction deviation between a new energy actual output and a preset target output, performing weighted processing on the prediction deviation according to a deviation direction of the current market scene and the prediction deviation, and obtaining a scene weighted error; updating a prediction correction confidence of the current time according to the scene weighted error, and calculating a correction coefficient according to the updated prediction correction confidence; and performing multi-step rolling correction on the prediction values of each future time period in the original prediction curve through the correction coefficient. The application solves the problems that the existing correction scheme is difficult to adaptively adjust the correction strategy according to the market operation risk, and lacks quantitative evaluation and risk coupling mechanism for the credibility of the correction result.

Description

Technical Field

[0001] This invention relates to the field of new energy technology, and in particular to a market boundary adaptive correction method and related apparatus. Background Technology

[0002] With the continuous increase in installed capacity of new energy sources such as wind power and photovoltaics, regional power systems are shifting from a traditional "controllable power source-dominated" operation mode to a "high proportion of new energy integration" operation mode. New energy output is affected by factors such as weather, season, terrain, and the status of power plant equipment, exhibiting significant randomness, volatility, and uncertainty. Its forecasting results directly impact multiple aspects, including day-ahead power generation planning, intraday rolling adjustments, real-time market clearing boundary calculations, and deviation assessments. Currently, many regions have initiated the construction of electricity spot markets. With the continuous increase in new energy power plants and their increasing market participation, market-based new energy deviations have become the most significant source of deviation in the spot system and the actual power grid, seriously affecting grid security and market fairness.

[0003] Currently, there are two main methods for correcting deviations in new energy forecasts. First, operators manually and subjectively correct the deviation between actual output and the forecast curve. This method relies heavily on the operator's personal experience, and different personnel may provide different correction results under the same deviation conditions. This leads to inconsistent correction standards and poor uniformity, and makes it difficult to respond in real-time to rapidly changing market conditions. Second, fixed parameters are used for correction, such as setting a fixed smoothing coefficient or correction ratio and adjusting all time periods according to the same rules. This method is simple to implement, but the fixed rules prevent dynamic adjustment of the correction strategy based on changes in actual operating conditions.

[0004] However, the above-mentioned solutions have at least the following problems: First, manual correction is highly subjective, with inconsistent correction standards, making it difficult to respond in real time to changes in market conditions. Second, fixed-parameter correction methods cannot adaptively adjust to market risk factors such as real-time electricity price fluctuations, changes in deviation assessment unit prices, and changes in system reserve margins, resulting in a mismatch between the correction effect and actual market risk requirements. Furthermore, existing correction methods lack a quantitative assessment mechanism for the reliability of correction results and are not effectively coupled with spot market clearing risks. Corrections may continue even in cases of data anomalies or communication interruptions, posing a security risk of introducing abnormal data into the market clearing process. In summary, existing correction schemes are unable to adaptively adjust correction strategies according to market operational risks and lack a quantitative assessment mechanism for the reliability of correction results and a risk coupling mechanism. Summary of the Invention

[0005] This invention provides a market boundary adaptive correction method and related apparatus to solve the problems that existing correction schemes are difficult to adaptively adjust correction strategies according to market operation risks, and lack quantitative assessment of the credibility of correction results and risk coupling mechanism.

[0006] In view of this, a first aspect of the present invention provides a market boundary adaptive correction method, the method comprising:

[0007] Obtain the original forecast curves of available power output from new energy sources for multiple future time periods, as well as the operating parameters at the current moment;

[0008] The current market scenario is identified based on the operating parameters, and the current market scenario includes normal scenario, high-risk scenario and abnormal scenario;

[0009] Calculate the prediction deviation between the actual output of new energy and the preset target output, and weight the prediction deviation according to the current market scenario and the direction of the deviation to obtain the scenario-weighted error;

[0010] The prediction correction confidence level at the current moment is updated based on the scenario weighted error, and the correction coefficient is calculated based on the updated prediction correction confidence level.

[0011] The original forecast curve for available renewable energy output is modified using the correction coefficient in multiple rolling steps for each future time period, and a modified forecast curve for available renewable energy output is constructed based on the modified forecast values.

[0012] Optionally, the step of identifying the current market scenario based on the operating parameters includes normal scenarios, high-risk scenarios, and abnormal scenarios, including:

[0013] A market scenario feature vector is constructed based on the operating parameters. The market scenario feature vector includes real-time electricity price, deviation assessment unit price, reserve ratio, and renewable energy volatility.

[0014] Determine whether the data quality of the operating parameters is normal;

[0015] When the data quality is abnormal, the current market scenario is determined to be an abnormal scenario;

[0016] When the data quality is normal, the real-time electricity price, the deviation assessment unit price, the reserve ratio, and the new energy volatility are compared with their respective thresholds, and the current market scenario is determined to be a normal scenario or a high-risk scenario based on the comparison results.

[0017] Optionally, the step of calculating the prediction deviation between the actual output of new energy and the preset target output, and weighting the prediction deviation according to the current market scenario and the direction of the deviation, to obtain a scenario-weighted error, includes:

[0018] Obtain the preset target output and calculate the difference between the actual output of the new energy source at the current moment and the target output to obtain the prediction deviation;

[0019] When the current market scenario is the normal scenario, the prediction deviation is used as the scenario-weighted error;

[0020] When the current market scenario is the high-risk scenario and the prediction deviation is negative, the prediction deviation is amplified with a first weight to obtain the scenario-weighted error;

[0021] When the current market scenario is the high-risk scenario and the prediction deviation is non-negative, the prediction deviation is reduced by the second weight to obtain the scenario-weighted error, where the first weight is greater than the second weight.

[0022] Optionally, updating the prediction correction confidence at the current moment based on the scenario-weighted error includes:

[0023] The adaptive smoothing coefficient is determined based on the current market scenario, and the normalized error contribution is calculated based on the scenario-weighted error.

[0024] Obtain the prediction correction confidence score from the previous time step, and recursively update it based on the adaptive smoothing coefficient, the normalization error contribution, and the prediction correction confidence score from the previous time step to obtain the prediction correction confidence score from the current time step.

[0025] Optionally, calculating the correction coefficient based on the updated prediction correction confidence includes:

[0026] The ideal correction coefficient is determined based on the updated prediction correction confidence; the actual reserve ratio of the region at the current time is obtained, and the upper limit of the correction coefficient is determined based on the actual reserve ratio of the region; the ideal correction coefficient and the upper limit of the correction coefficient are compared, and the smaller value of the two is taken as the final correction coefficient.

[0027] Optionally, the step of performing multi-step rolling correction on the predicted values ​​for each future period in the original prediction curve of available renewable energy output using the correction coefficient includes:

[0028] A correction direction factor is generated based on the updated prediction correction confidence.

[0029] The time decay factor corresponding to each future time period is determined based on the prediction step size corresponding to each future time period, and the time decay factor decreases as the prediction step size increases;

[0030] For each future time period, the original predicted value of the future time period is corrected according to the correction coefficient, the time decay factor corresponding to the future time period, and the correction direction factor, to obtain the corrected predicted value of the available power output of new energy for the future time period.

[0031] Optionally, the calculation expression for the revised predicted output of new energy sources for the future time period is as follows:

[0032] ;

[0033] ;

[0034] In the formula, For the future The revised forecast of available renewable energy output for each time period For the future Time decay factor for each time period, The correction coefficient is... For the future The original predicted values ​​for each time period, To correct the direction factor, This represents the region's installed capacity of new energy sources.

