A computing power center electricity price calculation method and device, electronic equipment and storage medium

Abstract

The application discloses a computing power center electricity price calculation method and device, electronic equipment and storage medium, which are used for solving the technical problem that the time-of-use electricity price division is fixed, the real-time process state of the power grid cannot be reflected, all users face the same electricity price in the same time period regardless of the actual behavior difference, and the incentive accuracy is insufficient. The application comprises the following steps: collecting load time series data of a power supply inlet of a computing power center in a current scoring period through a preset sampling frequency; performing feature extraction on the load time series data to obtain a load form characteristic value; obtaining a demand response interaction record and a green electricity consumption proportion; calculating a power grid friendliness comprehensive score value according to the load form characteristic value, the demand response interaction record and the green electricity consumption proportion; calculating an electricity price discount of the computing power center according to the power grid friendliness comprehensive score value; and calculating an actual settlement electricity price according to the electricity price discount.

Description

Technical Field

[0001] This invention relates to the field of electricity price calculation technology, and in particular to a method, apparatus, electronic device, and storage medium for calculating electricity prices in a computing center. Background Technology

[0002] With the rapid development of the digital economy, computing centers, as a new type of infrastructure, are being deployed on a large scale, and their share of total electricity consumption in the society continues to rise. At the same time, the construction of new power systems themed around new energy sources is accelerating, making the power grid's demand for load-side flexibility resources increasingly urgent. Due to their large electricity consumption and flexible task scheduling, computing centers are considered to be a demand-side response resource with enormous potential.

[0003] Currently, electricity pricing for computing centers primarily relies on industrial and commercial electricity catalog prices or direct power supply agreements for large users, supplemented by peak-valley time-of-use pricing mechanisms. This pricing method only considers the time of day and the amount of electricity used, failing to take into account the actual electricity consumption behavior of computing centers and its contribution to or conflict with the power grid. Computing centers lack economic incentives to proactively optimize their load patterns and cooperate with grid dispatch, thus their significant flexibility resources remain untapped.

[0004] To address the aforementioned issues, existing technologies offer a demand response scheme based on peak-valley time-of-use pricing. Its working principle is as follows: the power grid divides a day into peak, high, flat, and low periods, setting different electricity prices for each. The computing center then schedules some deferred tasks to be executed during the low-price periods based on the price differences between these periods. This scheme is applied to large industrial and commercial users, but its main drawback is that the fixed time-of-use pricing fails to reflect the real-time status of the power grid, resulting in all users facing the same electricity price regardless of their actual behavior within the same time period, leading to insufficient incentive precision. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for calculating electricity prices in computing centers, which solves the technical problem that fixed time-of-use pricing cannot reflect the real-time operation status of the power grid, resulting in all users facing the same electricity price regardless of their actual behavior during the same period, and insufficient incentive accuracy.

[0006] This invention provides a method for calculating electricity prices for computing power centers, including:

[0007] By using a preset sampling frequency, load time-series data of the power supply entrance of the computing center within the current scoring period is collected;

[0008] Feature extraction is performed on the load time series data to obtain load pattern feature values;

[0009] Obtain demand response interaction records and green energy consumption ratio;

[0010] Based on the load pattern characteristic value, the demand response interaction record, and the green energy consumption ratio, calculate the comprehensive grid friendliness score;

[0011] The electricity price discount for the computing center is calculated based on the comprehensive grid friendliness score.

[0012] The actual settlement price is calculated based on the aforementioned electricity price discount.

[0013] Optionally, the step of extracting features from the load time-series data to obtain load pattern feature values ​​includes:

[0014] The load time series data is filtered to obtain the effective load time series data;

[0015] Feature extraction is performed on the effective load time series data to obtain load shape feature values.

[0016] Optionally, the load shape characteristic values ​​include smoothness characteristic values, peak-to-valley difference characteristic values, and rate of change characteristic values; the step of extracting features from the effective load time series data to obtain load shape characteristic values ​​includes:

[0017] Extract the sampling point load values ​​of all sampling points within the current scoring period from the effective load time series data;

[0018] Calculate the first load difference between the load values ​​of all adjacent sampling points;

[0019] Generate smoothness feature values ​​based on the first load difference;

[0020] Obtain the maximum and minimum load values ​​from all sampled load values;

[0021] Calculate the average load of all sampled load values;

[0022] Calculate the second load difference between the maximum load value and the minimum load value;

[0023] Calculate the ratio of the second load difference to the average load to obtain the peak-valley difference characteristic value;

[0024] Calculate the average of the absolute values ​​of all the first load differences to obtain the rate of change characteristic value.

[0025] Optionally, after the step of extracting features from the load time-series data to obtain load pattern feature values, the method further includes:

[0026] The current load profile of the computing center is determined based on the load profile characteristic values.

[0027] Optionally, the step of determining the current load profile of the computing center based on the load profile characteristic value includes:

[0028] When the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value are all lower than the corresponding preset low threshold, the current load state of the computing center is determined to be stable.

[0029] When the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value are all higher than the corresponding preset high threshold, the current load mode of the computing center is determined to be drastic fluctuation type.

[0030] When the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value are all lower than the corresponding preset high threshold, and at least one of the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value is between the corresponding preset low threshold and the preset high threshold, the current load mode of the computing center is determined to be a slow-changing type.

[0031] Optionally, the step of calculating the comprehensive grid-friendliness score based on the load pattern characteristic value, the demand response interaction record, and the green energy consumption ratio includes:

[0032] The smoothness feature values ​​are converted into smoothness scores;

[0033] The peak-valley difference characteristic value is converted into a peak-valley difference fraction;

[0034] Convert the rate of change characteristic value into a rate of change fraction;

[0035] The load morphology dimension score is obtained by weighted summation of the smoothness score, the peak-to-valley difference score, and the rate of change score.

[0036] Obtain the demand response interaction records within the target scoring period; the target scoring period includes the current scoring period and several scoring periods preceding the current scoring period;

[0037] Calculate the response rate, response timeliness rate, and response depth compliance rate based on the aforementioned demand response interaction records;

[0038] The response rate is mapped to a response rate score, the response timeliness rate is mapped to a response timeliness rate score, and the response depth compliance rate is mapped to a response depth compliance rate score.

[0039] The response rate score, the response timeliness score, and the response depth compliance rate score are weighted and summed to obtain the demand response dimension score.

[0040] The green energy consumption ratio is mapped to a green energy consumption dimension score;

[0041] The scores for the load profile dimension, the demand response dimension, and the green energy consumption dimension are weighted and summed using preset dimension weights to obtain a comprehensive grid friendliness score.

