A method and system for intelligent identification of power load driving factors

Abstract

This application provides an intelligent identification method and system for power load driving factors, relating to the field of power load management technology. The method includes: real-time monitoring and acquisition of power load data; when the load change in the power load data exceeds a preset load change threshold, acquiring the power load change characteristics of the current load change; performing load change matching based on the power load change characteristics to determine multiple historical similar load change scenarios, each with a combination of driving factors; calculating the matching degree between each historical similar load change scenario and the current load change scenario, and selecting the driving factor combination corresponding to the historical similar load change scenario with the highest matching degree to determine the load change driving factor of the current load change scenario. This solves the technical problems of existing load driving factor identification methods, such as difficulty in handling complex load changes with multiple coupled factors, poor adaptability, and decreased identification accuracy when data is insufficient.

Description

Technical Field

[0001] This invention relates to the field of power load management, and in particular to an intelligent identification method and system for power load driving factors. Background Technology

[0002] Changes in power load directly affect the stability and economy of power grid operation. Accurately identifying the driving factors that cause load changes is crucial for power system dispatch and management.

[0003] However, traditional load driver identification methods mainly rely on simple correlation analysis or single statistical models, which have the following technical problems: First, existing methods are difficult to handle complex load changes under the combined influence of multiple factors, and often can only identify a single or a few main factors; second, they lack adaptability to the operating characteristics of different power networks, and the same load change may be driven by completely different factors in power grids of different sizes; third, when historical data is insufficient or new load change patterns are encountered, the identification accuracy of traditional methods drops significantly, and they cannot provide reliable analysis results.

[0004] Therefore, there is an urgent need for an intelligent identification method for the driving factors of power load. Summary of the Invention

[0005] This invention addresses the technical problems of existing load driver identification methods, such as difficulty in handling complex load changes with multiple coupled factors, poor adaptability, and decreased identification accuracy when data is insufficient. It provides an intelligent identification method and system for power load drivers.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0007] In a first aspect, the present invention provides an intelligent identification method for power load driving factors, comprising:

[0008] Real-time monitoring and acquisition of power load data; when the load change exceeds the preset load change threshold, the power load change characteristics of the current load change situation are obtained.

[0009] Based on the power load change characteristics, load change matching is performed to determine multiple historical similar load change scenarios, and each of the historical similar load change scenarios has a combination of driving factors.

[0010] Calculate the matching degree between each historical similar load change and the current load change, and select the driving factor combination corresponding to the historical similar load change with the highest matching degree to determine the load change driving factor of the current load change.

[0011] Secondly, the present invention provides an intelligent identification system for power load driving factors, comprising:

[0012] The real-time monitoring module is used to monitor and acquire power load data in real time. When the load change exceeds the preset load change threshold, it acquires the power load change characteristics of the current load change situation.

[0013] The change matching module is used to perform load change matching based on the power load change characteristics, determine multiple historical similar load change situations, and each of the historical similar load change situations has a combination of driving factors.

[0014] The factor determination module is used to calculate the matching degree between each historical similar load change and the current load change, and select the driving factor combination corresponding to the historical similar load change with the highest matching degree to determine the load change driving factor of the current load change.

[0015] The beneficial effects of this invention are:

[0016] Compared to existing technologies, this application first monitors and acquires power load data in real time. When the load change exceeds a preset load change threshold, it obtains the power load change characteristics of the current load change. Through load observation window monitoring and triggering by the preset load change threshold, it achieves precise screening of power load changes, filtering out noise without missing important changes, providing high-quality analysis objects for subsequent identification of driving factors. Secondly, based on the power load change characteristics, it performs load change matching to identify multiple historical similar load change scenarios. Each historical similar load change scenario has a combination of driving factors. Through the collaboration of historical data from multiple similar power grids, the sample coverage is expanded, providing a reliable historical reference for the driving factors of current power load changes. Finally, the matching degree between each historical similar load change and the current load change is calculated, and the driving factor combination corresponding to the historical similar load change with the highest matching degree is selected to determine the load change driving factor of the current load change. This improves the accuracy and reliability of driving factor identification. Furthermore, when the number of matching samples is less than the preset number of samples, multiple potential driving factors are predicted and verified to determine the load change driving factor of the current load change. This solves the problem of difficulty in identifying driving factors in scenarios with sparse historical data or rare load changes, and effectively improves the robustness and applicability of the load driving factor identification method.

[0017] Through the above technical solution, this application establishes an intelligent load observation window to monitor power load data in real time and extract key characteristic parameters of load changes. Then, it sends the current load change characteristics to multiple power networks of similar scale for historical pattern searching. It uses the historical load record databases of each network for feature matching and classifies and aggregates them according to combinations of driving factors. Furthermore, it obtains the currently occurring driving factors and calculates the matching degree with driving factor combinations of similar historical situations. The combination with the highest matching degree is selected as the driving factor. When historical matching samples are insufficient, a backup prediction verification mechanism is activated. By constructing multiple potential driving factor combinations, load prediction is performed based on a historical influence model, and the factor combination with the smallest prediction deviation is selected as the final result. Thus, through multi-dimensional feature and driving factor coupled analysis, the identification accuracy under complex load change scenarios is improved; the similar grid adaptation mechanism enhances adaptability to the operating characteristics of power grids of different scales; and the backup verification mechanism maintains high reliability even with insufficient data or new change patterns, promoting the intelligent and precise upgrading of power load driving factor identification. Attached Figure Description

[0018] Figure 1 A flowchart illustrating an intelligent identification method for power load driving factors provided by the present invention;

[0019] Figure 2 This is a schematic diagram of the structure of an intelligent identification system for power load driving factors provided by the present invention.