[0035] A second aspect of the present invention provides a market boundary adaptive correction system, the system comprising:

[0036] The acquisition unit is used to acquire the original forecast curves of the available power output of new energy sources for multiple future time periods, as well as the operating parameters at the current moment.

[0037] The identification unit is used to identify the current market scenario based on the operating parameters, wherein the current market scenario includes a normal scenario, a high-risk scenario, and an abnormal scenario.

[0038] The calculation unit is used to calculate the prediction deviation between the actual output of new energy and the preset target output, and to perform weighted processing on the prediction deviation according to the current market scenario and the direction of the deviation to obtain the scenario-weighted error.

[0039] The update unit is used to update the prediction correction confidence at the current moment according to the scenario weighted error, and to calculate the correction coefficient according to the updated prediction correction confidence.

[0040] The correction unit is used to perform multi-step rolling corrections on the predicted values ​​of each future time period in the original prediction curve of available renewable energy output using the correction coefficient, and to construct a corrected prediction curve of available renewable energy output based on the corrected predicted values.

[0041] A third aspect of the present invention provides a market boundary adaptive correction device, the device comprising a processor and a memory:

[0042] The memory is used to store program code and transmit the program code to the processor;

[0043] The processor is configured to execute the steps of the market boundary adaptive correction method as described in the first aspect above, according to the instructions in the program code.

[0044] A fourth aspect of the present invention provides a computer-readable storage medium for storing program code for executing the market boundary adaptive correction method described in the first aspect above.

[0045] As can be seen from the above technical solutions, the present invention has the following advantages:

[0046] This invention provides a market boundary adaptive correction method. First, it acquires the original predicted power output curves of renewable energy for multiple future time periods and the current operating parameters. Based on these operating parameters, it identifies the current market scenario, achieving dynamic coupling between the correction strategy and market operating risks. This solves the problem that existing fixed-parameter correction methods cannot adaptively adjust to market risk factors. Then, it calculates the prediction deviation between the actual renewable energy output and the preset target output, and applies differentiated weighting to this deviation based on the current market scenario and the direction of the deviation. This allows the system to automatically improve its sensitivity to key deviations in high-risk scenarios and maintain stability in normal scenarios. The invention achieves a stable response; then, it recursively updates the prediction correction confidence level through scenario-weighted error, and calculates the correction coefficient based on the confidence level, establishing a quantifiable and interpretable automatic correction rule, thus solving the problem that existing correction methods lack a mechanism for quantitatively evaluating the reliability of correction results. Finally, it performs multi-step rolling corrections on the predicted values ​​for each future time period using the correction coefficients and constructs a corrected prediction curve, enabling near-term predictions to respond more quickly to the latest operational deviations and far-term predictions to remain relatively stable. Simultaneously, when operational parameters are abnormal, it identifies an abnormal scenario and automatically stops correction, solving the safety risk problem of existing solutions continuing corrections when data is abnormal and introducing abnormal data into subsequent processes. This invention achieves end-to-end adaptive correction from scenario identification, deviation weighting, confidence quantification to rolling correction, reducing human experience intervention and improving the consistency and automation level of correction results. In summary, this invention solves the problems of existing correction schemes that are difficult to adaptively adjust correction strategies according to market operational risks and lack a mechanism for quantitatively evaluating the reliability of correction results and coupling with risk. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 A flowchart illustrating a market boundary adaptive correction method provided in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of a market boundary adaptive correction system provided in an embodiment of the present invention. Detailed Implementation

[0050] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0051] Example 1:

[0052] Please see Figure 1 An adaptive market boundary correction method provided in this embodiment of the invention includes:

[0053] Step 101: Obtain the original forecast curves of available power output of new energy sources for multiple future time periods and the operating parameters at the current moment;

[0054] It should be noted that, firstly, the raw forecast curves of available power output from new energy sources for multiple future time periods are obtained. New energy sources include wind power, photovoltaic power, and other renewable energy sources. The raw forecast curves of available power output from new energy sources are denoted as follows: Where t is the current time, i is the prediction step size (i=1,2,…,N), and N is the total number of prediction steps. In this embodiment, the time granularity is 15 minutes, and the prediction step size N is 16, which covers the next 4 hours. The original prediction value P for each time period is... forecast (t+i) can be read directly.

[0055] Then, the operating parameters at the current moment are obtained. To ensure the clarity and feasibility of the technical solution, the operating parameters in a preferred embodiment are given below, as shown in Table 1. In practical applications, each parameter can be adjusted according to the regional installed capacity, market rules, and operating characteristics.

[0056] Table 1

[0057]

[0058] Step 102: Identify the current market scenario based on the operating parameters. The current market scenario includes normal scenarios, high-risk scenarios, and abnormal scenarios.

[0059] It should be noted that the completeness and rationality of operating parameters are first assessed. This includes checking for communication interruptions, missing data, and whether actual output exceeds reasonable limits. If any of these anomalies occur, the current data is considered unreliable. Continuing the correction process in this case could introduce abnormal data, thus it is identified as an abnormal scenario and correction is stopped. Assuming normal data quality, the market operation risk level is further evaluated. For example, whether the real-time electricity price is too high, the deviation assessment unit price is too high, or the reserve ratio is insufficient. When all these risk indicators are within safe limits, it is identified as a normal scenario, and the correction logic maintains a normal response. However, when any risk indicator exceeds a threshold, it is identified as a high-risk scenario, and the correction logic needs to increase its sensitivity to deviations to cope with market risks. This scenario classification allows the correction strategy to adaptively adjust according to data status and market risks, proactively protecting against data anomalies, operating smoothly under low risk, and strengthening the response under high risk, thereby balancing the effectiveness of correction and system security.

[0060] Step 103: Calculate the prediction deviation between the actual output of new energy and the preset target output, and perform weighted processing on the prediction deviation according to the current market scenario and the direction of the prediction deviation to obtain the scenario-weighted error;

[0061] It should be noted that, firstly, the difference between the actual output of new energy and the preset target output is calculated to obtain the prediction deviation. This difference is used to quantify the degree of deviation between the actual operating state of new energy and the expected target at the current moment, and serves as the basis for subsequent corrections. Then, based on the current market scenario identified in step 102 and the direction of the prediction deviation, the prediction deviation is subjected to differentiated weighting processing to obtain the scenario-weighted error. For example, in normal scenarios, the deviation remains at its original value; in high-risk scenarios, if the deviation direction is negative, it is appropriately amplified; if the deviation direction is non-negative, it is appropriately reduced; and in abnormal scenarios, the current deviation is not used for correction. The above differentiated weighting processing enables the correction strategy to adaptively adjust the response sensitivity according to the market risk level and the direction of deviation, avoiding over-correction or under-correction caused by using a uniform correction intensity in different risk scenarios. The scenario-weighted error will be used as the input for subsequent confidence updates, allowing the cumulative effect of historical errors to be quantified and transmitted to the calculation of the correction coefficient.

[0062] Step 104: Update the prediction correction confidence at the current moment based on the scenario weighted error, and calculate the correction coefficient based on the updated prediction correction confidence.