[0042] Optionally, after the step of calculating the actual settlement electricity price based on the electricity price discount, the method further includes:

[0043] Improvement dimensions are determined based on the load pattern dimension score, the demand response dimension score, and the green energy consumption dimension score;

[0044] Computing task scheduling is performed based on the aforementioned improvement dimensions.

[0045] The present invention also provides a computing center electricity price calculation device, comprising:

[0046] The load time-series data acquisition module is used to collect load time-series data of the power supply entrance of the computing center within the current scoring period through a preset sampling frequency;

[0047] The feature extraction module is used to extract features from the load time series data to obtain load shape feature values;

[0048] The module for obtaining demand response interaction records and green energy consumption ratio is used to obtain demand response interaction records and green energy consumption ratio.

[0049] The grid friendliness comprehensive score calculation module is used to calculate the grid friendliness comprehensive score based on the load pattern characteristic value, the demand response interaction record and the green electricity consumption ratio;

[0050] The electricity price discount calculation module is used to calculate the electricity price discount for the computing center based on the comprehensive grid friendliness score.

[0051] The actual settlement electricity price calculation module is used to calculate the actual settlement electricity price based on the electricity price discount.

[0052] The present invention also provides an electronic device, the device comprising a processor and a memory:

[0053] The memory is used to store program code and transmit the program code to the processor;

[0054] The processor is used to execute the computing center electricity price calculation method as described above, according to the instructions in the program code.

[0055] The present invention also provides a computer-readable storage medium for storing program code for executing the power center electricity price calculation method as described in any of the preceding claims.

[0056] As can be seen from the above technical solution, the present invention has the following advantages: The present invention collects load time-series data of the power supply entrance of the computing center within the current scoring period through a preset sampling frequency; extracts features from the load time-series data to obtain load pattern characteristic values; acquires demand response interaction records and green electricity consumption ratios; calculates a comprehensive grid friendliness score based on the load pattern characteristic values, demand response interaction records, and green electricity consumption ratios; calculates the electricity price discount for the computing center based on the comprehensive grid friendliness score; and calculates the actual settlement electricity price based on the electricity price discount. The present invention directly links load pattern characteristics, demand response interaction records, and green electricity consumption ratios with economic incentives, forming a market-oriented mechanism that drives users to actively optimize their electricity consumption behavior based on price signals. This differs from existing unified pricing or static time-of-use pricing methods, and improves users' subjective initiative in optimizing their electricity consumption behavior. Attached Figure Description

[0057] 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.

[0058] Figure 1 A flowchart illustrating the steps of a method for calculating electricity prices at a computing center, as provided in an embodiment of the present invention;

[0059] Figure 2 A technical flowchart of a method for calculating electricity prices in a computing center, provided in an embodiment of the present invention;

[0060] Figure 3 A structural block diagram of a computing center electricity price calculation system provided in an embodiment of the present invention;

[0061] Figure 4 This is a schematic diagram of the internal processing flow of the load shape recognition module.

[0062] Figure 5 A schematic diagram of the data aggregation and scoring calculation process for the power grid friendliness scoring module;

[0063] Figure 6 A schematic diagram illustrating the collaborative workflow of an electricity price linkage calculation module and a computing power task scheduling optimization module, provided in an embodiment of the present invention;

[0064] Figure 7 This is a structural block diagram of a computing center electricity price calculation device provided in an embodiment of the present invention. Detailed Implementation

[0065] This invention provides a method, apparatus, electronic device, and storage medium for calculating electricity prices in computing centers, which addresses the technical problem that fixed time-of-use pricing cannot reflect the real-time operational status of the power grid, resulting in all users facing the same electricity price regardless of their actual behavior during the same period, leading to insufficient incentive accuracy.

[0066] 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.

[0067] Please see Figure 1 , Figure 1 A flowchart illustrating the steps of a method for calculating electricity prices at a computing center, as provided in an embodiment of the present invention.

[0068] This invention provides a method for calculating electricity prices in computing centers, which is applied to a computing center electricity price calculation system. The computing center electricity price calculation system includes a load data acquisition module, a load pattern recognition module, a power grid friendliness scoring module, and an electricity price linkage calculation module.

[0069] Before the electricity price calculation system in the computing center is put into operation, the key parameters of each module need to be initialized and configured. The configuration includes: the sampling frequency and window length of the scoring period in the load data acquisition module; the threshold values ​​of each feature value used for classifying load types in the load form identification module; the weight coefficients of the load form dimension, demand response coordination dimension, and green electricity consumption dimension in the grid friendliness scoring module, as well as the weight coefficients and scoring mapping rules of the sub-indicators within each dimension; and the benchmark score and discount cap of the scoring-discount mapping function in the electricity price linkage calculation module.

[0070] The method may specifically include the following steps:

[0071] Step 101: Collect load time-series data of the power supply entrance of the computing center within the current scoring period using a preset sampling frequency;

[0072] In this embodiment of the invention, after the system starts, the load data acquisition module runs continuously. The power metering sensor in the load data acquisition module can continuously measure the active power at the power supply entrance of the computing center according to a preset sampling frequency to obtain load time-series data. Each sampling point records the current timestamp and the corresponding active power value. Then, the collected raw data is time-aligned to ensure that the time interval between sampling points is uniform and consistent, and stored in the buffer in chronological order.

[0073] Step 102: Extract features from the load time series data to obtain load shape feature values;

[0074] To obtain the current load profile of a computing center, feature extraction can be performed on the collected load time-series data to obtain load profile feature values, which can then be used to analyze the current load profile of the computing center.

[0075] In one example, step 102 may include the following sub-steps:

[0076] S21, perform filtering operation on the load time series data to obtain effective load time series data;

[0077] S22, extract features from the effective load time series data to obtain load shape feature values.

[0078] In its implementation, after receiving the load time-series data for the current scoring period, the load pattern recognition module first performs filtering and noise reduction processing. This filtering process removes abnormal data points caused by factors such as measurement errors and transient impacts, retaining the effective load change trend that reflects the actual electricity consumption behavior of the computing center. The filtered data then proceeds to the feature extraction stage.

[0079] In one example, load shape characteristics include smoothness characteristics, peak-to-valley difference characteristics, and rate of change characteristics; the steps for extracting features from effective load time-series data to obtain load shape characteristics include:

[0080] S221, Extract the sampling point load values ​​of all sampling points within the current scoring period from the effective load time series data;

[0081] S222, calculate the first load difference between the load values ​​of all adjacent sampling points;

[0082] S223, Generate smoothness feature value based on the first load difference;

[0083] S224, obtain the maximum and minimum load values ​​among all sampled load values;

[0084] S225, calculate the average load of all sampled load values;

[0085] S226, calculate the second load difference between the maximum load value and the minimum load value;

[0086] S227, calculate the ratio of the second load difference to the average load to obtain the peak-valley difference characteristic value;

[0087] S228, calculate the average of the absolute values ​​of all first load differences to obtain the characteristic value of the rate of change.