[0020] In the attached diagram, the components represented by each number are as follows:

[0021] Real-time monitoring module 11, change matching module 12, factor determination module 13. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0024] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0025] Example 1, as Figure 1 As shown, this embodiment of the invention provides an intelligent identification method for power load driving factors, including:

[0026] S10: Real-time monitoring and acquisition of power load data. When the load change of the power load data exceeds the preset load change threshold, the power load change characteristics of the current load change situation are obtained.

[0027] Power load data is susceptible to transient noise interference on a short time scale, such as instantaneous equipment start-up and shutdown, measurement errors, etc. Furthermore, the reasonable range of power load data variation is strongly correlated with parameters such as the power generation scale of the power grid. Therefore, using a fixed threshold lacks scenario adaptability.

[0028] To address the aforementioned issues, this application monitors and acquires power load data in real time. When the load change in the power load data exceeds a preset load change threshold, it acquires the power load change characteristics of the current load change situation.

[0029] Specifically, step S10 in the method includes:

[0030] A load observation window is established, and the real-time acquired power load data is transmitted to the load observation window for load change monitoring to obtain the load change amount;

[0031] When the load change is greater than or equal to the preset load change threshold, load change characteristic analysis is performed through the load observation window to obtain the power load change characteristics of the current load change situation.

[0032] In this embodiment, a load observation window is first established. Real-time acquired power load data is transmitted to the load observation window for load change monitoring to obtain the load change amount. The load observation window is a fixed-duration time window, such as 30 minutes or 1 hour, continuously receiving real-time power load data streams. The load observation window acts as a data buffer, using a first-in-first-out (FIFO) mechanism to update data, ensuring that the load status within the latest time period is always analyzed. For example, within the load observation window, the load change amount for that period is obtained by calculating the difference between the load value at the end of the load observation window and the load value at the beginning of the window. For instance, if the load observation window is 1 hour, the load value at the beginning of the window is 50 MW, and the load value at the end of the window is 58 MW, then the load change amount is 8 MW. The load change amount quantifies the overall fluctuation of the load within the load observation window.

[0033] Secondly, when the load change is greater than or equal to a preset load change threshold, load change characteristic analysis is performed through the load observation window to obtain the power load change characteristics of the current load change situation. For example, when the calculated load change is greater than or equal to the preset load change threshold (e.g., 6MW), load change characteristic analysis is performed through the load observation window; if it is less than the preset load change threshold, monitoring continues without further analysis. This avoids ineffective processing of minor fluctuations (such as normal electricity consumption errors), saves computational resources, and ensures that the analysis focuses on load disturbances that truly need to be attributed.

[0034] Specifically, the steps for setting the "preset load change threshold" include:

[0035] Obtain the current power generation capacity and load capacity of the power network, and search for multiple similar power networks based on the power generation capacity and load capacity;

[0036] Historical power load data of each of the similar power networks is extracted, and the historical power load data is divided into scenarios to obtain multiple load change scenarios;

[0037] Box plot analysis was performed on the load change amounts for each load change scenario to obtain the load change distribution range corresponding to each load change scenario;

[0038] The upper boundary value of each load change distribution interval is determined as the preset load change threshold for the corresponding load change scenario.

[0039] In this embodiment, the current power generation capacity and load capacity of the current power network are first obtained, and multiple similar power networks are retrieved based on these capacities. Specifically, the current power generation capacity and load capacity are obtained because the load change threshold is closely related to the power network's capacity and load capacity. For example, an 8MW load change is a huge change for a 50MW power network, but may be a small fluctuation for a 5000MW power network. Therefore, it is necessary to filter out multiple similar power networks with similar capacity and load from the historical database based on the current power network's capacity (e.g., total installed capacity) and load capacity (e.g., average load, peak load). For example, using the current power network's capacity and load capacity as search constraints, multiple similar power networks in cities B and C within the same province are retrieved. The historical data of these similar power networks can serve as a reference benchmark for the current power network.

[0040] Secondly, historical power load data from similar power networks is extracted, and the historical power load data is divided into scenarios to obtain multiple load change scenarios. Specifically, the patterns of load changes are strongly correlated with specific scenarios. For example, the load change characteristics of summer high temperatures and winter cold waves are different, as are those of weekdays and holidays. Therefore, it is necessary to classify the historical power load data of similar power networks according to scenario dimensions (such as season, time period, holiday type, extreme weather, etc.) to obtain multiple load change scenarios. For example, the historical power load data of each similar power network is divided into scenarios such as spring weekdays, spring rest days, summer weekdays, summer rest days, autumn weekdays, autumn rest days, winter weekdays, and winter rest days according to seasonal weekdays and rest days. Each scenario corresponds to a set of historical power load data.

[0041] Next, box plot analysis was performed on the load change amounts for each load change scenario to obtain the load change distribution range corresponding to each load change scenario. The box plot analysis used quartiles (Q1, Q2, Q3) to describe the data distribution characteristics: Q1 is the lower quartile, representing 25% of the sample data being less than this value, and Q3 is the upper quartile, representing 75% of the sample data being less than this value. The interquartile range (IQR) was calculated as IQR = Q3 - Q1, which is used to reflect the concentration of the data. The upper boundary was defined as Q3 + 1.5 × IQR, which is the upper limit of the normal fluctuation of the load change amount under this scenario.

