Electric power marketing business online auditing all-in-one machine digital auditing system and method

Abstract

The invention relates to the technical field of electric power marketing, in particular to an electric power marketing business online auditing all-in-one machine digital inspection system, which comprises a data acquisition and classification module, a curve modeling and feature extraction module, a dynamic threshold generation module, a database management module and an anomaly detection and alarm module. According to the method, curve modeling is carried out by combining historical normal data, standardization processing is carried out through multiple curve modeling, a new standard curve is established, the curvature between points on the curve is used as a reference voucher, a threshold range is established, and the power consumption curve is automatically produced by monitoring existing power consumption data. And the calculated curvature is compared with a threshold value, so that electricity consumption abnormity can be judged, and inspection is facilitated.

Description

Technical Field

[0001] The present invention relates to the field of power marketing technology, and in particular to a digital audit system and method for an online audit all-in-one machine for power marketing business. Background Art

[0002] With the widespread adoption of smart meters and the explosive growth of power marketing data, traditional power marketing auditing approaches face numerous challenges. First, data collection and processing are inefficient, making it difficult to meet real-time requirements. Traditional methods rely on manual or simple automated tools, resulting in long data collection cycles, low accuracy, and an inability to rapidly classify and deeply mine massive amounts of data. Second, anomaly detection lacks accuracy, relying on fixed thresholds and simple rules. These methods struggle to adapt to complex and changing power usage patterns, leading to a significant number of anomalous behaviors remaining undetected.

[0003] Publication No. CN119513760A discloses a digital audit method and system for power marketing that acquires real-time power data from a specified area. This data includes real-time power consumption data from smart meters and real-time power transmission data from sensors installed at key nodes in the power grid. The audit results corresponding to the real-time power data are generated using a preset audit model. The audit model is used to identify potential abnormal behaviors corresponding to the real-time power data based on the characteristics of the real-time power consumption and transmission data. If potential abnormal behaviors are found in the audit results, corresponding power adjustment measures are triggered based on the corresponding behavior type. The primary purpose is to improve the accuracy of power marketing audits and reduce power losses.

[0004] In the above method, real-time electricity consumption data is monitored by external sensors, and anomalies are determined by analyzing the audit results using models. When determining anomalies in instantaneous electricity consumption data, it is impossible to make specific anomaly determinations based on the electricity consumption habits in the area. Therefore, we propose a digital audit system and method for online auditing of power marketing business. A comparison model is established through normal data in the existing database, and a model comparison is established for the existing electricity consumption data. Anomalies are determined through comparison thresholds, and judgments can be made for different situations during electricity audits. Summary of the Invention

[0005] To achieve the above objectives, the present invention proposes a digital audit system for online audit of power marketing business, which includes a data acquisition and classification module, a curve modeling and feature extraction module, a dynamic threshold generation module, a database management module, and an anomaly detection and alarm module;

[0006] The data collection and classification module collects electricity consumption data in real time from smart meters and collects user attribute data through the power grid database. The classification module performs multi-level classification based on electricity consumption patterns, generates a user tag library based on the classification, and stores it in association with its electricity consumption data.

[0007] The curve modeling and feature extraction module generates curves based on daily and monthly electricity consumption through time dimension modeling;

[0008] The dynamic threshold generation module generates the initial threshold and predicts the curvature threshold range for the next 24 hours. If the curvature exceeds the threshold for five consecutive times, the adaptive feedback mechanism is triggered to recalculate the threshold.

[0009] The database management module includes a standard database, sub-databases, and a data synchronization mechanism. The standard database stores standardized curve templates, curvature threshold ranges, and characteristic parameters for each user type. The sub-database is divided into sub-databases according to user type, storing raw power consumption data and associated curves in real time. The sub-databases are partitioned by week and month to improve query efficiency. The data synchronization mechanism aggregates sub-database data into the standard database, adopting an incremental update method to modify only the data.