[0063] It should be noted that, firstly, based on the scenario-weighted error obtained in step 103, the prediction correction confidence level of the previous moment is recursively updated to obtain the prediction correction confidence level of the current moment. The prediction correction confidence level is used to quantify the credibility and correction direction of the current prediction correction result. Its update method employs an adaptive smoothing mechanism, assigning different weights to the current error and historical information under different market scenarios. For example, in high-risk scenarios, the weight of the current error is increased to make the confidence level respond more quickly to the latest deviation changes, while in normal scenarios, a lower weight is maintained to filter out short-term fluctuations. Then, a correction coefficient is calculated based on the updated prediction correction confidence level. This correction coefficient is used to control the magnitude of subsequent prediction corrections; a higher confidence level results in a larger correction magnitude, while a lower confidence level results in a smaller correction magnitude or stops correction. The above confidence level update and correction coefficient calculation enable the correction strategy to adaptively adjust the correction intensity based on the cumulative effect of historical deviations, avoiding over-correction caused by a single deviation fluctuation. Simultaneously, the confidence level is quantitatively assessed to evaluate the credibility of the correction result, providing a quantifiable basis for subsequent correction steps.

[0064] Step 105: Perform multi-step rolling correction on the predicted values ​​of each future time period in the original prediction curve of available new energy output using the correction coefficient, and construct the corrected prediction curve of available new energy output based on the corrected predicted values.

[0065] It should be noted that the correction coefficients calculated in step 104 are used to correct the predicted values ​​for each future period in the original forecast curve of available renewable energy output, and all corrected predicted values ​​are combined into a complete corrected forecast curve of available renewable energy output. Specifically, the correction coefficient represents the correction strength determined based on the cumulative effect of historical deviations at the current moment. When applied to multiple future periods, it is necessary to consider the time distance between different periods and the current moment for differentiated processing. For example, a stronger correction effect is applied to periods closer to the current moment to enable them to respond more quickly to the latest operational deviations; a weaker correction effect is applied to periods farther from the current moment to maintain relative stability and avoid drastic jumps in long-term forecasts due to accidental fluctuations at the current moment. After the above differentiated correction, the corrected predicted values ​​for each future period are arranged in chronological order to construct a complete corrected forecast curve of available renewable energy output, which serves as input data for subsequent market clearing or scheduling decisions. The above multi-step rolling correction mechanism enables the predicted values ​​for each future period to maintain the overall trend while being differentiated according to the latest operational deviations, thereby improving the accuracy of the forecast curve in the near term and the stability in the far term.

[0066] In one embodiment, step 102 includes:

[0067] Step 1021: Construct a market scenario feature vector based on the operating parameters. The market scenario feature vector includes real-time electricity price, deviation assessment unit price, reserve ratio, and renewable energy volatility.

[0068] Specifically, at the current moment Construct market scenario feature vectors :

[0069] ;

[0070] In the formula, This is the real-time electricity price, in yuan / MWh; This is the day-ahead electricity price, in yuan / MWh; The unit price for deviation assessment is RMB / MWh; This represents the actual reserve ratio for the region. The volatility of new energy sources is expressed in MW. This indicates the data quality status, with values ​​of "normal" or "abnormal".

[0071] in, ;

[0072] in, For the current available power generation capacity, This represents the current regional load.

[0073] New energy volatility The standard deviation of actual output over four 15-minute intervals within the past hour can be used:

[0074] ;

[0075] in, , For the current realities of new energy, This represents the average actual output of new energy sources over the past hour. The loop variable in the summation formula is used to iterate through the various sampling periods in the past hour, and it represents the j-th historical sampling point.

[0076] Step 1022: Determine whether the data quality of the operating parameters is normal;

[0077] Specifically, the process begins by checking for missing operating parameters. For example, it checks if the missing rate of metering data exceeds a preset threshold (e.g., 20%). If the missing rate exceeds 20%, the data quality is deemed abnormal. It also checks for communication interruptions. If the interruption time exceeds a preset continuous sampling period threshold (e.g., two periods), the data quality is deemed abnormal. Furthermore, it checks for equipment alarms. If an alarm is received from a renewable energy power station or the data acquisition system, the data quality is deemed abnormal. Finally, it checks if the actual output of renewable energy is within a preset physical reasonable range. For example, whether the actual output is greater than the installed capacity or less than zero. If so, the data quality is deemed abnormal. Finally, it checks if key operating parameters are complete and available. For example, whether real-time electricity price, deviation assessment unit price, or reserve ratio are missing. If missing, the data quality is deemed abnormal. Then, if all the above checks are normal, the data quality of the operating parameters is deemed normal. If any of the above checks is abnormal, the data quality is deemed abnormal. This data quality assessment is used for subsequent market scenario identification. When data quality is abnormal, it is directly identified as an abnormal scenario and correction is stopped to prevent abnormal data from participating in the correction process.

[0078] Step 1023: When the data quality is abnormal, determine that the current market scenario is an abnormal scenario;

[0079] Specifically, the determination of an abnormal scenario (denoted as S2) includes at least the following situations: the metering data missing rate is greater than a preset missing rate threshold (e.g., 20%), the communication interruption time exceeds a preset continuous sampling period threshold (e.g., 2 periods), receiving equipment alarm information from a new energy power station or the data acquisition system, physical anomalies such as the actual output of new energy exceeding the installed capacity or being less than zero, and missing key operating parameters such as real-time electricity price, deviation assessment unit price, or reserve ratio. When any of the above situations occurs, the data quality is determined to be abnormal, and the corresponding current market scenario is an abnormal scenario. In this embodiment, the existence of a communication interruption is first checked. If the communication interruption time exceeds 2 continuous sampling periods (e.g., 30 minutes), it is directly determined to be an abnormal scenario. Then, the existence of equipment alarm information is checked. If the data acquisition system sends equipment alarms, it is also determined to be an abnormal scenario. Once a scenario is determined to be an abnormal scenario, the subsequent correction process will stop executing automatic correction and output alarms to avoid abnormal data affecting the accuracy of the correction results.

[0080] Step 1024: When the data quality is normal, compare the real-time electricity price, the deviation assessment unit price, the reserve ratio, and the new energy volatility with their respective thresholds, and determine whether the current market scenario is a normal scenario or a high-risk scenario based on the comparison results.

[0081] Specifically, a scenario is considered normal (S0) when all of the following conditions are met simultaneously: the real-time electricity price is less than the preset high real-time electricity price threshold (e.g., 600 yuan / MWh), the deviation assessment unit price is less than the preset high deviation assessment threshold (e.g., 200 yuan / MWh), the reserve ratio is greater than or equal to the preset minimum reserve ratio (e.g., 0.08), the renewable energy volatility is less than or equal to the preset volatility threshold (e.g., 150MW), and the data quality status is normal. If the data quality is normal, but any one of the above conditions is not met (real-time electricity price greater than or equal to 600 yuan / MWh, or deviation assessment unit price greater than or equal to 200 yuan / MWh, or reserve ratio less than 0.08), it is considered a high-risk scenario (S1). In this embodiment, the current real-time electricity price is first compared with a preset high real-time electricity price threshold. If the real-time electricity price is 650 yuan / MWh, which is greater than 600 yuan / MWh, it is directly identified as a high-risk scenario. If the real-time electricity price is less than 600 yuan / MWh, the deviation assessment unit price, reserve ratio, and renewable energy volatility are further compared. If all threshold conditions are met, it is identified as a normal scenario. Through the above scenario division, the correction strategy can adaptively adjust according to the current market operation risk. That is, in high-risk scenarios, the response sensitivity to prediction deviations is increased, while in normal scenarios, it is handled in a conventional manner.

[0082] In one embodiment, step 103 includes:

[0083] Step 1031: Obtain the preset target output and calculate the difference between the actual output of the new energy source at the current moment and the target output to obtain the prediction deviation;

[0084] Specifically, the preset target output Contributing to the recent market clearing plan Given that an intraday market correction plan has already been generated, the pre-set target output will contribute to the intraday market correction plan. If neither the daily nor intraday plan has been generated, the target output is set at the previously revised forecast value. or original predicted value .