[0042] Finally, since the upper boundary of the box plot represents the upper limit of normal fluctuations in that scenario, the upper boundary value of each load change distribution interval is determined as the preset load change threshold for the corresponding load change scenario. When the load change is greater than or equal to the preset load change threshold, it exceeds the normal fluctuation range of that scenario and requires further analysis. In this way, through similar power network references and scenario division, the preset load change threshold can adapt to both the actual power generation and load scale of the current power grid and the load characteristics of different scenarios, significantly reducing misjudgments and omissions caused by unreasonable thresholds.

[0043] Furthermore, the phrase "analyzing load change characteristics through the load observation window to obtain the power load change characteristics of the current load change situation" includes:

[0044] Calculate the load change amplitude, load change rate, and load change duration of the power load data within the load observation window;

[0045] The load change magnitude, the load change rate, and the load change duration are used as the characteristics of the current load change situation.

[0046] In this embodiment of the application, the load change amplitude, load change rate, and load change duration of the power load data within the load observation window are first calculated. The load change amplitude represents the maximum deviation of the load from the initial value within the load observation window, the load change rate is the intensity of load change per unit time, and the load change duration is the length of time that the load deviates from the initial level and maintains a significant change. For example, the load change amplitude can be obtained by acquiring the extreme value and initial value of the power load data within the load observation window, calculating the absolute difference between the extreme value and the initial value, and then taking the maximum value of the absolute difference. For example, if the initial value is 50MW, the maximum value rises to 58MW, and the minimum value drops to 45MW, then the load change amplitude = max(|58-50|, |45-50|) = 8MW; the load change rate can be obtained by acquiring the time required to reach the load change amplitude, and then calculating the ratio. For example, if it takes 5 minutes to reach the load change amplitude of 8MW, then the load change rate = 8 / 5 = 1.6MW / min; the load change duration can be calculated by statistically analyzing the duration for which the load remains in a significantly changing state (e.g., ≥80% × load change amplitude). For example, the total time from the moment the load first crosses the load change amplitude × 80% to the moment it finally falls back to within 80% × load change amplitude is 10 minutes, which is used as the load change duration.

[0047] Secondly, the magnitude, rate, and duration of load changes are used as characteristics of current load changes. For example, the magnitude, rate, and duration of load changes characterize load changes from three dimensions: intensity, speed, and duration, respectively. For instance, load changes caused by extreme weather may manifest as high magnitude, high rate, and long duration of load changes. This transforms load changes into calculable and comparable data indicators, ensuring a unified standard for judging the similarity between different load change scenarios.

[0048] In summary, compared to existing technologies, this application monitors and acquires power load data in real time. When the load change exceeds a preset load change threshold, it obtains the power load change characteristics of the current load change situation. Thus, through load observation window monitoring and preset load change threshold triggering, it achieves precise screening of power load changes, filtering out noise without missing important changes, providing high-quality analytical data for subsequent identification of driving factors.

[0049] S20: Based on the power load change characteristics, load change matching is performed to determine multiple historical similar load change scenarios, each of which has a combination of driving factors.

[0050] The aforementioned steps retrieve multiple similar power networks based on the current power generation and load scale. The historical data of these multiple similar power networks is an important reference. By matching them with the current load change characteristics, similar historical data can be used to identify the driving factors of the current load change and to point out potential factors.

[0051] To address the aforementioned issues, this application performs load change matching based on the aforementioned power load change characteristics to determine multiple historical similar load change scenarios, each of which has a combination of driving factors.

[0052] Specifically, step S20 in the method includes:

[0053] The power load change characteristics are sent to the multiple similar power networks. Each of the similar power networks has a historical load record database. Each of the historical load record databases contains multiple historical load change characteristics and corresponding labeled driving factor combinations.

[0054] Each of the similar power networks performs feature matching in its respective historical load record database and feeds back the matched historical load feature set;

[0055] Collect matching historical load feature sets from the feedback of each of the similar power networks, and classify and aggregate them according to the combination of driving factors;

[0056] Based on the classification and aggregation results, multiple historical similar load changes are identified, and each of these historical similar load changes has a corresponding combination of driving factors.

[0057] In this embodiment, the power load change characteristics are first sent to multiple similar power networks. Each similar power network has a historical load record database, which contains multiple historical load change characteristics and corresponding labeled driving factor combinations. These labeled driving factor combinations are manually labeled with driving factors for the historical load change characteristics after historical analysis and confirmation. Examples include cold wave weather + residential electric heating + weekdays, large-scale sporting events + temporary commercial electricity use, etc. For example, since multiple similar power networks are retrieved based on the current power generation and load scale of the power network, the historical load record data of similar power networks has strong reference value. The power load change characteristics (such as load change amplitude of 8MW, load change rate of 1.6MW / min, and load change duration of 10 minutes) are sent to multiple similar power networks, and the driving factor combinations are retrieved from the historical load record databases of each similar power network.

[0058] Secondly, each similar power network performs feature matching in its respective historical load record database and feeds back a set of matched historical load features. For example, feature matching can calculate the similarity between power load change features and historical load change features, such as calculating Euclidean distance, cosine similarity, etc., to filter out historical load change features with high similarity, forming a set of matched historical load features that includes power load change features and corresponding combinations of driving factors, and then feeding it back to the current power network.