[0010] The anomaly detection and alarm module includes a real-time computing engine, alarm strategy, and a visual audit interface.

[0011] In one example, the multi-level classification is divided into major categories according to industry attributes for primary classification, and within the same major category, subcategories are subdivided according to power consumption time distribution and load volatility through a clustering algorithm.

[0012] In one example, the daily electricity consumption curve has the X-axis representing time and the Y-axis representing instantaneous electricity consumption. The power outage period is marked as an invalid interval. The monthly electricity consumption curve has the X-axis representing date and the Y-axis representing daily total electricity consumption. If the data for a single day is missing by more than 50%, the average of the three consecutive days is used to fill the gap.

[0013] In one example, the initial threshold is calculated by counting the mean μ and standard deviation σ of the curvature of each user type. The initial threshold is set to .

[0014] In one example, the real-time computing engine receives real-time electricity consumption data through Apache Kafka, Spark Streaming generates curves by window, calls a pre-trained polynomial model, and outputs curvature values in real time. When the instantaneous curvature exceeds the threshold in the alarm strategy, a low-level alarm is triggered, and when three consecutive sampling points exceed the standard, a high-level alarm is triggered and pushed to the manager's mobile terminal. The rationality of the anomaly is judged in combination with weather and holiday information. The visual audit interface can display a comparison chart of the abnormal curve, provide historical data overlay analysis function, and support the export of abnormality reports.

[0015] In one example, a digital audit method for an online auditing machine for power marketing business includes the following steps:

[0016] Step 1: Data statistics: organize the original electricity consumption data and classify the electricity users according to the different types of electricity consumption;

[0017] Step 2: Data processing: Electricity consumption of the same type is sorted by time, divided into daily and monthly consumption, and a consumption curve is created. Feature analysis is performed on the sorted electricity consumption data curve to extract key features and generate a standard curve. The curve is smoothed using mathematical methods to generate a standardized curve. The curvature between the midpoints of the curve is calculated. Threshold ranges are established based on the characteristics of different types of standardized curves. The threshold ranges are automatically adjusted based on changes in historical and real-time data.

[0018] Step 3, database establishment: save the curve and curvature data established in steps 1 and 2 in a standard database;

[0019] Step 4: Existing data collection: Save the existing electricity consumption data in the existing database, and record the user type, establish a sub-database, and divide the sub-database according to the user type;

[0020] Step 5, data arrangement: Create a curve graph with the data in the database and calculate the curvature of the curve;

[0021] Step 6: Data comparison: Compare the calculated curve curvature with the threshold in the standard database in step 3. When it exceeds or falls below the threshold, a warning will be issued and abnormal power consumption curves will be retrieved for timely inspection.

[0022] Step 7, alarm processing: notify operation and maintenance personnel via email, push abnormal signals to the mobile app and trigger voice prompts, display the comparison chart of abnormal curve and standard curve, support drilling to view historical data for the same period, and export abnormality reports.

[0023] In one example, in step 1, user attribute data is extracted from the original database, and the user attribute data includes user industry type, equipment list, and address information.

[0024] In one example, in the processing of the power outage time period in step 2, if the data before and after the power outage period are complete, linear interpolation is used for filling; if the data is seriously missing, the average value of the data of the same sub-category users during the same period is used for filling.

[0025] In one example, in step 2, feature analysis is performed on the electricity consumption data curve to extract key features. The curvature is calculated as follows: , where Δy is the change in power consumption between adjacent points and Δx is the time interval.

[0026] In one example, in step 2, the future threshold range is predicted by a machine learning algorithm, the initial threshold is set, the mean μK and standard deviation σK of the curvature data of the past 90 days are counted, the threshold range is, the real-time threshold is adjusted, and the difference order d is determined by the time series prediction model through the ADF test, and the ACF / PACF graph determines (p, q); the historical 30-day curvature data is input, and the threshold boundary of the next 24 hours is predicted. If the curvature is detected to exceed the threshold for 5 consecutive times, the threshold recalculation is triggered.