[0085] Then, the difference between the actual output of the new energy source at the current moment and the preset target output is calculated to obtain the prediction deviation, i.e. For example, if the actual output of new energy is 2900MW, and the preset target output is 3100MW (as per the day-ahead market clearing plan), then the prediction deviation is -200MW. A negative value indicates that the actual output is less than the target output, posing a risk of under-generation. If the actual output of new energy is 3300MW, and the preset target output is 3100MW, then the prediction deviation is +200MW. A positive value indicates that the actual output is greater than the target output, posing a risk of over-generation or curtailment. These prediction deviations are used for subsequent weighted processing based on the current market scenario and the direction of the deviation, in order to respond differently to under-generation and over-generation deviations under different market risks.

[0086] Furthermore, in one embodiment, when the regional spot market system uses day-ahead plan deviation assessment as the primary assessment method, the day-ahead market clearing plan output is selected as the preset target output, i.e. = When prediction deviation When the value is less than 0, it indicates that the actual output of new energy is less than the preset target output, posing a risk of under-output deviation. In this case, if it is in a high-risk scenario, the negative deviation needs to be amplified to strengthen the correction. A value greater than 0 indicates that the actual output of new energy sources exceeds the preset target output, suggesting a risk of over-generation or curtailment. In high-risk scenarios, this positive deviation should be reduced to avoid excessive upward adjustment of the forecast. For details on how the deviation direction is coupled with the market scenario, please refer to step 1032.

[0087] Step 1032: When the current market scenario is the normal scenario, the prediction deviation is used as the scenario-weighted error;

[0088] Specifically, a normal scenario indicates low real-time electricity prices and deviation assessment costs, sufficient reserves, and acceptable fluctuations in renewable energy, resulting in low market operation risk. Therefore, no additional amplification or reduction processing of the deviation is required; the predicted deviation calculated in step 1031 is directly used as the scenario-weighted error, i.e., r(t) = e(t). In this embodiment, assuming the predicted deviation e(t) is -200MW, since the current market scenario is a normal scenario, the scenario-weighted error r(t) is directly taken as -200MW without any scaling adjustment. First, no weighting processing is applied to the predicted deviation under normal scenarios; then, the predicted deviation is directly used as the scenario-weighted error and passed to the subsequent confidence update step. This processing method ensures that during periods of low market risk, the correction logic maintains a normal response, preventing the introduction of additional fluctuations due to over-correction.

[0089] Step 1033: When the current market scenario is the high-risk scenario and the prediction deviation is negative, the prediction deviation is amplified with the first weight to obtain the scenario-weighted error;

[0090] Specifically, the first weight is greater than 1, and in the previous embodiment, it is taken as 1.5 times. High-risk scenarios indicate high real-time electricity prices, high deviation assessment unit prices, or insufficient reserve ratios. The system is more sensitive to new energy deviations, and a negative prediction deviation (actual output less than target output) means there is a risk of under-generation deviation, which may lead to higher deviation assessment costs or even jeopardize system safety. For example, if the prediction deviation e(t) is -200MW, the current market scenario is a high-risk scenario, and the prediction deviation is negative, amplifying the prediction deviation with the first weight of 1.5 yields a scenario-weighted error r(t) of -300MW. First, it is determined that the current market scenario is a high-risk scenario and the prediction deviation is negative. Then, the prediction deviation is amplified with the first weight to obtain the scenario-weighted error. This processing method makes the system more sensitive to under-generation deviations in high-risk scenarios, and the corrected prediction value is more conservative, helping to reduce the risk of insufficient actual output due to over-prediction and ensuring the safe operation of the power system.

[0091] Step 1034: When the current market scenario is the high-risk scenario and the prediction deviation is non-negative, the prediction deviation is reduced by the second weight to obtain the scenario-weighted error, where the first weight is greater than the second weight.

[0092] Specifically, a non-negative prediction deviation (actual output greater than or equal to target output) indicates a risk of over-generation or curtailment. In high-risk scenarios, directly amplifying this deviation could lead to an over-adjusted prediction value, resulting in an overly aggressive clearing outcome. Therefore, in the previous embodiment, the second weight is set to 0.8 to reduce the prediction deviation. For example, if the prediction deviation e(t) is +200MW, the current market scenario is a high-risk scenario, and the prediction deviation is non-negative, using the second weight of 0.8 to reduce the prediction deviation yields a scenario-weighted error r(t) of +160MW. First, it is determined that the current market scenario is a high-risk scenario and the prediction deviation is non-negative; then, the prediction deviation is reduced using the second weight to obtain the scenario-weighted error. This approach ensures that in high-risk scenarios, the system's response to over-generation deviation is relatively conservative, avoiding an overly aggressive clearing outcome due to excessive upward adjustment of the prediction, thereby reducing the risk of curtailment. Meanwhile, since the first weight is greater than the second weight, in high-risk scenarios, the degree to which less frequent deviations (negative deviations) are amplified is greater than the degree to which excessive deviations (non-negative deviations) are reduced, reflecting a higher priority focus on the risk of less frequent deviations.

[0093] The process from steps 1032 to 1034 above can be represented as follows:

[0094] ;

[0095] In one embodiment, step 104 includes:

[0096] Step 1041: Determine the adaptive smoothing coefficient based on the current market scenario, and calculate the normalized error contribution based on the scenario weighted error;

[0097] It should be noted that in this embodiment, the initial value of the prediction correction confidence level, b(0), is set to 0. This value represents that at the start of the correction process, the system has not yet accumulated any historical deviation information, so the tendency to correct the prediction results remains neutral. As subsequent time periods are updated, the confidence level will gradually converge to a reasonable level based on the actual deviation.

[0098] Specifically, firstly, an adaptive smoothing coefficient is determined based on the current market scenario, wherein the adaptive smoothing coefficient... This value controls the weighting of old and new information during confidence level updates, taking different values ​​for different market scenarios: 0.3 in normal scenarios, indicating that historical information is the primary factor and the current error contribution is secondary; 0.6 in high-risk scenarios, indicating that the current error contribution is the primary factor and historical information is secondary, improving the sensitivity to the latest deviations; and 0 in abnormal scenarios, which stops confidence level updates to avoid abnormal data affecting the reliability of the confidence level.

[0099] Then, the normalized error contribution is calculated based on the scene weighted error obtained in step 103. Specifically, the hyperbolic tangent function is used. (.) The scenario-weighted error is normalized, mapping it to the interval [-1, 1], where the normalization benchmark is 10% of the regional new energy installed capacity. The formula for calculating the contribution of the normalized error is as follows:

[0100] ;

[0101] For example, the scene weighted error r(t) is -300MW, and the normalized benchmark... It is 500MW, that is If the power density is 500MW, then the normalized error contribution u(t) = tanh(-300 / 500) ≈ -0.537. The above adaptive smoothing coefficient and normalized error contribution are used for subsequent recursive updates of the confidence level, enabling the confidence level to adaptively adjust its update speed under different market risks, while preventing a single abnormal error from excessively affecting the confidence level, thus improving the stability and reliability of the confidence level update.

[0102] Step 1042: Obtain the prediction correction confidence of the previous time step, and recursively update it based on the adaptive smoothing coefficient, the normalization error contribution, and the prediction correction confidence of the previous time step to obtain the prediction correction confidence of the current time step.