[0059] Next, the historical load feature sets of feedback from similar power networks are collected and categorized and aggregated according to the combination of driving factors. For example, by summarizing the historical load feature sets of feedback from similar power networks, a large number of historical cases are obtained. Since different historical cases have the same or similar combinations of driving factors, these historical cases are categorized and aggregated according to the combination of driving factors. For example, all historical load features with the combination of driving factors labeled as "high temperature + residential electricity consumption" are grouped into one category, and all historical load features with the combination of driving factors labeled as "temporary commercial electricity consumption + weekdays" are grouped into another category. In this way, by categorizing and aggregating, the randomness of a single case can be reduced and the robustness of data analysis can be improved.

[0060] Finally, based on the classification and aggregation results, multiple historical similar load change scenarios are identified, each with a corresponding combination of driving factors. For example, the top 3-5 data sets with the highest frequency and average similarity can be selected as multiple historical similar load change scenarios. Each scenario has a corresponding combination of driving factors, which serves as a direct reference for subsequently determining which factors are driving the current load change. In this way, by retrieving similar power grid data and filtering out low-similarity, infrequent, and incidental cases, the problem of insufficient coverage of historical data from a single power grid is solved, improving the comprehensiveness and accuracy of the matching.

[0061] In summary, compared to existing technologies, this application matches load changes based on the aforementioned power load change characteristics to identify multiple historical similar load change scenarios, each of which has a combination of driving factors. Thus, through the collaboration of historical data from multiple similar power grids, the sample coverage is expanded, providing a reliable historical reference for the driving factors of current power load changes.

[0062] S30: Calculate the matching degree between each of the historical similar load change scenarios and the current load change scenario, and select the driving factor combination corresponding to the historical similar load change scenario with the highest matching degree to determine the load change driving factor of the current load change scenario.

[0063] Historical data on similar load changes contains a large number of similar load change cases with clear combinations of driving factors. Therefore, by matching these historical cases with the current situation, the driving factors of the current load change can be determined.

[0064] To address the aforementioned issues, this application calculates the matching degree between each of the historical similar load changes and the current load change, and selects the combination of driving factors corresponding to the historical similar load change with the highest matching degree to determine the load change driving factors for the current load change.

[0065] Specifically, step S30 in the method includes:

[0066] Obtain the actual driving factors when the current load change occurs;

[0067] The actual driving factors are compared with the driving factors of each historical similar load change to obtain the matching driving factors of each historical similar load change.

[0068] Based on the matching driving factors and corresponding combinations of driving factors for each historical similar load change, the matching degree between each historical similar load change and the current load change is obtained.

[0069] In this embodiment, the actual driving factors at the time of the current load change are first obtained. These actual driving factors are various objectively existing factors that may affect the power load during the period of the current load change, such as real-time temperature, precipitation, wind speed, weekday / holiday status, whether it is a peak electricity consumption period, and large-scale events. For example, the actual driving factors at the time of the current load change can be collected through real-time monitoring systems, such as weather forecasts, sensors, public information platforms, and enterprise-reported data. For instance, the actual driving factors at the time of the current load change can be obtained through weather forecasts or public information platforms as [high temperature 38℃, weekday, peak electricity consumption period, large-scale sporting event]. This ensures the authenticity and timeliness of the data, providing a benchmark for subsequent comparisons.

[0070] Secondly, the actual driving factors are compared with the driving factor combinations of various historical similar load changes to obtain the matching driving factors for each historical similar load change. For example, the actual driving factors are compared with the driving factor combinations of various historical similar load changes one by one, and the matching factors are screened and the number is counted. For example, the actual driving factors are [high temperature 38℃, weekday, peak electricity consumption period, large-scale sports event], and the driving factor combination of a certain historical similar load change is [high temperature 38℃, weekday, concentrated electricity consumption in commercial areas], and the matching factors are [high temperature, weekday].

[0071] Finally, based on the matching driving factors and corresponding combinations of driving factors for each historical similar load change, the matching degree between each historical similar load change and the current load change is obtained, where the matching degree = the number of matching driving factors for each historical similar load change / max (the actual number of driving factors and the number of combinations of driving factors for each historical similar load change). For example, if the actual driving factors are [high temperature 38℃, weekday, peak electricity consumption period, large-scale sports event], and the driving factor combination for a historical similar load change is [high temperature 38℃, weekday, concentrated electricity consumption in commercial areas], and the matching factor is [high temperature, weekday], then the number of matching driving factors for the historical similar load change is 2, the number of actual driving factors is 4, and the number of driving factor combinations for the historical similar load change is 3. Then the matching degree = 2 / max(4,3) = 0.5. In this way, following the same method, the matching driving factors for each historical similar load change are traversed, and multiple matching degrees are calculated. The closer the matching degree is to 1, the higher the degree of matching. The driving factor combination corresponding to the historical similar load change with the highest matching degree is selected as the load change driving factor for the current load change. Compared with the actual driving factors, the load change driving factors can more accurately pinpoint the core cause that truly causes load fluctuations, and can filter out the interference of irrelevant or secondary factors, providing a clear causal orientation and decision-making basis for power grid operation and control.

[0072] In summary, compared to existing technologies, this application calculates the matching degree between each historical similar load change and the current load change, and selects the driving factor combination corresponding to the historical similar load change with the highest matching degree to determine the load change driving factor for the current load change. Thus, based on the driving factor data of historical similar load change cases, the driving factor for the current load change is determined through a precise comparison between the current actual driving factor and the historical driving factor combination, effectively improving the accuracy and reliability of driving factor identification.