[0088] In the specific implementation, the smoothness feature value is obtained by calculating the statistical dispersion of the first load difference of all adjacent sampling points in the current scoring period. Specifically, the standard deviation or root mean square value of the first load difference of all adjacent sampling points in the current scoring period can be obtained. The smaller the value, the smoother the load curve. The peak-valley difference feature value is obtained by calculating the second load difference between the maximum load value and the minimum load value, and then calculating the ratio of the second load difference to the average load. The rate of change feature value is obtained by calculating the average of the absolute values ​​of the first load difference between all adjacent sampling points.

[0089] Furthermore, after extracting the load morphology features, the following steps may also be included:

[0090] The current load profile of the computing center is determined based on the load profile characteristic values.

[0091] In one example, the steps for determining the current load profile of a computing center based on load profile characteristics include:

[0092] S1, when the smoothness characteristic value, peak-valley difference characteristic value and rate of change characteristic value are all lower than the corresponding preset low threshold, the current load mode of the computing center is determined to be stable.

[0093] S2, when the smoothness characteristic value, peak-valley difference characteristic value and rate of change characteristic value are all higher than the corresponding preset high threshold, the current load mode of the computing center is determined to be drastic fluctuation type.

[0094] S3. When the smoothness feature value, peak-valley difference feature value and rate of change feature value are all lower than the corresponding preset high threshold, and at least one of the smoothness feature value, peak-valley difference feature value and rate of change feature value is between the corresponding preset low threshold and preset high threshold, the current load mode of the computing center is determined to be slow change type.

[0095] In practice, the smoothness feature value, peak-to-valley difference feature value, and rate of change feature value can be compared with preset thresholds to determine whether the current load pattern is stable, slowly changing, or drastically fluctuating. The feature values ​​and pattern category labels are then output to the grid friendliness scoring module.

[0096] The specific morphology classification rules are as follows: Two threshold levels are set for the smoothness characteristic value, peak-to-valley difference characteristic value, and rate of change characteristic value: low thresholds (θ_s1, θ_p1, θ_r1) and high thresholds (θ_s2, θ_p2, θ_r2). When all three characteristic values ​​are below their respective low thresholds (i.e., smoothness characteristic value < θ_s1, peak-to-valley difference characteristic value < θ_p1, and rate of change characteristic value < θ_r1), it is classified as a stable type, indicating minimal load curve fluctuations, weak peak-to-valley differences, and slow change rate, resulting in minimal impact on the power grid. When any one of the three characteristic values ​​exceeds its corresponding high threshold (i.e., smoothness characteristic value > θ_s2, peak-to-valley difference characteristic value > θ_p2, or rate of change characteristic value > θ_r2), it is classified as a violently fluctuating type, indicating significant load curve fluctuations, large peak-to-valley differences, or rapid load changes, resulting in a significant impact on power grid operation. When none of the three characteristic values ​​trigger the criteria for a drastic fluctuation type, and at least one characteristic value is between its low and high thresholds, it is judged as a slow-changing type, indicating that the load curve has a certain degree of fluctuation but is generally controllable.

[0097] Different load patterns reflect the varying degrees of impact of computing center electricity consumption on the power grid, providing a basis for subsequent grid friendliness scoring.

[0098] Step 103: Obtain the demand response interaction records and the green electricity consumption ratio;

[0099] Step 104: Calculate the comprehensive grid friendliness score based on load pattern characteristics, demand response interaction records, and green energy consumption ratio;

[0100] The grid-friendliness comprehensive score reflects the grid-friendly nature of a computing center's electricity consumption behavior. This invention decomposes the evaluation of a computing center's grid-friendliness into three relatively independent yet complementary evaluation dimensions, and integrates the scores of these three dimensions into a unified comprehensive score through weighting coefficients. The load pattern dimension assesses the inherent characteristics of user electricity consumption behavior; the demand response coordination dimension assesses the initiative and reliability of user interaction with the grid; and the green energy consumption dimension assesses the user's contribution to clean energy consumption. These three dimensions measure the user's value to the grid from different perspectives, and the comprehensive score is more comprehensive and impartial than any single-dimensional evaluation. By calculating the grid-friendliness comprehensive score of the computing center, electricity price regulation can be implemented based on this score, incentivizing computing centers to proactively optimize their load patterns and cooperate with grid dispatch.

[0101] In one example, step 104 may include the following sub-steps:

[0102] S41, converts the smoothness feature values ​​into smoothness scores;

[0103] S42 converts the peak-valley difference characteristic value into the peak-valley difference fraction;

[0104] S43, converts the characteristic value of the rate of change into a fraction of the rate of change;

[0105] S44, the load morphology dimension score is obtained by weighted summation of the smoothness score, peak-to-valley difference score and rate of change score;

[0106] In its implementation, the grid friendliness scoring module receives three types of input data simultaneously: load pattern characteristic values, demand response interaction records, and green electricity consumption ratio, and then begins the scoring calculation.

[0107] In the load pattern dimension, the smoothness feature value, peak-to-valley difference feature value, and rate of change feature value are converted into scores through their respective mapping functions. The specific mapping functions are as follows: For the smoothness feature value V_s, its reasonable range is set to [0, V_s_max], where V_s_max is the preset upper limit of the smoothness feature value. Then, the smoothness score F_s = S_max_dim × (1 - V_s / V_s_max). When V_s ≥ V_s_max, F_s is 0, and when V_s ≤ 0, F_s is S_max_dim, where S_max_dim is the full score of a single indicator within the dimension.

[0108] For the peak-valley difference feature value V_p, its reasonable range is set to [0, V_p_max]. Then the peak-valley difference score F_p = S_max_dim × (1 - V_p / V_p_max). The boundary processing rules are the same as those for the smoothness score.

[0109] For the rate of change eigenvalue V_r, its reasonable range is set to [0, V_r_max]. Then, the rate of change score F_r = S_max_dim × (1 - V_r / V_r_max), with the same boundary handling rules. All three mapping functions are linearly decreasing functions, ensuring that the smaller the eigenvalue, the higher the corresponding score.

[0110] The load pattern dimension score is obtained by weighting and summing the three feature scores using sub-weights within the dimension: F_load = w_s × F_s + w_p × F_p + w_r × F_r, where w_s, w_p, and w_r are the sub-weight coefficients for smoothness, peak-to-valley difference, and rate of change, respectively, and w_s + w_p + w_r = 1. Smaller feature values ​​(smoother load, smaller peak-to-valley difference, slower change) correspond to higher scores. The load pattern dimension score is obtained by weighting and summing the three feature scores using sub-weights within the dimension.