[0103] Specifically, the prediction correction confidence level of the previous time step is first obtained. When t=1 (i.e., the first calculation time), the prediction correction confidence is initialized to b(0)=0, indicating that without any historical bias information, the initial correction tendency is zero and does not favor any direction. When t>1, the prediction confidence is directly obtained from the recursive update of the previous time. For example, the revised confidence level of the prediction at the previous time step. It is 0.1. Then, based on the adaptive smoothing coefficient... Normalized error contribution Confidence level of prediction revised from the previous time step The prediction correction confidence level is obtained by updating the prediction using the following recursive formula. :

[0104]

[0105] For example, in a high-risk market scenario, the adaptive smoothing coefficient... =0.6, normalized error contribution ≈-0.537, the revised confidence level of the previous time step prediction =0.1, then the prediction correction confidence level at the current time is 0.1. =0.6×(-0.537)+(1-0.6)×0.1≈-0.282. This recursive update allows the confidence score to incorporate historical bias information and current error contribution, achieving a smooth update. When A value greater than 0 indicates that the actual output in the recent period is relatively high compared to the target output, and subsequent forecasts can be revised upwards; when... When <0, it indicates that the actual output in the near term is relatively low compared to the target output, and subsequent forecasts can be revised downwards; when A confidence level close to 0 indicates that the recent deviation direction is unstable or the need for correction is weak. This confidence level is used in the calculation of subsequent correction coefficients, enabling the output of the correction coefficients to adaptively adjust according to changes in the confidence level.

[0106] Step 1043: Determine the ideal correction coefficient based on the updated prediction correction confidence; obtain the actual reserve ratio of the region at the current time, and determine the upper limit of the correction coefficient based on the actual reserve ratio of the region; compare the ideal correction coefficient with the upper limit of the correction coefficient, and take the smaller value of the two as the final correction coefficient.

[0107] Specifically, the confidence level is first adjusted based on the updated prediction. Determine the ideal correction factor The updated predictions will be adjusted to improve confidence. Substitute into the following formula:

[0108] ;

[0109] In the formula, Adjust the confidence level for the updated prediction. , as well as The preset confidence threshold, This represents the actual reserve ratio for the region.

[0110] It should be noted that the preset confidence threshold in this embodiment is set as follows: =-0.3, =0.2, =0.6.

[0111] Then, to prevent over-correction during periods of low reserve capacity, an upper limit for the correction factor is set. ;

[0112] ;

[0113] The final correction factor is:

[0114] ;

[0115] in, Used for subsequent multi-step rolling corrections.

[0116] The aforementioned final correction coefficient is used to perform multi-step rolling corrections on the predicted values ​​for future periods, so that the correction magnitude matches the confidence level. At the same time, the correction magnitude is automatically limited when the reserve is strained to prevent excessive correction from affecting system security.

[0117] In one embodiment, step 105 involves performing multi-step rolling corrections on the predicted values ​​for each future period in the original forecast curve of available renewable energy output using the correction coefficient, including:

[0118] Step 1051: Generate a correction direction factor based on the updated prediction correction confidence.

[0119] Specifically, define the correction direction factor. for:

[0120] ;

[0121] As can be seen from the above formula, when b(t)>0, θ(t)>0, indicating that the subsequent correction direction tends to be upward; when b(t)<0, θ(t)<0, indicating that the subsequent correction direction tends to be downward; the larger the absolute value of b(t), the larger the absolute value of θ(t), indicating that the correction tendency is stronger.

[0122] Step 1052: Determine the time decay factor corresponding to each future time period based on the prediction step size corresponding to each future time period. The time decay factor decreases as the prediction step size increases.

[0123] Specifically, let the time decay factor be:

[0124] ;

[0125] in, =1, ≈0.74, ≈0.50, ≈0.22. This setting indicates that the closer to the current time, the larger the correction magnitude; the farther away from the current time, the smaller the correction magnitude gradually becomes.

[0126] Step 1053: For each future time period, correct the original predicted value of the future time period according to the correction coefficient, the time decay factor corresponding to the future time period and the correction direction factor, to obtain the corrected predicted value of the available power output of new energy for the future time period.

[0127] Specifically, the revised forecast of available renewable energy output for the i-th time period in the future. The following calculation formula is used:

[0128] ;

[0129] In the formula, This represents the original predicted value for the i-th time period in the future.

[0130] To ensure that the corrected predictions meet the physical constraints, boundary restrictions are also required:

[0131] ;

[0132] in, This represents the region's installed capacity of new energy sources.

[0133] Understandably, the process begins by obtaining the original predicted values ​​for each future time period. Then, the final correction coefficient, the time decay factor corresponding to that future time period, and the correction direction factor are substituted into the aforementioned formula for correction, yielding the corrected predicted value of available renewable energy output for that future time period. Furthermore, to ensure that the corrected predicted value meets physical constraints, boundary restrictions are also necessary: ​​when the corrected predicted value of available renewable energy output is less than a preset minimum output limit (e.g., 0), the preset minimum output limit is used as the final predicted value of available renewable energy output for that future time period; when the corrected predicted value of available renewable energy output is greater than the preset maximum output limit (i.e., the regional renewable energy installed capacity), the predicted value of available renewable energy output is greater than the preset maximum output limit (i.e., the regional renewable energy installed capacity). When the maximum output limit is preset, the final predicted value of available renewable energy output for that future period is used. Finally, based on the corrected predicted value, a revised prediction curve of available renewable energy output is constructed as follows:

[0134] ;

[0135] Furthermore, after obtaining the corrected renewable energy available output prediction curve, the system can also output the corrected renewable energy available output prediction curve, along with the corresponding current confidence level, final correction coefficient, current market scenario label, and anomaly alarm information. The aforementioned corrected renewable energy available output prediction curve can serve as the upper limit or reference boundary for renewable energy output during market clearing, participating in subsequent power generation planning, reserve dispatch, or real-time dispatch. Simultaneously, when the current market scenario is determined to be an abnormal scenario, or when the final correction coefficient is 0 for multiple consecutive time periods, it indicates that the current data quality is insufficient to support automatic correction, or the correction magnitude remains zero, indicating weak necessity for correction. This can trigger a manual review process, where operators review and confirm the correction results to prevent abnormal data or invalid correction results from affecting the accuracy of subsequent decisions, further improving the safety and reliability of system operation.

[0136] This invention provides a market boundary adaptive correction method. First, it acquires the original predicted power output curves of renewable energy for multiple future time periods and the current operating parameters. Based on these operating parameters, it identifies the current market scenario, achieving dynamic coupling between the correction strategy and market operating risks. This solves the problem that existing fixed-parameter correction methods cannot adaptively adjust to market risk factors. Then, it calculates the prediction deviation between the actual renewable energy output and the preset target output, and applies differentiated weighting to this deviation based on the current market scenario and the direction of the deviation. This allows the system to automatically improve its sensitivity to key deviations in high-risk scenarios and maintain stability in normal scenarios. The invention achieves a stable response; then, it recursively updates the prediction correction confidence level through scenario-weighted error, and calculates the correction coefficient based on the confidence level, establishing a quantifiable and interpretable automatic correction rule, thus solving the problem that existing correction methods lack a mechanism for quantitatively evaluating the reliability of correction results. Finally, it performs multi-step rolling corrections on the predicted values ​​for each future time period using the correction coefficients and constructs a corrected prediction curve, enabling near-term predictions to respond more quickly to the latest operational deviations and far-term predictions to remain relatively stable. Simultaneously, when operational parameters are abnormal, it identifies an abnormal scenario and automatically stops correction, solving the safety risk problem of existing solutions continuing corrections when data is abnormal and introducing abnormal data into subsequent processes. This invention achieves end-to-end adaptive correction from scenario identification, deviation weighting, confidence quantification to rolling correction, reducing human experience intervention and improving the consistency and automation level of correction results. In summary, this invention solves the problems of existing correction schemes that are difficult to adaptively adjust correction strategies according to market operational risks and lack a mechanism for quantitatively evaluating the reliability of correction results and coupling with risk.