[0073] Furthermore, if the number of matched samples is insufficient, misjudgments are likely to occur due to a lack of data representativeness. For example, when faced with rare load change events, if the matched samples are insufficient to support a reliable judgment, the accurate identification of driving factors can be achieved through the prediction and verification of potential driving factors, thereby compensating for the limitations of relying solely on historical data. Specifically:

[0074] The number of matching samples matching historical load features from multiple similar power network feedback sets is counted.

[0075] When the number of matched samples is less than the preset number of samples, multiple potential driving factors are predicted and verified to determine the driving factors of the current load change.

[0076] In this embodiment, the number of matching samples in the historical load feature sets of feedback from multiple similar power networks is first counted. The number of matching samples is the total number of historical cases in the feedback matching historical load feature sets of the similar power networks. For example, if the feedback matching historical load feature sets of 5 similar power networks include a total of 8 matching samples.

[0077] Secondly, when the number of matched samples is less than the preset sample size, multiple potential driving factors are predicted and verified to determine the load change driving factors for the current load change situation. Specifically, the preset sample size is a threshold set based on statistical reliability. For example, it can be determined through experience or experimentation that the preset sample size is 10. Those skilled in the art can dynamically adjust it according to the actual situation. If the number of matched samples is less than the preset sample size, it indicates that the representativeness of similar historical data is insufficient, and directly relying on these cases may lead to misjudgment. Therefore, the process of predicting and verifying multiple potential driving factors is triggered. For example, if the preset sample size is 10 and the current number of matched samples is 8, then multiple potential driving factors are predicted and verified to determine the load change driving factors for the current load change situation.

[0078] Specifically, the phrase "predicting and validating multiple potential driving factors to determine the driving factors of current load changes" includes:

[0079] Multiple combinations of potential driving factors are established based on the aforementioned multiple potential driving factors;

[0080] Based on the load change, the driving factors of the multiple potential driving factors combination are verified to determine the load change driving factors of the current load change situation.

[0081] In this embodiment, multiple combinations of potential driving factors are first established based on multiple potential driving factors. These potential driving factors are known factors that may affect changes in power load, such as meteorological factors (including temperature, precipitation, wind, extreme weather, etc.), time factors (including weekdays / holidays, peak hours, seasons, etc.), socio-economic factors (including industrial operating rates, commercial activities, policy adjustments, electricity price changes, etc.), and special events (including major sporting events, public emergencies, equipment failures, etc.). For example, since changes in power load are usually the result of multiple factors working together, individual potential driving factors need to be combined into combinations to cover possible interactive effects. For instance, if the potential driving factors are [high temperature, weekend, peak hours], multiple combinations of potential driving factors can be generated: [high temperature, weekend], [high temperature, peak hours], [weekend, peak hours].

[0082] Secondly, based on the load change, multiple potential driver combinations are validated to determine the driving factors of the current load change. This transforms the problem of insufficient data into a systematic validation of known potential factors, covering possible combinations of potential drivers and avoiding the omission of key drivers.

[0083] Furthermore, the step of "verifying the driving factors of the multiple potential driving factor combinations based on the load change amount, and determining the load change driving factors of the current load change situation" includes:

[0084] Obtain potential driver data for each of the aforementioned potential driver combinations during the period in which the current load change occurs;

[0085] Based on the potential driver data of each potential driver combination, the load change is predicted to obtain the predicted change for each potential driver combination.

[0086] The deviation between each predicted change and the load change is calculated to obtain the prediction deviation of each combination of potential driving factors.

[0087] The driver factor selected from the combination of potential drivers with the smallest prediction deviation is chosen as the load change driver for the current load change situation.

[0088] In this embodiment, the potential driving factor data for each combination of potential driving factors during the period in which the current load change occurs is first obtained. For example, for each combination of potential driving factors, the actual data of each factor in that combination during the period in which the current load change occurs are collected. For instance, if the combination of potential driving factors is [high temperature, weekend], then the actual temperature and day of the week during the period in which the current load change occurs are obtained, such as 38℃, Saturday, to obtain the potential driving factor data [high temperature 38℃, Saturday], providing reliable input data for subsequent forecasting.

[0089] Secondly, load change prediction is performed based on the potential driving factor data for each combination of potential driving factors to obtain the predicted change for each combination of potential driving factors. For example, a load change prediction model can be constructed based on LSTM, taking the potential driving factor data as input and predicting the load change output. For instance, taking the potential driving factor data [high temperature 38℃, Saturday] as input, the predicted load change output is: predicted load change amplitude of 10MW, predicted load change rate of 1.5MW / min, and predicted load change duration of 15 minutes.