[0111] S45, Obtain the demand response interaction records within the target scoring period; the target scoring period includes the current scoring period and several scoring periods preceding the current scoring period;

[0112] S46, calculate the response rate, response timeliness rate and response depth compliance rate based on the demand response interaction records;

[0113] S47, map the response rate to the response rate score, the response timeliness rate to the response timeliness rate score, and the response depth compliance rate to the response depth compliance rate score.

[0114] S48, the response rate score, response timeliness score and response depth compliance rate score are weighted and summed to obtain the demand response dimension score;

[0115] In the demand response coordination dimension, the system queries demand response interaction records for the current period and several previous scoring periods, calculates the response rate, response timeliness rate, and response depth compliance rate. These three indicators are each mapped to a score, and then weighted according to their sub-weights to obtain the demand response dimension score. The specific mapping process is as follows: Response rate R_rate is defined as the ratio of the actual number of responses N_resp to the total number of instructions issued N_cmd, i.e., R_rate = N_resp / N_cmd; Response timeliness rate R_timely is defined as the ratio of the number of responses completed within the required time frame N_timely to the actual number of responses N_resp, i.e., R_timely = N_timely / N_resp; Response depth compliance rate R_depth is defined as the ratio of the number of times the actual load adjustment reaches the required range N_depth to the actual number of responses N_resp, i.e., R_depth = N_depth / N_resp. The values ​​of the above three indicators are all in the range of [0, 1]. Multiplying them by the single full score value S_max_dr of the demand response dimension yields the corresponding scores: response rate score F_rate = S_max_dr × R_rate, response timeliness score F_timely = S_max_dr × R_timely, and response depth compliance score F_depth = S_max_dr × R_depth. The three scores are weighted and summed according to their sub-weights to obtain the demand response dimension score F_dr = w_rate × F_rate + w_timely × F_timely + w_depth × F_depth, where w_rate, w_timely, and w_depth are the sub-weight coefficients of response rate, response timeliness, and response depth compliance, respectively, and w_rate + w_timely + w_depth = 1. When the power grid does not issue any demand response instructions during the scoring period, the demand response dimension score is taken as the weighted average of the previous several valid scoring periods as the substitute score for the current period.

[0116] S49, mapping the green electricity consumption ratio to a green electricity consumption dimension score;

[0117] The green electricity consumption ratio is the ratio of the amount of green electricity consumed by the computing center to its total electricity consumption during the current scoring period.

[0118] The process of mapping the green energy consumption ratio to a green energy consumption dimension is as follows:

[0119] Let E_green be the amount of green electricity consumed by the computing center in the current scoring period, and E_total be the total electricity consumption. Then, the green electricity consumption ratio G_ratio = E_green / E_total, and the value of G_ratio ranges from [0, 1]. The mapping of the green electricity consumption dimension score F_green adopts a piecewise linear function: a baseline green electricity consumption ratio is set. And the target ratio G_target, when When, F_green = 0; when hour, When G_ratio > G_target, F_green = S_max_green, where S_max_green is the maximum score for the green energy consumption dimension. This mapping function ensures that no points are awarded when the green energy consumption ratio is below the benchmark value, and full marks are awarded when the target value is reached, increasing linearly between the benchmark and the target value.

[0120] S410 uses preset dimension weights to weight and sum the scores of load form dimension, demand response dimension, and green electricity consumption dimension to obtain the comprehensive grid friendliness score.

[0121] The overall grid friendliness score is calculated by weighting and summing the scores from the load profile, demand response, and green energy consumption dimensions according to preset dimension weighting coefficients. The overall score falls within a set range. When the calculated overall score exceeds the set range, the system performs a truncation process: if the overall score is below the lower limit S_min, it is forcibly set to S_min; if the overall score is above the upper limit S_max, it is forcibly set to S_max. That is, the final overall score S_final = max(S_min, min(S_max, S_raw)), where S_raw is the original overall score obtained by weighted summation. Simultaneously, when truncation occurs, the system records the truncation event and the corresponding original score in the historical database of the closed-loop feedback control module for subsequent analysis to determine whether adjustments to the dimension weighting coefficients and mapping function parameters are needed.

[0122] This invention extracts morphological features from three dimensions: smoothness of the load curve, peak-to-valley difference, and rate of change. It combines demand response coordination and green energy consumption ratio to conduct a multi-dimensional comprehensive evaluation, thereby achieving a precise quantitative evaluation of the electricity consumption behavior of computing centers.

[0123] Step 105: Calculate the electricity price discount for the computing center based on the comprehensive grid friendliness score;

[0124] Step 106: Calculate the actual settlement electricity price based on the electricity price discount.

[0125] In this embodiment of the invention, after receiving the comprehensive grid friendliness score, the electricity price linkage calculation module substitutes it into a preset score-discount mapping function for calculation. When the comprehensive score is lower than the benchmark score, the discount coefficient is zero, and the computing center settles the bill at the standard electricity price; when the comprehensive score is higher than the benchmark score, the discount coefficient is calculated according to the score-discount mapping function, and the higher the score, the greater the discount, but the discount coefficient has an upper limit to ensure a basic level of electricity revenue.

[0126] The calculated discount factor is applied to the standard electricity price for the current period to generate the actual settlement price.

[0127] Specifically, when the overall score is higher than the benchmark score, the score-discount mapping function is defined as follows:

[0128] Let S be the overall score for grid friendliness, and let the baseline score be... The maximum value is S_max, and the upper limit of the discount coefficient is d_max. The calculation rule for the discount coefficient d is: when... When d = 0; when hour, Therefore, the discount factor d increases linearly with the rating S within the interval [0, d_max]. The actual settlement electricity price P_actual = P_standard × (1 - d), where P_standard is the standard electricity price for the current period. For example, when When the score is set to 60, S_max is set to 100, and d_max is set to 0.15, the discount coefficient d = 0.15 × (80 - 60) / (100 - 60) = 0.075 for a computing center with a comprehensive score of 80, which means it enjoys a 7.5% discount on electricity prices.

[0129] This invention establishes a direct mapping between ratings and electricity price discounts, enabling the real-time conversion of electricity consumption behavior evaluation results from computing centers into differentiated electricity price incentive signals. This mechanism directly links user behavior evaluations with economic incentives, forming a market-based mechanism that drives users to proactively optimize their electricity consumption behavior through price signals.

[0130] In this embodiment of the invention, after the step of calculating the actual settlement electricity price based on the electricity price discount, the method further includes:

[0131] Improvement dimensions are determined based on scores for load profile, demand response, and green energy consumption; computing power tasks are then scheduled based on these improvement dimensions.