[0137] Example 2: Downward revision of forecasts in high-risk market scenarios:

[0138] In a certain provincial-level spot market system, the total installed capacity of new energy is 5000MW, the time granularity of the spot market is 15 minutes, and the prediction step size is 16 time periods.

[0139] Note: The regional spot market system is the technical platform for the electricity spot market, used to organize and manage electricity spot transactions within a specific geographical area. The regional spot market system determines real-time electricity prices and trading plans through market mechanisms, typically including trading instruments at different time scales such as day-ahead, intraday, and real-time markets. It is the core technical system for the operation of the regional electricity market.

[0140] The current time is 10:00, and the system has collected the data shown in Table 2:

[0141] Table 2

[0142]

[0143] 1. Scene recognition:

[0144] because: =650≥600 and: =220≥200, therefore the current situation is determined to be a high-risk scenario S1.

[0145] 2. Prediction deviation calculation:

[0146] Take the effort planned for the previous day as the target for effort: = 3100; then: =2900-3100=-200MW.

[0147] A negative deviation indicates that the actual output is lower than the planned output, and there is a risk of under-output.

[0148] 3. Calculation of scene-weighted error:

[0149] Because the current scenario is high-risk S1, and <0, therefore:

[0150] =1.5×(-200)=-300MW;

[0151] 4. Confidence update:

[0152] Take in high-risk scenarios: =0.6;

[0153] The contribution of normalization error is:

[0154] (-300 / 500)=tanh(-0.6)≈-0.537;

[0155] The confidence level is updated to: =0.6×(-0.537)+(1-0.6)×0.1、 ≈-0.322+0.04=-0.282;

[0156] 5. Calculation of correction factor:

[0157] because: =-0.3< =-0.282≤ =0.2;

[0158] therefore =0.5( - ) = 0.5(-0.282+0.3) = 0.009;

[0159] And because =0.06≥0.055, therefore: =1;

[0160] so: =0.009;

[0161] The results indicate that although the current scenario is high-risk, the system only makes minimal corrections because the confidence level is close to the low threshold, thus avoiding overcorrection caused by erroneous data.

[0162] 6. Multi-step rolling correction:

[0163] For the first 15-minute period in the future =1, Correction direction factor:

[0164] =-0.282 / (1+0.282)≈-0.220;

[0165] If the original forecast value for the first future time period is:

[0166] =3150MW;

[0167] but: =(1+0.009×1×(-0.220))×3150, ≈3143.8MW;

[0168] This means that the forecast value for the first period in the future will be slightly lowered.

[0169] For the 8th time period in the future, The correction magnitude is approximately 0.50, further reducing the magnitude of the correction. This mechanism ensures that the near-term period is strongly affected, while the long-term period remains relatively stable.

[0170] Example 3: Automatically shutting down correction in abnormal scenarios:

[0171] If a communication interruption occurs at the current moment, and no actual output of new energy is received for two consecutive sampling cycles, or if the metering data missing rate exceeds 20%, the market scenario identification unit will determine it as an abnormal scenario S2.

[0172] at this time: =0、 =0、 =0;

[0173] The forecast correction device no longer adjusts the forecast curve based on current abnormal data, but instead outputs anomaly alarms to the regional spot market system. The regional spot market system can participate in market clearing using the original forecast curve of available renewable energy output, the backup forecast curve, or the results of manual verification.

[0174] Example 4:

[0175] Please see Figure 2 An adaptive market boundary correction system provided in this embodiment of the invention includes:

[0176] The acquisition unit 201 is used to acquire the original prediction curves of the available power output of new energy for multiple future time periods and the operating parameters at the current time.

[0177] It should be noted that, firstly, the raw forecast curves of available power output from new energy sources for multiple future time periods are obtained. New energy sources include wind power, photovoltaic power, and other renewable energy sources. The raw forecast curves of available power output from new energy sources are denoted as follows: Where t is the current time, i is the prediction step size (i=1,2,…,N), and N is the total number of prediction steps. In this embodiment, the time granularity is 15 minutes, and the prediction step size N is 16, which covers the next 4 hours. The original prediction value Pforecast(t+i) for each time period can be directly read. Then, the running parameters at the current time are obtained.

[0178] The identification unit 202 is used to identify the current market scenario based on the operating parameters, wherein the current market scenario includes a normal scenario, a high-risk scenario, and an abnormal scenario;

[0179] It should be noted that the completeness and rationality of operating parameters are first assessed. This includes checking for communication interruptions, missing data, and whether actual output exceeds reasonable limits. If any of these anomalies occur, the current data is considered unreliable. Continuing the correction process in this case could introduce abnormal data, thus it is identified as an abnormal scenario and correction is stopped. Assuming normal data quality, the market operation risk level is further evaluated. For example, whether the real-time electricity price is too high, the deviation assessment unit price is too high, or the reserve ratio is insufficient. When all these risk indicators are within safe limits, it is identified as a normal scenario, and the correction logic maintains a normal response. However, when any risk indicator exceeds a threshold, it is identified as a high-risk scenario, and the correction logic needs to increase its sensitivity to deviations to cope with market risks. This scenario classification allows the correction strategy to adaptively adjust according to data status and market risks, proactively protecting against data anomalies, operating smoothly under low risk, and strengthening the response under high risk, thereby balancing the effectiveness of correction and system security.

[0180] The calculation unit 203 is used to calculate the prediction deviation between the actual output of new energy and the preset target output, and to perform weighted processing on the prediction deviation according to the current market scenario and the deviation direction of the prediction deviation to obtain the scenario-weighted error.

[0181] It should be noted that, firstly, the difference between the actual output of new energy and the preset target output is calculated to obtain the prediction deviation. This difference is used to quantify the degree of deviation between the actual operating state of new energy and the expected target at the current moment, and serves as the basis for subsequent corrections. Then, based on the current market scenario identified in step 102 and the direction of the prediction deviation, the prediction deviation is subjected to differentiated weighting processing to obtain the scenario-weighted error. For example, in normal scenarios, the deviation remains at its original value; in high-risk scenarios, if the deviation direction is negative, it is appropriately amplified; if the deviation direction is non-negative, it is appropriately reduced; and in abnormal scenarios, the current deviation is not used for correction. The above differentiated weighting processing enables the correction strategy to adaptively adjust the response sensitivity according to the market risk level and the direction of deviation, avoiding over-correction or under-correction caused by using a uniform correction intensity in different risk scenarios. The scenario-weighted error will be used as the input for subsequent confidence updates, allowing the cumulative effect of historical errors to be quantified and transmitted to the calculation of the correction coefficient.

[0182] The update unit 204 is used to update the prediction correction confidence at the current moment according to the scenario weighted error, and calculate the correction coefficient according to the updated prediction correction confidence.