[0090] For example, the load change prediction model can be trained through the following technical path: 1. Data preparation: Collect the time series of potential driving factors in historical data and the actual load change in the corresponding period. By removing outliers and normalizing, the feature values ​​are mapped to the [0,1] interval to eliminate the influence of the dimension, forming a sample set. The sample set is divided into a training set (for model parameter learning) and a validation set (for monitoring overfitting) in an 8:2 ratio. 2. Model Construction: An LSTM (Long Short-Term Memory) architecture is adopted, leveraging its ability to capture temporal dependencies to adapt to the dynamic characteristics of load changes. It mainly consists of an input layer, hidden layers, and an output layer. The input layer's dimension is (time step, number of features). The time step is set to N consecutive historical time periods, such as N=24, corresponding to a 24-hour sequence. The number of features corresponds to the types of potential driving factors; for example, if there are three types of factors—temperature, time period, and operating rate—then the number of features is 3. The hidden layer contains two LSTM units: the first layer has 64 neurons, and the second layer has 32 neurons. Each layer is followed by a dropout layer (dropout rate = 0.2) to suppress overfitting. The ReLU activation function enhances the nonlinear fitting ability. The output layer contains one neuron, using a linear activation function, directly outputting the predicted value of load change for a single time period. 3. Model Training: Using time-series data of potential driving factors in the training set as input and the actual load change in the corresponding time period as supervision label, the Adam optimizer (learning rate = 0.001) is used to minimize the loss function (mean squared error, i.e., the mean squared difference between the predicted value and the actual value). The training rounds are set to 200 rounds. After each round, the MSE of the validation set is calculated. When the prediction bias of the validation set is ≤5%, the model is considered to have converged, training is stopped, and the trained load change prediction model is obtained.

[0091] Next, the deviations between each predicted change and the load change are calculated to obtain the prediction deviation for each combination of potential driving factors. The prediction deviation for each predicted change is calculated as (|each predicted change - load change|) / load change. The prediction deviation for a combination of potential driving factors is the sum of the prediction deviations for each predicted change. For example, if the predicted load change is 10MW, the predicted load change rate is 1.5MW / min, and the predicted load change duration is 15 minutes, while the actual load change is 8MW, the load change rate is 1.6MW / min, and the load change duration is 10 minutes, then the prediction deviation for the current combination of potential driving factors is (|10-8|) / 8 + (|1.5-1.6|) / 1.6 + (|15-10|) / 10 = 0.8125. Similarly, the prediction deviation for each combination of potential driving factors is calculated. The smaller the prediction deviation of a combination of potential driving factors, the better it explains the actual load change; that is, the combination of potential driving factors is more likely to be the actual driving factor.

[0092] Finally, among all potential driver combinations, the driver combination with the smallest prediction deviation is selected as the load change driver for the current load change situation. The driver factors contained in this combination are the most likely factors for the current load change.

[0093] In summary, compared to existing technologies, this application predicts and verifies multiple potential driving factors when the number of matched samples is less than a preset number of samples, thereby determining the driving factors of load changes in the current load situation. This solves the problem of difficulty in identifying driving factors in scenarios with sparse historical data or rare load changes, effectively improving the robustness and applicability of the load driving factor identification method.

[0094] In summary, the embodiments of this application have at least the following technical effects:

[0095] Compared to existing technologies, this application first monitors and acquires power load data in real time. When the load change exceeds a preset load change threshold, it acquires the power load change characteristics of the current load change situation. In this way, through load observation window monitoring and preset load change threshold triggering, it achieves accurate screening of power load changes, filtering out noise without missing important changes, providing high-quality analysis objects for subsequent identification of driving factors.

[0096] Secondly, this application performs load change matching based on the aforementioned power load change characteristics to identify multiple historical similar load change scenarios, each of which has a combination of driving factors. Thus, through the collaboration of historical data from multiple similar power grids, the sample coverage is expanded, providing a reliable historical reference for the driving factors of current power load changes.

[0097] Furthermore, this application calculates the matching degree between each of the historical similar load change scenarios and the current load change scenario, and selects the driving factor combination corresponding to the historical similar load change scenario with the highest matching degree to determine the load change driving factor for the current load change scenario. Thus, based on the driving factor data of historical similar load change cases, the driving factor for the current load change is determined through a precise comparison between the current actual driving factor and the historical driving factor combination, effectively improving the accuracy and reliability of driving factor identification.

[0098] Finally, when the number of matched samples is less than the preset number of samples, this application performs prediction verification on multiple potential driving factors to determine the load change driving factors for the current load change situation. In this way, it solves the problem of difficulty in identifying driving factors in scenarios with sparse historical data or rare load changes, effectively improving the robustness and applicability of the load driving factor identification method.

[0099] Through the above technical solution, this application fully considers the impact of chronic inflammation in patients with renal anemia on the monitoring results. By adjusting the monitoring influence coefficient, the preset iron metabolism monitoring cycle is dynamically optimized to obtain an optimized monitoring cycle. This improves the accuracy of iron metabolism monitoring in renal anemia.

[0100] Example 2, as Figure 2 As shown, based on the same inventive concept as the intelligent identification method for power load driving factors provided in Embodiment 1, this embodiment of the invention also provides an intelligent identification system for power load driving factors, including:

[0101] Real-time monitoring module 11 is used to monitor and acquire power load data in real time. When the load change of the power load data exceeds the preset load change threshold, it acquires the power load change characteristics of the current load change situation.

[0102] The change matching module 12 is used to perform load change matching based on the power load change characteristics, determine multiple historical similar load change situations, and each of the historical similar load change situations has a combination of driving factors;

[0103] The factor determination module 13 is used to calculate the matching degree between each of the historical similar load change situations and the current load change situation, and select the driving factor combination corresponding to the historical similar load change situation with the highest matching degree to determine the load change driving factor of the current load change situation.