[0132] In the specific implementation, the electricity price linkage calculation module generates two output signals: the first signal contains the actual settlement electricity price and is sent to the electricity billing and settlement system for electricity billing calculation in the current scoring period; the second signal contains complete information on the grid friendliness comprehensive score, the scores of each dimension, the discount coefficient and the actual settlement electricity price, and is sent to the task scheduling and optimization module of the computing center through the electricity price signal transmission interface module.

[0133] After receiving the electricity price signal, the computing power task scheduling optimization module analyzes the scoring details and electricity price discount level of the current scoring period, identifying the dimensions with lower scores in the current scoring as priority improvement directions. The module then analyzes the queue of tasks to be executed, identifying rigid tasks (real-time computing tasks that must be completed within a specified time and cannot be delayed) and flexible tasks (batch processing tasks, offline training tasks, etc., with relaxed completion time requirements and can be delayed within a certain range). For flexible tasks, the scheduling module, based on the improvement needs of the load pattern dimension, migrates some flexible tasks from peak load periods to off-peak load periods to reduce the peak-to-valley difference and improve load smoothness. When the power grid issues a demand response command in the current period or the upcoming period, the scheduling module immediately suspends or reduces the computing resource allocation of non-critical flexible tasks, releasing corresponding power load margins to cooperate with the power grid's peak shaving requirements. For improvements in the green electricity consumption dimension, the scheduling module, combined with green electricity supply forecast information, prioritizes the start of high-energy-consuming large-scale training tasks during periods with abundant green electricity supply. After the scheduling strategy adjustment is completed, the computing center operates according to the new task orchestration scheme, and its actual electricity load pattern changes accordingly.

[0134] This invention integrates grid friendliness scores and electricity price discount signals into the internal scheduling system of the computing center. With maximizing the friendliness score as one of the optimization objectives, it automatically adjusts the timing of elastic tasks, load reduction during demand response, and priority scheduling of tasks during green electricity periods. This achieves automated transmission of grid scheduling needs to computing task scheduling, eliminating the delay caused by manual intervention.

[0135] Furthermore, after task scheduling optimization, the computing center generates new electricity load. The load data acquisition module continuously collects new load data. Once the data window for the next scoring cycle is full, the system begins a new round of pattern recognition, score calculation, electricity price linkage, and scheduling optimization processes from step three. The closed-loop feedback control module records complete data for each scoring cycle at the end of each cycle, including load pattern characteristic values, scores for each dimension, comprehensive score, electricity price discount coefficient, task scheduling adjustments, and their actual improvement effect on load pattern. Through cumulative analysis of historical cycle data, system administrators can adjust the scoring weight coefficients, mapping function parameters, and pattern classification thresholds based on actual results, continuously improving the system's incentive effect and scoring accuracy. Under closed-loop incentives, the computing center continuously optimizes its electricity consumption behavior, and the power grid guides the computing center's load pattern towards a direction favorable to power grid operation through continuous price signals, achieving long-term collaborative optimization between the power grid and the computing center.

[0136] This invention directly links load characteristics, demand response interaction records, and green energy consumption ratios with economic incentives, forming a market-based mechanism that drives users to proactively optimize their electricity consumption behavior based on price signals. This differs from existing uniform pricing or static time-of-use pricing methods, enhancing users' initiative in optimizing their electricity consumption behavior and helping to improve the flexibility of computing power scheduling.

[0137] Please see Figure 2 , Figure 2 This invention provides a technical flowchart for a method of calculating electricity prices for computing centers. The overall technical framework includes five core components: load data acquisition and pattern recognition, multi-dimensional grid friendliness scoring, calculation linking scoring results with electricity prices, task scheduling optimization driven by electricity prices, and closed-loop feedback between scoring and scheduling. These components are closely interconnected, forming a complete closed-loop link from data acquisition to behavior optimization. Starting with load data acquisition and pattern recognition at the computing center, the process involves multi-dimensional grid friendliness scoring, calculation linking scoring with electricity prices, and task scheduling optimization driven by electricity prices. The optimized new load data is then re-input into the pattern recognition module via closed-loop feedback, forming a complete cyclical link from data acquisition to behavior optimization. This closed-loop structure ensures that the computing center continuously improves its electricity consumption behavior under economic incentives, while the power grid continuously guides the load pattern of the computing center through dynamic scoring and electricity price linkage mechanisms.

[0138] In the load data acquisition and morphology recognition stage, this embodiment of the invention deploys high-frequency power metering sensors at the power supply entrance of the computing center to collect time-series data of the computing center's electricity load within a preset current scoring period window. After preprocessing, the collected load data undergoes morphological feature extraction of the load curve. The main components of the morphological feature extraction include: a load curve smoothness index, used to measure the severity of load fluctuations over time; a load peak-to-valley difference index, used to measure the difference between the maximum and minimum loads; and a load change rate index, used to measure the rate of load change between adjacent time points. Through comprehensive analysis of these features, the load morphology of the computing center is categorized into several preset morphology types, such as stable, slowly changing, and drastically fluctuating. Different load morphology types reflect different degrees of impact of the computing center's electricity consumption behavior on the power grid, providing a basic input for subsequent grid friendliness scoring.

[0139] In the multi-dimensional scoring of grid friendliness, this embodiment of the invention constructs a comprehensive scoring model covering multiple evaluation dimensions. The scoring dimensions include at least the following three aspects: load pattern dimension, which scores the smoothness, peak-to-valley ratio, and rate of change of the computing center's load curve based on the aforementioned load pattern identification results; demand response dimension, which scores the computing center based on indicators such as the number of times it responds to grid demand response commands, response speed, response depth, and response duration within a historical period; the more times it responds, the faster its response speed, and the closer its response magnitude is to the command requirements, the higher its score; and green energy consumption dimension, which scores the computing center based on the proportion of clean energy actually consumed to its total electricity consumption; the higher the proportion of green energy consumption, the higher its score. The scores of the above dimensions are weighted and combined according to preset weighting coefficients to calculate the comprehensive grid friendliness score of the computing center for the current period. The scoring results continuously change within a preset scoring interval, reflecting the real-time quantitative level of the computing center's electricity consumption behavior's grid friendliness.