[0027] The digital audit system and method for online auditing of electric power marketing business proposed by the present invention can bring the following beneficial effects:

[0028] 1. The present invention conducts curve modeling by combining historical normal data, and performs standardization through multiple curve modeling to establish a new standard curve. The curvature between points on the curve is used as a reference, and a threshold range is established. By monitoring the existing power consumption data to automatically generate the power consumption curve, the curvature is calculated and compared with the threshold to determine abnormal power consumption, which is convenient for inspection.

[0029] 2. The present invention establishes different databases and different threshold ranges for different electricity usage properties, thereby ensuring that the system can make different judgments based on different electricity usage properties, and can automatically change the thresholds to ensure the accuracy of abnormal judgments. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0031] Figure 1 A schematic diagram of the first-person perspective structure of a digital audit system and method for online auditing of power marketing business;

[0032] Figure 2 A schematic diagram of the process and structure of a digital audit system and method for online auditing of power marketing business; DETAILED DESCRIPTION

[0033] In order to more clearly illustrate the overall concept of the present invention, a detailed description is given below in an exemplary manner in conjunction with the accompanying drawings.

[0034] In the description of the present invention, it should be understood that the terms "center", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "axial", "radial", "circumferential", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the present invention.

[0035] Furthermore, 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 the technical features being referred to. Thus, a feature identified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0036] In the present invention, unless otherwise expressly specified or limited, terms such as "mounted," "connected," "connect," and "fixed" should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection, electrical connection, or communication; direct connection or indirect connection through an intermediate medium; and internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0037] In the present invention, unless otherwise clearly specified and limited, a first feature "above" or "below" a second feature may be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. In the description of this specification, the descriptions with reference to the terms "one scheme", "some schemes", "examples", "specific examples", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the scheme or example are included in at least one scheme or example of the present invention. In this specification, the schematic expressions of the above terms do not necessarily refer to the same scheme or example. Moreover, the specific features, structures, materials or characteristics described may be combined in an appropriate manner in any one or more schemes or examples.

[0038] like Figures 1 to 2 As shown, the present invention proposes a digital audit system and method for online auditing of power marketing business, including a data acquisition and classification module, a curve modeling and feature extraction module, a dynamic threshold generation module, a database management module, and an anomaly detection and alarm module.

[0039] The data collection and classification module collects electricity consumption data (timestamp, electricity value, user ID) in real time from smart meters, and collects user attribute data (address, industry type, equipment list) through the power grid database. The classification module performs multi-level classification based on electricity consumption patterns, and divides major categories according to industry attributes for primary classification. Within the same major category, clustering algorithms (such as K-means) are used to subdivide subcategories according to electricity consumption time distribution and load volatility. A user tag library is generated based on the classification and stored in association with its electricity consumption data.

[0040] The curve modeling and feature extraction module uses time dimension modeling to generate curves based on daily and monthly electricity consumption. For the daily electricity consumption curve, the sampling frequency can be selected as 15 minutes or 1 hour. The X-axis is time (00:00-24:00), and the Y-axis is instantaneous electricity consumption. Power outage periods (electricity consumption at ≥2 consecutive sampling points is zero) are linearly interpolated or marked as invalid intervals. In the monthly electricity consumption curve, the X-axis is the date (from the 1st to the 30th / 31st) and the Y-axis is the total daily electricity consumption. If the data for a single day is missing by more than 50%, the average of the three adjacent days is used to fill the gap. After the curve is established, it is standardized and a 3rd-5th order polynomial is selected to fit the original curve to eliminate random fluctuations.