[0104] The real-time monitoring module 11 is specifically used for:

[0105] A load observation window is established, and the real-time acquired power load data is transmitted to the load observation window for load change monitoring to obtain the load change amount;

[0106] When the load change is greater than or equal to the preset load change threshold, load change characteristic analysis is performed through the load observation window to obtain the power load change characteristics of the current load change situation.

[0107] Specifically, the steps for setting the "preset load change threshold" include:

[0108] Obtain the current power generation capacity and load capacity of the power network, and search for multiple similar power networks based on the power generation capacity and load capacity;

[0109] Historical power load data of each of the similar power networks are extracted, and the historical power load data is divided into scenarios to obtain multiple load change scenarios;

[0110] Box plot analysis was performed on the load change amounts for each load change scenario to obtain the load change distribution range corresponding to each load change scenario;

[0111] The upper boundary value of each load change distribution interval is determined as the preset load change threshold for the corresponding load change scenario.

[0112] Specifically, the phrase "analyzing load change characteristics through the load observation window to obtain the power load change characteristics of the current load situation" includes:

[0113] Calculate the load change amplitude, load change rate, and load change duration of the power load data within the load observation window;

[0114] The load change magnitude, the load change rate, and the load change duration are used as the characteristics of the current load change situation.

[0115] Specifically, the change matching module 12 is used for:

[0116] The power load change characteristics are sent to the multiple similar power networks. Each of the similar power networks has a historical load record database. Each of the historical load record databases contains multiple historical load change characteristics and corresponding labeled driving factor combinations.

[0117] Each of the similar power networks performs feature matching in its respective historical load record database and feeds back the matched historical load feature set;

[0118] Collect matching historical load feature sets from the feedback of each of the similar power networks, and classify and aggregate them according to the combination of driving factors;

[0119] Based on the classification and aggregation results, multiple historical similar load changes are identified, and each of these historical similar load changes has a corresponding combination of driving factors.

[0120] The factor determination module 13 is specifically used for:

[0121] Obtain the actual driving factors when the current load change occurs;

[0122] The actual driving factors are compared with the driving factors of each historical similar load change to obtain the matching driving factors of each historical similar load change.

[0123] Based on the matching driving factors and corresponding combinations of driving factors for each historical similar load change, the matching degree between each historical similar load change and the current load change is obtained.

[0124] Furthermore, the factor determination module 13 is specifically used for:

[0125] The number of matching samples matching historical load features from multiple similar power network feedback sets is counted.

[0126] When the number of matched samples is less than the preset number of samples, multiple potential driving factors are predicted and verified to determine the driving factors of the current load change.

[0127] Specifically, the phrase "predicting and validating multiple potential driving factors to determine the driving factors of current load changes" includes:

[0128] Multiple combinations of potential driving factors are established based on the aforementioned multiple potential driving factors;

[0129] Based on the load change, the driving factors of the multiple potential driving factors combination are verified to determine the load change driving factors of the current load change situation.

[0130] Specifically, the step of "verifying the driving factors of the multiple potential driving factor combinations based on the load change amount, and determining the load change driving factors of the current load change situation" includes:

[0131] Obtain potential driver data for each of the aforementioned potential driver combinations during the period in which the current load change occurs;

[0132] Based on the potential driver data of each potential driver combination, the load change is predicted to obtain the predicted change for each potential driver combination.

[0133] The deviation between each predicted change and the load change is calculated to obtain the prediction deviation of each combination of potential driving factors.

[0134] The driver factor selected from the combination of potential drivers with the smallest prediction deviation is chosen as the load change driver for the current load change situation.

[0135] In summary, the embodiments of this application have at least the following technical effects:

[0136] Compared to existing technologies, this application firstly uses a real-time monitoring module to acquire power load data in real time. When the load change exceeds a preset load change threshold, it obtains the power load change characteristics of the current load change. Through load observation window monitoring and triggering by the preset load change threshold, it achieves precise screening of power load changes, filtering out noise without missing important changes, providing high-quality analysis objects for subsequent identification of driving factors. Secondly, through a change matching module, it performs load change matching based on power load change characteristics to identify multiple historical similar load change scenarios. Each historical similar load change scenario has a combination of driving factors. By collaborating with historical data from multiple similar power grids, the sample coverage is expanded, providing a reliable historical reference for the driving factors of current power load changes. Finally, the factor determination module calculates the matching degree between historical similar load changes and current load changes, and selects the driving factor combination corresponding to the historical similar load change with the highest matching degree to determine the load change driving factor for the current load change. This improves the accuracy and reliability of driving factor identification. Furthermore, when the number of matching samples is less than the preset number of samples, multiple potential driving factors are predicted and verified to determine the load change driving factor for the current load change. This solves the problem of difficulty in identifying driving factors in scenarios with sparse historical data or rare load changes, effectively improving the robustness and applicability of the load driving factor identification method. Thus, by coupling multi-dimensional features with driving factors, the identification accuracy in complex load change scenarios is improved; by leveraging the similar grid adaptation mechanism, the adaptability to the operating characteristics of grids of different scales is enhanced; and by employing a backup verification mechanism, high reliability is maintained even with insufficient data or new change patterns, promoting the intelligent and precise upgrading of power load driving factor identification.

[0137] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0138] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied 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.

[0139] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (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 computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0140] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function 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.

[0141] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable 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.