[0140] In the calculation of the linkage between the scoring results and electricity prices, this embodiment of the invention establishes a mapping relationship between the grid friendliness score and the electricity price discount. Specifically, when the comprehensive grid friendliness score of the computing center is at a high level, the system automatically generates a larger electricity price discount, making the actual settlement price of the computing center lower than the standard electricity price; when the score is at a medium level, the electricity price discount is reduced accordingly; when the score is below a preset threshold, the computing center settles at the standard electricity price and does not enjoy the discount. The calculation of the electricity price discount follows a monotonically increasing principle that the higher the score, the larger the discount, ensuring that the incentive direction of the price signal is consistent with the direction of user behavior expected by the grid. The result of the linkage calculation is output in real time in the form of an electricity price signal, which is sent to the power billing system for actual electricity bill settlement on the one hand, and to the internal scheduling system of the computing center on the other hand to guide task scheduling decisions.

[0141] In the task scheduling optimization process driven by electricity price signals, the internal scheduling system of the computing center, upon receiving the electricity price linkage signal, incorporates the current grid friendliness score and corresponding electricity price discount level as one of the optimization objectives for task scheduling in the scheduling decision. Under the premise of meeting the service quality constraints of computing tasks, the scheduling system automatically adjusts the allocation strategy of computing tasks in the time dimension. Specific adjustments include: migrating deferable batch processing tasks, model training tasks, and other elastic loads to off-peak periods to smooth the load curve; proactively reducing the allocation of computing resources for non-critical tasks to cooperate with peak shaving when the grid issues demand response instructions; and prioritizing the scheduling of computing tasks during periods of abundant green electricity supply to increase the proportion of green electricity consumption. Through the automatic adjustment of the above task orchestration strategies, the load pattern of the computing center evolves towards a smoother and more grid-coordinated direction, thereby obtaining a higher grid friendliness score and a larger electricity price discount in the next scoring cycle.

[0142] In the closed-loop feedback loop of scoring and scheduling, the new load pattern data generated by the computing center after task scheduling optimization is re-collected and input into the load pattern recognition module, initiating a new round of grid friendliness score calculation. This closed-loop mechanism ensures a continuous positive iterative cycle between the computing center's electricity consumption behavior, grid friendliness score, and electricity price discount. The computing center continuously optimizes its load pattern to pursue lower electricity costs, while the grid continuously guides the computing center's electricity consumption behavior towards a direction conducive to the safe and stable operation of the grid through price signals, ultimately achieving coordinated optimization between grid scheduling and computing power scheduling.

[0143] Please see Figure 3 , Figure 3This invention provides a structural block diagram of a computing center electricity price calculation system. It includes seven functional modules: a load data acquisition module, a load pattern recognition module, a grid friendliness scoring module, an electricity price linkage calculation module, an electricity price signal transmission interface module, a computing power task scheduling optimization module, and a closed-loop feedback control module. These modules are connected via data communication links, collaboratively completing the entire process from load data perception to electricity price linkage output and task scheduling optimization. The load data acquisition module is located at the beginning of the link, sequentially passing through the load pattern recognition module, grid friendliness scoring module, electricity price linkage calculation module, and electricity price signal transmission interface module to the computing power task scheduling optimization module. Finally, it returns to the load data acquisition module through the closed-loop feedback control module, forming a complete circular data flow path. Each module has clearly defined responsibilities and interfaces; the output of the upstream module serves as the input of the downstream module, ensuring automatic operation of the entire link from load perception to behavior optimization.

[0144] The load data acquisition module is deployed at the power supply input side of the computing center. It is responsible for continuously collecting electrical parameters such as active power, reactive power, current, and voltage at a preset sampling frequency, generating time-series load data records. This module includes power metering sensors, a data buffer unit, and a data transmission unit. The power metering sensors perform real-time measurement of electrical parameters; the data buffer unit temporarily stores and aligns the collected raw data; and the data transmission unit uploads the pre-processed load time-series data to the load pattern recognition module.

[0145] The load pattern recognition module receives load time-series data uploaded by the load data acquisition module. After filtering, denoising, and standardizing the data, it extracts multi-dimensional pattern features of the load curve. The core function of this module is to calculate the smoothness feature value, peak-to-valley difference feature value, and rate of change feature value of the load curve within the current scoring period, and classify the current load pattern into a preset pattern category based on these feature values. The smoothness feature value reflects the fluctuation amplitude of the load curve and is obtained by calculating the statistical dispersion of the load difference between adjacent sampling points; the peak-to-valley difference feature value is the ratio of the difference between the maximum and minimum load values ​​within the scoring period to the average load value; the rate of change feature value is the average of the load change amplitudes between adjacent sampling points. The load pattern recognition module outputs the extracted feature values ​​and pattern category labels to the power grid friendliness scoring module.

[0146] like Figure 4 As shown, Figure 4This is a schematic diagram of the internal processing flow of the load pattern recognition module. After filtering and denoising, the raw load time-series data enters three parallel processing paths: smoothness feature extraction, peak-valley difference feature extraction, and rate of change feature extraction. These three feature values ​​are ultimately merged into the pattern category determination stage, outputting the load pattern category label. The load pattern recognition module transforms continuous time-series data into discrete pattern features and discrimination information, providing structured input for subsequent scoring calculations.

[0147] The grid friendliness scoring module is the core calculation module of this system, responsible for calculating the current comprehensive grid friendliness score of the computing center based on input data from multiple evaluation dimensions. This module receives three types of input data: load pattern characteristic values ​​and category labels from the load pattern identification module, demand response coordination records from the grid dispatching system, and green electricity consumption ratio data from the power trading platform or green electricity certification system. The scoring module internally consists of dimensional scoring sub-units and a comprehensive scoring sub-unit. The dimensional scoring sub-units independently score the load pattern dimension, demand response coordination dimension, and green electricity consumption dimension, with each dimension's score value continuously taking values ​​within a preset range. The comprehensive scoring sub-units weight and sum the scores from each dimension according to preset weighting coefficients to obtain the comprehensive grid friendliness score. The comprehensive score value is output to the electricity price linkage calculation module.

[0148] like Figure 5 As shown, Figure 5 This diagram illustrates the data aggregation and scoring calculation process for the grid friendliness scoring module. Three types of input data from different sources—load pattern characteristic values, demand response matching interaction records, and green energy consumption ratio data—simultaneously enter the dimensional scoring sub-units for independent scoring. The scores from each dimension are then aggregated into the comprehensive scoring sub-unit for weighted summation, ultimately outputting a comprehensive grid friendliness score. This multi-dimensional input ensures that the scoring results comprehensively reflect the various characteristics of the computing center's electricity consumption behavior.

[0149] The electricity price linkage calculation module receives the comprehensive grid friendliness score and calculates the electricity price discount coefficient corresponding to the current scoring period according to the pre-set mapping rules between the score and the electricity price discount. The mapping rules ensure that the higher the score, the larger the discount coefficient. The discount coefficient is applied to the standard electricity price to generate the actual settlement price. This module simultaneously outputs two signals: one is the actual settlement price signal, which is sent to the electricity billing and settlement system for cost accounting; the other is an optimization guidance signal containing the grid friendliness score and electricity price discount information, which is sent to the computing power task scheduling and optimization module.