[0041] Dynamic threshold generation module generates the initial threshold, calculates the mean (μ) and standard deviation (σ) of the curvature of each user type, and sets the initial threshold to The threshold is dynamically adjusted, and the ARIMA or LSTM model is used to predict the curvature threshold range for the next 24 hours. The (p, d, q) order is determined by the autocorrelation diagram (ACF) and partial autocorrelation diagram (PACF). If the curvature exceeds the threshold for five consecutive times, the adaptive feedback mechanism is triggered to recalculate the threshold.

[0042] The database management module includes a standard database, a sub-database, and a data synchronization mechanism. The standard database stores standardized curve templates, curvature threshold ranges, and characteristic parameters for each user type, and establishes a multidimensional index table based on "user type-time granularity (day / month)". The sub-database is divided into sub-databases based on user type, and stores raw electricity consumption data and associated curves in real time. The sub-databases are partitioned by week or month to improve query efficiency. The data synchronization mechanism executes ETL (Extract-Transform-Load) tasks every morning to aggregate the sub-database data into the standard database. Incremental updates are used to synchronize only newly added or modified data.

[0043] The anomaly detection and alarm module includes a real-time computing engine, an alarm strategy, and a visual audit interface. The real-time computing engine receives real-time electricity consumption data through Apache Kafka. Spark Streaming generates curves by window (15 minutes / 1 hour), calls a pre-trained polynomial model, and outputs the curvature value in real time. In the alarm strategy, a low-level alarm is triggered when the instantaneous curvature exceeds the threshold. A high-level alarm is triggered when three consecutive sampling points exceed the standard and is pushed to the manager's mobile terminal. The rationality of the anomaly is judged based on weather and holiday information. For example, a sudden drop in electricity consumption on a typhoon day is considered normal. The visual audit interface can display a comparison chart of the anomaly curve (current curve vs. standard curve), provide historical data overlay analysis function, and support the export of anomaly reports (PDF / Excel format).

[0044] The specific method of using the above-mentioned online audit integrated digital audit system includes the following steps:

[0045] Step 1: Data statistics: Organize the original electricity usage data and classify the electricity users according to the nature of electricity usage, such as ordinary residents, factory users, shopping mall users, etc. User attribute data is obtained from the power company database, including user industry type, equipment list, and address information.

[0046] Step 2: Data processing: Electricity consumption of the same type is sorted by time, divided into daily and monthly electricity consumption, and produced into an electricity consumption curve. In daily electricity consumption, time is used as the X-axis and electricity consumption is used as the Y-axis to generate a daily electricity consumption curve. The sampling interval is dynamically adjusted according to the user type. Factory users use a 15-minute granularity, and ordinary households use a 1-hour granularity. For power outage periods (electricity consumption is zero at ≥2 consecutive sampling points), the following strategies are adopted: linear interpolation. If the data before and after the power outage period is complete, linear interpolation is used to fill in the data. If the data is seriously missing, the average data of the same sub-category of users during the same period is used to fill in the data. In monthly electricity consumption, date is used as the X-axis and electricity consumption is used as the Y-axis to generate a monthly electricity consumption curve. The curve sampling frequency is daily. If a day is marked as an invalid day, the moving average of the three days before and after is used instead.

[0047] Perform feature analysis on the organized electricity consumption data curve, extract key features (such as peak values, valley values, average values, etc.), and generate a standard curve. Smooth the curve using mathematical methods (such as polynomial fitting, Fourier transform, etc.) to generate a standardized curve, and calculate the curvature between the midpoints of the curve. The curvature calculation formula is: , where Δy is the change in electricity consumption at adjacent points, Δx is the time interval, and the threshold range is established according to the characteristics of different types of standardized curves. The threshold range is automatically adjusted according to the changes in historical data and real-time data. For example, the future threshold range is predicted by machine learning algorithms (such as time series prediction). The initial threshold is set by counting the mean μK and standard deviation σK of the curvature data for the past 90 days. The threshold range is , real-time threshold adjustment, through the time series prediction model, the difference order d is determined by the ADF test, and the ACF / PACF graph is used to determine (p, q); input the historical 30-day curvature data and predict the threshold boundary for the next 24 hours. If the curvature exceeds the threshold for 5 consecutive times, the threshold recalculation is triggered.