[0142] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0143] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for intelligent identification of power load driving factors, characterized in that, The method includes: Real-time monitoring and acquisition of power load data; when the load change exceeds a preset load change threshold, acquisition of the power load change characteristics of the current load change; the steps for setting the preset load change threshold include: Obtain the current power generation capacity and load capacity of the power network, and search for multiple similar power networks based on the power generation capacity and load capacity; Historical power load data of each of the similar power networks are extracted, and the historical power load data is divided into scenarios to obtain multiple load change scenarios; Box plot analysis was performed on the load change amounts for each load change scenario to obtain the load change distribution range corresponding to each load change scenario; The upper boundary value of each load change distribution interval is determined as the preset load change threshold for the corresponding load change scenario; Based on the aforementioned power load change characteristics, load change matching is performed to determine multiple historical similar load change scenarios. Each of these historical similar load change scenarios has a combination of driving factors, including: The power load change characteristics are sent to the multiple similar power networks. Each of the similar power networks has a historical load record database. Each of the historical load record databases contains multiple historical load change characteristics and corresponding labeled driving factor combinations. Each of the similar power networks performs feature matching in its respective historical load record database and feeds back the matching historical load feature set; Collect matching historical load feature sets from the feedback of each of the similar power networks, and classify and aggregate them according to the combination of driving factors; Based on the classification and aggregation results, multiple historical similar load changes are identified, and each of the historical similar load changes has a corresponding combination of driving factors. Calculate the matching degree between each historical similar load change and the current load change, and select the driving factor combination corresponding to the historical similar load change with the highest matching degree to determine the load change driving factor of the current load change.

2. The method according to claim 1, characterized in that, Real-time monitoring and acquisition of power load data; when the load change exceeds a preset load change threshold, acquisition of the current load change characteristics, including: A load observation window is established, and the real-time acquired power load data is transmitted to the load observation window for load change monitoring to obtain the load change amount; When the load change is greater than or equal to the preset load change threshold, load change characteristic analysis is performed through the load observation window to obtain the power load change characteristics of the current load change situation.

3. The method according to claim 2, characterized in that, By analyzing load change characteristics through the load observation window, the power load change characteristics of the current load change situation are obtained, including: Calculate the load change amplitude, load change rate, and load change duration of the power load data within the load observation window; The load change magnitude, the load change rate, and the load change duration are used as the characteristics of the current load change situation.

4. The method according to claim 1, characterized in that, Calculate the matching degree between each of the aforementioned historical similar load changes and the current load change, including: Obtain the actual driving factors when the current load change occurs; The actual driving factors are compared with the driving factors of each historical similar load change to obtain the matching driving factors of each historical similar load change. Based on the matching driving factors and corresponding combinations of driving factors for each historical similar load change, the matching degree between each historical similar load change and the current load change is obtained.

5. The method according to claim 1, characterized in that, The method further includes: The number of matching samples matching historical load features from multiple similar power network feedback sets is counted. When the number of matched samples is less than the preset number of samples, multiple potential driving factors are predicted and verified to determine the driving factors of the current load change.

6. The method according to claim 5, characterized in that, Predictive validation of multiple potential drivers is performed to identify the drivers of current load changes, including: Multiple combinations of potential driving factors are established based on the aforementioned potential driving factors; Based on the load change, the driving factors of the multiple potential driving factors combination are verified to determine the load change driving factors of the current load change situation.

7. The method according to claim 6, characterized in that, Based on the load change, the driving factors of the multiple potential driving factors combination are verified to determine the load change driving factors of the current load change situation, including: Obtain potential driver data for each of the aforementioned potential driver combinations during the period in which the current load change occurs; Based on the potential driver data of each potential driver combination, the load change is predicted to obtain the predicted change for each potential driver combination. The deviation between each predicted change and the load change is calculated to obtain the prediction deviation of each combination of potential driving factors. The driver factor selected from the combination of potential drivers with the smallest prediction deviation is chosen as the load change driver for the current load change situation.

8. An intelligent identification system for power load driving factors, characterized in that, For performing the method according to any one of claims 1-7, comprising: The real-time monitoring module is used to monitor and acquire power load data in real time. When the load change exceeds a preset load change threshold, it acquires the power load change characteristics of the current load change. The steps for setting the preset load change threshold include: Obtain the current power generation capacity and load capacity of the power network, and search for multiple similar power networks based on the power generation capacity and load capacity; Historical power load data of each of the similar power networks are extracted, and the historical power load data is divided into scenarios to obtain multiple load change scenarios; Box plot analysis was performed on the load change amounts for each load change scenario to obtain the load change distribution range corresponding to each load change scenario; The upper boundary value of each load change distribution interval is determined as the preset load change threshold for the corresponding load change scenario; The load change matching module is used to perform load change matching based on the power load change characteristics, and to determine multiple historical similar load change scenarios. Each of the historical similar load change scenarios has a combination of driving factors, including: The power load change characteristics are sent to the multiple similar power networks. Each of the similar power networks has a historical load record database. Each of the historical load record databases contains multiple historical load change characteristics and corresponding labeled driving factor combinations. Each of the similar power networks performs feature matching in its respective historical load record database and feeds back the matching historical load feature set; Collect matching historical load feature sets from the feedback of each of the similar power networks, and classify and aggregate them according to the combination of driving factors; Based on the classification and aggregation results, multiple historical similar load changes are identified, and each of the historical similar load changes has a corresponding combination of driving factors. The factor determination module is used to calculate the matching degree between each historical similar load change and the current load change, and select the driving factor combination corresponding to the historical similar load change with the highest matching degree to determine the load change driving factor of the current load change.

Images