[0183] It should be noted that, firstly, based on the scenario-weighted error, the prediction correction confidence level of the previous moment is recursively updated to obtain the prediction correction confidence level of the current moment. This prediction correction confidence level quantifies the credibility and direction of the current prediction correction result. Its update method employs an adaptive smoothing mechanism, assigning different weights to the current error and historical information under different market scenarios. For example, in high-risk scenarios, the weight of the current error is increased to allow the confidence level to respond more quickly to the latest deviation changes, while in normal scenarios, a lower weight is maintained to filter out short-term fluctuations. Then, a correction coefficient is calculated based on the updated prediction correction confidence level. This correction coefficient controls the magnitude of subsequent prediction corrections; a higher confidence level results in a larger correction magnitude, while a lower confidence level results in a smaller correction magnitude or stops correction altogether. This confidence level update and correction coefficient calculation enable the correction strategy to adaptively adjust the correction intensity based on the cumulative effect of historical deviations, avoiding over-correction caused by a single deviation fluctuation. Simultaneously, the confidence level is quantitatively assessed to evaluate the credibility of the correction result, providing a quantifiable basis for subsequent correction steps.

[0184] The correction unit 205 is used to perform multi-step rolling correction on the predicted values ​​of each future time period in the original prediction curve of available new energy output using the correction coefficient, and to construct a corrected prediction curve of available new energy output based on the corrected predicted values.

[0185] It should be noted that correction coefficients are used to correct the predicted values ​​for each future period in the original forecast curve of available renewable energy output, and all corrected predicted values ​​are combined into a complete corrected forecast curve of available renewable energy output. Specifically, the correction coefficient represents the correction strength determined based on the cumulative effect of historical deviations at the current moment. When applied to multiple future periods, it is necessary to consider the time distance between different periods and the current moment for differentiated processing. For example, a stronger correction effect is applied to periods closer to the current moment to enable them to respond more quickly to the latest operational deviations; a weaker correction effect is applied to periods farther from the current moment to maintain relative stability and avoid drastic jumps in long-term forecasts due to accidental fluctuations at the current moment. After the above differentiated correction, the corrected predicted values ​​for each future period are arranged in chronological order to construct a complete corrected forecast curve of available renewable energy output, which serves as input data for subsequent market clearing or scheduling decisions. This multi-step rolling correction mechanism enables the predicted values ​​for each future period to maintain the overall trend while being differentiated according to the latest operational deviations, thereby improving the accuracy of the forecast curve in the near term and the stability in the long term.

[0186] In one embodiment, the acquisition unit 202 is specifically used for:

[0187] A market scenario feature vector is constructed based on the operating parameters. The market scenario feature vector includes real-time electricity price, deviation assessment unit price, reserve ratio, and renewable energy volatility.

[0188] Determine whether the data quality of the operating parameters is normal;

[0189] When the data quality is abnormal, the current market scenario is determined to be an abnormal scenario;

[0190] When the data quality is normal, the real-time electricity price, the deviation assessment unit price, the reserve ratio, and the new energy volatility are compared with their respective thresholds, and the current market scenario is determined to be a normal scenario or a high-risk scenario based on the comparison results.

[0191] In one embodiment, the acquisition unit 203 is specifically used for:

[0192] Obtain the preset target output and calculate the difference between the actual output of the new energy source at the current moment and the target output to obtain the prediction deviation;

[0193] When the current market scenario is the normal scenario, the prediction deviation is used as the scenario-weighted error;

[0194] When the current market scenario is the high-risk scenario and the prediction deviation is negative, the prediction deviation is amplified with a first weight to obtain the scenario-weighted error;

[0195] When the current market scenario is the high-risk scenario and the prediction deviation is non-negative, the prediction deviation is reduced by the second weight to obtain the scenario-weighted error, where the first weight is greater than the second weight.

[0196] In one embodiment, the acquisition unit 204 is specifically used for:

[0197] The adaptive smoothing coefficient is determined based on the current market scenario, and the normalized error contribution is calculated based on the scenario-weighted error.

[0198] Obtain the prediction correction confidence of the previous time step, and recursively update it based on the adaptive smoothing coefficient, the normalization error contribution, and the prediction correction confidence of the previous time step to obtain the prediction correction confidence of the current time step.

[0199] The ideal correction coefficient is determined based on the updated prediction correction confidence; the actual reserve ratio of the region at the current time is obtained, and the upper limit of the correction coefficient is determined based on the actual reserve ratio of the region; the ideal correction coefficient and the upper limit of the correction coefficient are compared, and the smaller value of the two is taken as the final correction coefficient.

[0200] In one embodiment, the acquisition unit 205 performs multi-step rolling corrections on the predicted values ​​for each future time period in the original prediction curve of available renewable energy output using the correction coefficient, including:

[0201] A correction direction factor is generated based on the updated prediction correction confidence.

[0202] The time decay factor corresponding to each future time period is determined based on the prediction step size corresponding to each future time period, and the time decay factor decreases as the prediction step size increases;

[0203] For each future time period, the original predicted value of the future time period is corrected according to the correction coefficient, the time decay factor corresponding to the future time period, and the correction direction factor, to obtain the corrected predicted value of the available power output of new energy for the future time period.

[0204] This invention provides a market boundary adaptive correction system. First, it acquires the original predicted power output curves of renewable energy for multiple future time periods and the current operating parameters. Based on these operating parameters, it identifies the current market scenario, achieving dynamic coupling between the correction strategy and market operating risks. This solves the problem that existing fixed-parameter correction methods cannot adaptively adjust to market risk factors. Then, it calculates the prediction deviation between the actual renewable energy output and the preset target output, and applies differentiated weighting to this deviation based on the current market scenario and the direction of the deviation. This allows the system to automatically improve its sensitivity to key deviations in high-risk scenarios and maintain stability in normal scenarios. The invention achieves a stable response; then, it recursively updates the prediction correction confidence level through scenario-weighted error, and calculates the correction coefficient based on the confidence level, establishing a quantifiable and interpretable automatic correction rule, thus solving the problem that existing correction methods lack a mechanism for quantitatively evaluating the reliability of correction results. Finally, it performs multi-step rolling corrections on the predicted values ​​for each future time period using the correction coefficients and constructs a corrected prediction curve, enabling near-term predictions to respond more quickly to the latest operational deviations and far-term predictions to remain relatively stable. Simultaneously, when operational parameters are abnormal, it identifies an abnormal scenario and automatically stops correction, solving the safety risk problem of existing solutions continuing corrections when data is abnormal and introducing abnormal data into subsequent processes. This invention achieves end-to-end adaptive correction from scenario identification, deviation weighting, confidence quantification to rolling correction, reducing human experience intervention and improving the consistency and automation level of correction results. In summary, this invention solves the problems of existing correction schemes that are difficult to adaptively adjust correction strategies according to market operational risks and lack a mechanism for quantitatively evaluating the reliability of correction results and coupling with risk.

[0205] Example 5:

[0206] This invention also provides a market boundary adaptive correction device, the device including a processor and a memory:

[0207] The memory is used to store program code and transmit the program code to the processor;

[0208] The processor is configured to execute the steps of the market boundary adaptive correction method as described in the above method embodiments, according to the instructions in the program code.