[0150] The electricity price signal transmission interface module is responsible for establishing a standardized data communication channel between the grid-side system and the internal scheduling system of the computing center. This module encapsulates and encrypts the signals output by the electricity price linkage calculation module before transmitting them to the receiving port on the computing center side via the communication network. Simultaneously, this module also receives task scheduling status information and load forecast data transmitted back from the computing center side, enabling bidirectional data interaction.

[0151] The computing power task scheduling optimization module, deployed within the computing center, is a key module connecting grid-side price signals with the actual electricity consumption behavior of the computing center. After receiving grid friendliness scores and electricity price discount information, the module transforms them into optimization constraints for task scheduling. The core logic of this module is to select a task orchestration scheme that maximizes the grid friendliness score in the next scoring cycle (thus minimizing electricity costs) under constraints of task latency and service quality. Specifically, the module reorders delayable tasks based on the priority, delayability, and expected computing resource requirements of each task in the current task queue, making the expected load curve of the computing center smoother in the next scoring cycle. When the grid issues a demand response command, the scheduling module immediately identifies interruptible or load-reducing tasks and executes load reduction to improve the demand response coordination score. Regarding green electricity consumption, the scheduling module combines green electricity supply forecast data to prioritize high computing power demand tasks during periods of abundant green electricity supply.

[0152] like Figure 6 As shown, Figure 6 This is a schematic diagram illustrating the collaborative workflow of an electricity price linkage calculation module and a computing power task scheduling optimization module, provided in an embodiment of the present invention. After the grid friendliness comprehensive score is entered into the score-discount mapping calculation, it outputs the actual settlement electricity price for cost accounting and generates an optimization guidance signal that is transmitted to the scheduling optimization module. The task scheduling optimization module executes three optimization strategies based on this signal: task time-series rearrangement to smooth the load curve, load reduction during demand response to improve the compliance score, and scheduling high-computing-power tasks during periods of abundant green electricity to increase the proportion of green electricity consumption. These three strategies work together to evolve the load pattern of the computing power center towards grid friendliness.

[0153] The closed-loop feedback control module is responsible for re-sending the actual load data generated by the computing center after task scheduling optimization into the load data acquisition module, triggering the start of the next scoring cycle. Simultaneously, this module records the scoring results, electricity price discounts, task scheduling adjustment strategies, and their effects for each scoring cycle, forming a historical database. The data in the historical database can be used to optimize scoring weight coefficients and adjust electricity price discount mapping rules, achieving continuous improvement in the overall system performance.

[0154] Please see Figure 7 , Figure 7This is a structural block diagram of a computing center electricity price calculation device provided in an embodiment of the present invention.

[0155] This invention provides a computing center electricity price calculation device, comprising:

[0156] The load time series data acquisition module 701 is used to acquire the load time series data of the power supply entrance of the computing center within the current scoring period through a preset sampling frequency;

[0157] The feature extraction module 702 is used to extract features from load time series data to obtain load shape feature values;

[0158] The demand response interaction record and green energy consumption ratio acquisition module 703 is used to acquire the demand response interaction record and green energy consumption ratio.

[0159] The grid friendliness comprehensive score calculation module 704 is used to calculate the grid friendliness comprehensive score based on load pattern characteristics, demand response interaction records and green electricity consumption ratio.

[0160] The electricity price discount calculation module 705 is used to calculate the electricity price discount for the computing center based on the comprehensive grid friendliness score.

[0161] The actual settlement electricity price calculation module 706 is used to calculate the actual settlement electricity price based on the electricity price discount.

[0162] In this embodiment of the invention, the feature extraction module 702 includes:

[0163] The filtering submodule is used to perform filtering operations on the load time series data to obtain the effective load time series data.

[0164] The feature extraction submodule is used to extract features from the effective load time series data to obtain load shape feature values.

[0165] In this embodiment of the invention, the load morphology characteristic values ​​include smoothness characteristic values, peak-to-valley difference characteristic values, and rate of change characteristic values; the feature extraction submodule includes:

[0166] The sampling point load value extraction unit is used to extract the sampling point load values ​​of all sampling points within the current scoring period from the effective load time series data;

[0167] The first load difference calculation unit is used to calculate the first load difference between the sample load values ​​of all adjacent sample points;

[0168] A smoothness feature value generation unit is used to generate smoothness feature values ​​based on the first load difference.

[0169] The maximum and minimum load value acquisition unit is used to acquire the maximum and minimum load values ​​among all sampled load values;

[0170] The average load calculation unit is used to calculate the average load of all sampled load values.

[0171] The second load difference calculation unit is used to calculate the second load difference between the maximum load value and the minimum load value;

[0172] The peak-valley difference characteristic value calculation unit is used to calculate the ratio of the second load difference to the average load to obtain the peak-valley difference characteristic value;

[0173] The rate of change characteristic value calculation unit is used to calculate the average of the absolute values ​​of all first load differences to obtain the rate of change characteristic value.

[0174] In this embodiment of the invention, it further includes:

[0175] The current load condition determination module is used to determine the current load condition of the computing center based on load condition characteristic values.

[0176] In this embodiment of the invention, the current load condition determination module includes:

[0177] The stability determination submodule is used to determine that the current load state of the computing center is stable when the smoothness feature value, peak-valley difference feature value and rate of change feature value are all lower than the corresponding preset low threshold.

[0178] The violent fluctuation type determination submodule is used to determine that the current load mode of the computing center is violent fluctuation type when the smoothness feature value, peak-valley difference feature value and change rate feature value are all higher than the corresponding preset high threshold.

[0179] The gradual change type determination submodule is used to determine that the current load mode of the computing center is gradually changing when the smoothness feature value, peak-valley difference feature value and rate of change feature value are all lower than the corresponding preset high threshold, and at least one of the smoothness feature value, peak-valley difference feature value and rate of change feature value is between the corresponding preset low threshold and preset high threshold.

[0180] In this embodiment of the invention, the power grid friendliness comprehensive score calculation module 704 includes:

[0181] The smoothness score conversion submodule is used to convert smoothness feature values ​​into smoothness scores;

[0182] The peak-valley difference fraction conversion submodule is used to convert peak-valley difference feature values ​​into peak-valley difference fractions;

[0183] The rate of change fraction conversion submodule is used to convert rate of change feature values ​​into rate of change fractions;

[0184] The load form dimension score calculation submodule is used to calculate the load form dimension score by weighted summation of the smoothness score, peak-to-valley difference score and rate of change score.