[0048] Step 3: Database creation: Save the curve and curvature data created in steps 1 and 2 in a standard database.

[0049] Step 4: Existing data collection: Save the existing electricity consumption data in the existing database, and record the user type, establish a sub-database, and divide the sub-database according to the user type.

[0050] Step 5, data arrangement: Create a curve graph with the data in the database, create a curve with time as the x-axis and power consumption as the y-axis, and calculate the curvature of the curve.

[0051] Step 6: Data comparison: Compare the calculated curve curvature with the threshold in the standard database in step 3. When it exceeds or falls below the threshold, a warning will be issued and abnormal power consumption curves will be retrieved to facilitate timely audits.

[0052] Step 7, alarm processing: Notify operation and maintenance personnel via email, or push abnormal signals to the mobile app and trigger voice prompts, display a comparison chart of the abnormal curve and the standard curve, support drilling down to view historical data for the same period, and export abnormality reports (including timestamps, exceedance magnitude, and possible causes).

[0053] This system realizes automated online auditing of power marketing business through real-time data collection from smart meters and collaborative analysis on the cloud, combined with multi-level user classification (industry attributes, power consumption time distribution, load fluctuation) and dynamic curve modeling technology. It constructs five core modules: data collection and classification module, curve modeling and feature extraction module, dynamic threshold generation module, database management module and anomaly detection and alarm module. The system takes curvature quantification of power consumption fluctuations as the core, combines dynamic thresholds with scenario-based alarms to form a "collection-modeling-monitoring-feedback" closed loop, and generates multi-dimensional reports after automatically completing data audits in the background. It can not only avoid interfering with users' normal business, but also significantly improve the accuracy of anomaly detection and processing efficiency, providing intelligent and highly reliable audit support for the digital transformation of State Grid's marketing business.

[0054] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0055] The foregoing is merely an embodiment of the present invention and is not intended to limit the present invention. It will be apparent to those skilled in the art that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention are intended to be included within the scope of the claims of the present invention.

Claims

1. A digital audit system for online auditing of power marketing business, characterized by: It includes data acquisition and classification module, curve modeling and feature extraction module, dynamic threshold generation module, database management module, anomaly detection and alarm module; The data collection and classification module collects electricity consumption data in real time from smart meters and collects user attribute data through the power grid database. The classification module performs multi-level classification based on electricity consumption patterns, generates a user tag library based on the classification, and stores it in association with its electricity consumption data. The curve modeling and feature extraction module generates curves based on daily and monthly electricity consumption through time dimension modeling; The dynamic threshold generation module generates the initial threshold and predicts the curvature threshold range for the next 24 hours. If the curvature exceeds the threshold for five consecutive times, the adaptive feedback mechanism is triggered to recalculate the threshold. The database management module includes a standard database, sub-databases, and a data synchronization mechanism. The standard database stores standardized curve templates, curvature threshold ranges, and characteristic parameters for each user type. The sub-database is divided into sub-databases according to user type, storing raw power consumption data and associated curves in real time. The sub-databases are partitioned by week and month to improve query efficiency. The data synchronization mechanism aggregates sub-database data into the standard database, adopting an incremental update method to modify only the data. The anomaly detection and alarm module includes a real-time computing engine, alarm strategy, and a visual audit interface.

2. The digital audit system for online auditing of power marketing business according to claim 1 is characterized by: The multi-level classification is divided into major categories according to industry attributes for primary classification. Within the same major category, subcategories are divided according to power consumption period distribution and load volatility through clustering algorithm.

3. The digital audit system for online auditing of power marketing business according to claim 1 is characterized by: In the daily electricity consumption curve, the X-axis is time, the Y-axis is instantaneous electricity consumption, and the power outage period is marked as an invalid interval. In the monthly electricity consumption curve, the X-axis is date, and the Y-axis is total daily electricity consumption. If the data for a single day is missing by more than 50%, the average of the three consecutive days is used to fill in the gaps.