[0209] It should be noted that this embodiment of the invention also provides a market boundary adaptive correction device, which includes a processor and a memory. The memory and the processor can be coupled to each other via a bus or other means to realize the transmission of data and instructions. The memory is used to store program code and transmit the program code to the processor. The memory may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device or other non-volatile solid-state memory device. The processor is used to execute the steps of the market boundary adaptive correction method as described in the above method embodiment according to the instructions in the program code. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. In this embodiment, the processor, by running program code stored in memory, can obtain the original forecast curves of available renewable energy output for multiple future time periods and the current operating parameters. Based on the operating parameters, it identifies the current market scenario, calculates the prediction deviation and performs scenario-weighted error calculation, updates the prediction correction confidence level, and calculates the correction coefficient. Finally, it uses the correction coefficient to perform multi-step rolling corrections on the predicted values ​​for each future time period in the original forecast curve of available renewable energy output, constructing a corrected forecast curve of available renewable energy output, thereby achieving adaptive correction processing throughout the entire process. This device enables automatic execution of the correction method, improving correction efficiency and reducing manual intervention costs. Simultaneously, the collaborative work of the processor and memory ensures the high efficiency of data reading, logical operations, and result storage, guaranteeing the stability and reliability of the correction process.

[0210] Example 6:

[0211] This invention also provides a computer-readable storage medium for storing program code for executing the market boundary adaptive correction method described in the above method embodiments.

[0212] It should be noted that the embodiments of the present invention also provide a computer-readable storage medium, which includes, but is not limited to, USB flash drives, mobile hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, and other media capable of storing program code. The computer-readable storage medium is used to store program code. When the program code is read and executed by one or more processors, it can implement all or part of the steps of the market boundary adaptive correction method described in the above method embodiments, including obtaining the original prediction curve of the available output of new energy for multiple future time periods and the operating parameters at the current time; identifying the current market scenario based on the operating parameters; calculating the prediction deviation between the actual output of new energy and the preset target output and performing weighted processing based on the deviation direction of the current market scenario and the prediction deviation to obtain the scenario weighted error; updating the prediction correction confidence at the current time based on the scenario weighted error and calculating the correction coefficient; and performing multi-step rolling correction on the prediction values ​​of each future time period in the original prediction curve of available output of new energy through the correction coefficient and constructing the corrected prediction curve of available output of new energy. The computer-readable storage medium described above allows the program code of the correction method to be stored, distributed, or sold independently, facilitating flexible deployment and operation on different computing devices without the need to repeatedly write program code for different devices, thereby reducing application costs and improving the portability and ease of promotion of the solution.

[0213] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0214] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0215] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0216] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0217] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0218] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A market boundary adaptive correction method, characterized in that, include: Obtain the original forecast curves of available power output from new energy sources for multiple future time periods, as well as the operating parameters at the current moment; The current market scenario is identified based on the operating parameters, and the current market scenario includes normal scenario, high-risk scenario and abnormal scenario; Calculate the prediction deviation between the actual output of new energy and the preset target output, and weight the prediction deviation according to the current market scenario and the direction of the deviation to obtain the scenario-weighted error; The prediction correction confidence level at the current moment is updated based on the scenario weighted error, and the correction coefficient is calculated based on the updated prediction correction confidence level. The original forecast curve for available renewable energy output is modified using the correction coefficient in multiple rolling steps for each future time period, and a modified forecast curve for available renewable energy output is constructed based on the modified forecast values.

2. The market boundary adaptive correction method according to claim 1, characterized in that, The step of identifying the current market scenario based on the operating parameters, wherein the current market scenario includes normal scenarios, high-risk scenarios, and abnormal scenarios, including: A market scenario feature vector is constructed based on the operating parameters. The market scenario feature vector includes real-time electricity price, deviation assessment unit price, reserve ratio, and renewable energy volatility. Determine whether the data quality of the operating parameters is normal; When the data quality is abnormal, the current market scenario is determined to be an abnormal scenario; When the data quality is normal, the real-time electricity price, the deviation assessment unit price, the reserve ratio, and the new energy volatility are compared with their respective thresholds, and the current market scenario is determined to be a normal scenario or a high-risk scenario based on the comparison results.

3. The market boundary adaptive correction method according to claim 1, characterized in that, The calculation of the prediction deviation between the actual output of new energy and the preset target output, and the weighting of the prediction deviation according to the current market scenario and the direction of the deviation, to obtain the scenario-weighted error, includes: Obtain the preset target output and calculate the difference between the actual output of the new energy source at the current moment and the target output to obtain the prediction deviation; When the current market scenario is the normal scenario, the prediction deviation is used as the scenario-weighted error; When the current market scenario is the high-risk scenario and the prediction deviation is negative, the prediction deviation is amplified with a first weight to obtain the scenario-weighted error; When the current market scenario is the high-risk scenario and the prediction deviation is non-negative, the prediction deviation is reduced by the second weight to obtain the scenario-weighted error, where the first weight is greater than the second weight.

4. The market boundary adaptive correction method according to claim 1, characterized in that, The step of updating the prediction correction confidence at the current moment based on the scenario-weighted error includes: The adaptive smoothing coefficient is determined based on the current market scenario, and the normalized error contribution is calculated based on the scenario-weighted error. Obtain the prediction correction confidence score from the previous time step, and recursively update it based on the adaptive smoothing coefficient, the normalization error contribution, and the prediction correction confidence score from the previous time step to obtain the prediction correction confidence score from the current time step.

5. The market boundary adaptive correction method according to claim 1, characterized in that, The step of calculating the correction coefficient based on the updated prediction correction confidence includes: The ideal correction coefficient is determined based on the updated prediction correction confidence; the actual reserve ratio of the region at the current time is obtained, and the upper limit of the correction coefficient is determined based on the actual reserve ratio of the region; the ideal correction coefficient and the upper limit of the correction coefficient are compared, and the smaller value of the two is taken as the final correction coefficient.

6. The market boundary adaptive correction method according to claim 1, characterized in that, The process of performing multi-step rolling corrections on the predicted values ​​for each future period in the original forecast curve of available new energy output using the correction coefficient includes: A correction direction factor is generated based on the updated prediction correction confidence. The time decay factor corresponding to each future time period is determined based on the prediction step size corresponding to each future time period, and the time decay factor decreases as the prediction step size increases; For each future time period, the original predicted value of the future time period is corrected according to the correction coefficient, the time decay factor corresponding to the future time period, and the correction direction factor, to obtain the corrected predicted value of the available power output of new energy for the future time period.

7. The market boundary adaptive correction method according to claim 6, characterized in that, The calculation expression for the revised forecast value of available new energy output for the future time period is as follows: ； ； In the formula, For the future The revised forecast of available renewable energy output for each time period For the future Time decay factor for each time period, The correction coefficient is... For the future The original predicted values ​​for each time period, To correct the direction factor, This represents the region's installed capacity of new energy sources.

8. A market boundary adaptive correction system, characterized in that, include: The acquisition unit is used to acquire the original forecast curves of the available power output of new energy sources for multiple future time periods, as well as the operating parameters at the current moment. The identification unit is used to identify the current market scenario based on the operating parameters, wherein the current market scenario includes a normal scenario, a high-risk scenario, and an abnormal scenario. The calculation unit is used to calculate the prediction deviation between the actual output of new energy and the preset target output, and to perform weighted processing on the prediction deviation according to the current market scenario and the direction of the deviation to obtain the scenario-weighted error. The update unit is used to update the prediction correction confidence at the current moment according to the scenario weighted error, and to calculate the correction coefficient according to the updated prediction correction confidence. The correction unit is used to perform multi-step rolling corrections on the predicted values ​​of each future time period in the original prediction curve of available renewable energy output using the correction coefficient, and to construct a corrected prediction curve of available renewable energy output based on the corrected predicted values.

9. A market boundary adaptive correction device, characterized in that, The device includes a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is configured to execute the market boundary adaptive correction method according to any one of claims 1-7 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store program code for executing the market boundary adaptive correction method according to any one of claims 1-7.

Images