[0185] The Demand Response Interaction Record Acquisition Submodule is used to acquire demand response interaction records within the target scoring period; the target scoring period includes the current scoring period and several scoring periods preceding the current scoring period.

[0186] The response rate, response timeliness rate, and response depth compliance rate calculation submodule is used to calculate the response rate, response timeliness rate, and response depth compliance rate based on the demand response interaction records.

[0187] The score mapping submodule is used to map response rate to response rate score, response timeliness rate to response timeliness score, and response depth compliance rate to response depth compliance rate score.

[0188] The demand response dimension score calculation submodule is used to calculate the weighted sum of the response rate score, response timeliness score, and response depth compliance rate score to obtain the demand response dimension score.

[0189] The green energy consumption dimension score mapping submodule is used to map the green energy consumption ratio to the green energy consumption dimension score;

[0190] The grid friendliness comprehensive score calculation submodule is used to perform weighted summation of the load form dimension score, demand response dimension score, and green energy consumption dimension score using preset dimension weights to obtain the grid friendliness comprehensive score.

[0191] In this embodiment of the invention, it further includes:

[0192] The improvement dimension determination module is used to determine the improvement dimensions based on the load pattern dimension score, demand response dimension score, and green energy consumption dimension score.

[0193] The computing power task scheduling module is used to schedule computing power tasks based on the improvement dimensions.

[0194] This invention also provides an electronic device, which includes a processor and a memory:

[0195] The memory is used to store program code and transfer the program code to the processor;

[0196] The processor is used to execute the power center electricity price calculation method of this invention according to the instructions in the program code.

[0197] This invention also provides a computer-readable storage medium for storing program code for executing the computing center electricity price calculation method of this invention.

[0198] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0199] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0200] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0201] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0202] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0203] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0204] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0205] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0206] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0207] The above-described 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 method for calculating electricity prices for computing power centers, characterized in that, include: By using a preset sampling frequency, load time-series data of the power supply entrance of the computing center within the current scoring period is collected; Feature extraction is performed on the load time series data to obtain load pattern feature values; Obtain demand response interaction records and green energy consumption ratio; Based on the load pattern characteristic value, the demand response interaction record, and the green energy consumption ratio, calculate the comprehensive grid friendliness score; The electricity price discount for the computing center is calculated based on the comprehensive grid friendliness score. The actual settlement price is calculated based on the aforementioned electricity price discount.

2. The method according to claim 1, characterized in that, The step of extracting features from the load time-series data to obtain load pattern feature values ​​includes: The load time series data is filtered to obtain the effective load time series data; Feature extraction is performed on the effective load time series data to obtain load shape feature values.

3. The method according to claim 2, characterized in that, The load shape characteristic values ​​include smoothness characteristic values, peak-to-valley difference characteristic values, and rate of change characteristic values; the step of extracting features from the effective load time series data to obtain load shape characteristic values ​​includes: Extract the sampling point load values ​​of all sampling points within the current scoring period from the effective load time series data; Calculate the first load difference between the load values ​​of all adjacent sampling points; Generate smoothness feature values ​​based on the first load difference; Obtain the maximum and minimum load values ​​from all sampled load values; Calculate the average load of all sampled load values; Calculate the second load difference between the maximum load value and the minimum load value; Calculate the ratio of the second load difference to the average load to obtain the peak-valley difference characteristic value; Calculate the average of the absolute values ​​of all the first load differences to obtain the rate of change characteristic value.

4. The method according to claim 3, characterized in that, After the step of extracting features from the load time-series data to obtain load pattern feature values, the method further includes: The current load profile of the computing center is determined based on the load profile characteristic values.

5. The method according to claim 4, characterized in that, The step of determining the current load profile of the computing center based on the load profile characteristic value includes: When the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value are all lower than the corresponding preset low threshold, the current load state of the computing center is determined to be stable. When the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value are all higher than the corresponding preset high threshold, the current load mode of the computing center is determined to be drastic fluctuation type. When the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value are all lower than the corresponding preset high threshold, and at least one of the smoothness feature value, the peak-to-valley difference feature value, and the rate of change feature value is between the corresponding preset low threshold and the preset high threshold, the current load mode of the computing center is determined to be a slow-changing type.

6. The method according to claim 3, characterized in that, The step of calculating the comprehensive grid friendliness score based on the load pattern characteristic value, the demand response interaction record, and the green energy consumption ratio includes: The smoothness feature values ​​are converted into smoothness scores; The peak-valley difference characteristic value is converted into a peak-valley difference fraction; Convert the rate of change characteristic value into a rate of change fraction; The load morphology dimension score is obtained by weighted summation of the smoothness score, the peak-to-valley difference score, and the rate of change score. Obtain the demand response interaction records within the target scoring period; the target scoring period includes the current scoring period and several scoring periods preceding the current scoring period; Calculate the response rate, response timeliness rate, and response depth compliance rate based on the aforementioned demand response interaction records; The response rate is mapped to a response rate score, the response timeliness rate is mapped to a response timeliness rate score, and the response depth compliance rate is mapped to a response depth compliance rate score. The response rate score, the response timeliness score, and the response depth compliance rate score are weighted and summed to obtain the demand response dimension score. The green energy consumption ratio is mapped to a green energy consumption dimension score; The scores for the load profile dimension, the demand response dimension, and the green energy consumption dimension are weighted and summed using preset dimension weights to obtain a comprehensive grid friendliness score.

7. The method according to claim 6, characterized in that, After the step of calculating the actual settlement electricity price based on the electricity price discount, the method further includes: Improvement dimensions are determined based on the load pattern dimension score, the demand response dimension score, and the green energy consumption dimension score; Computing task scheduling is performed based on the aforementioned improvement dimensions.

8. A computing center electricity price calculation device, characterized in that, include: The load time-series data acquisition module is used to collect load time-series data of the power supply entrance of the computing center within the current scoring period through a preset sampling frequency; The feature extraction module is used to extract features from the load time series data to obtain load shape feature values; The module for obtaining demand response interaction records and green energy consumption ratio is used to obtain demand response interaction records and green energy consumption ratio. The grid friendliness comprehensive score calculation module is used to calculate the grid friendliness comprehensive score based on the load pattern characteristic value, the demand response interaction record and the green electricity consumption ratio; The electricity price discount calculation module is used to calculate the electricity price discount for the computing center based on the comprehensive grid friendliness score. The actual settlement electricity price calculation module is used to calculate the actual settlement electricity price based on the electricity price discount.

9. An electronic 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 used to execute the electricity price calculation method for computing centers according to any one of claims 1-7, based on 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 power center electricity price calculation method according to any one of claims 1-7.

Images