4. The digital audit system for online auditing of power marketing business according to claim 1 is characterized by: The initial threshold is used to calculate the mean μ and standard deviation σ of the curvature of each user type. The initial threshold is set to .

5. The digital audit system for online auditing of power marketing business according to claim 1 is characterized by: The real-time computing engine receives real-time electricity consumption data through Apache Kafka, uses Spark Streaming to generate curves by window, calls a pre-trained polynomial model, and outputs curvature values in real time. When the instantaneous curvature exceeds the threshold in the alarm strategy, a low-level alarm is triggered. When three consecutive sampling points exceed the threshold, a high-level alarm is triggered and pushed to the manager's mobile terminal. The rationality of the anomaly is judged in combination with weather and holiday information. The visual audit interface can display a comparison chart of the anomaly curve, provide historical data overlay analysis function, and support the export of anomaly reports.

6. A digital audit method for an online auditing machine for power marketing business, used to implement requirements 1 to 5, characterized by: The following steps are involved: Step 1: Data statistics: organize the original electricity consumption data and classify the electricity users according to the different types of electricity consumption; Step 2: Data processing: Electricity consumption of the same type is sorted by time, divided into daily and monthly consumption, and a consumption curve is created. Feature analysis is performed on the sorted electricity consumption data curve to extract key features and generate a standard curve. The curve is smoothed using mathematical methods to generate a standardized curve. The curvature between the midpoints of the curve is calculated. Threshold ranges are established based on the characteristics of different types of standardized curves. The threshold ranges are automatically adjusted based on changes in historical and real-time data. Step 3, database establishment: save the curve and curvature data established in steps 1 and 2 in a standard database; Step 4: Existing data collection: Save the existing electricity consumption data in the existing database, and record the user type, establish a sub-database, and divide the sub-database according to the user type; Step 5, data arrangement: Create a curve graph with the data in the database and calculate the curvature of the curve; Step 6: Data comparison: Compare the calculated curve curvature with the threshold in the standard database in step 3. When it exceeds or falls below the threshold, a warning will be issued and abnormal power consumption curves will be retrieved for timely inspection. Step 7, alarm processing: notify operation and maintenance personnel via email, push abnormal signals to the mobile app and trigger voice prompts, display the comparison chart of abnormal curve and standard curve, support drilling to view historical data for the same period, and export abnormality reports.

7. The digital audit method for online auditing of electric power marketing business according to claim 6, characterized in that: In the step 1, user attribute data is extracted from the original database, and the user attribute data includes the user's industry type, equipment list, and address information.

8. The digital audit method for online auditing of electric power marketing business according to claim 6 is characterized by: In the processing of the power outage period in step 2, if the data before and after the power outage period are complete, linear interpolation is used to fill in the data. If the data is seriously missing, the data is filled in by referring to the average value of the data of the same sub-category users during the same period.

9. The digital audit method for online auditing of electric power marketing business according to claim 6, characterized in that: In the second step, the power consumption data curve is analyzed to extract key features. The curvature is calculated as follows: , where Δy is the change in power consumption between adjacent points and Δx is the time interval.

10. The digital audit method for online auditing of electric power marketing business according to claim 6, characterized in that: In the second step, the future threshold range is predicted by the machine learning algorithm, the initial threshold is set, the mean μK and standard deviation σK of the curvature data of the past 90 days are counted, the threshold range is, the real-time threshold is adjusted, and the difference order d is determined by the time series prediction model through the ADF test, and the ACF / PACF graph determines (p, q); the historical 30-day curvature data is input to predict the threshold boundary for the next 24 hours. If the curvature is detected to exceed the threshold for 5 consecutive times, the threshold recalculation is triggered.

Images