Alarm device of electric energy metering experiment scheduling management platform and application method thereof

Abstract

The invention relates to an alarm device of an electric energy metering experiment scheduling management platform and an application method thereof, which can replace manual analysis and perform statistics on an operation and maintenance log of an electric energy metering experiment and a verification qualified rate of a verification assembly line so as to perform fault analysis and automatic alarm supervision of experiment operation quality. The method avoids spending a lot of time to browse logs and statistical data, reduces the huge workload and time consumption, quickly provides an effective analysis result for the quality operation of an electric energy metering experiment, and improves the verification efficiency. The fault AI detection model can intelligently identify whether the operation and maintenance log A has the fault log feature or not, so that fault analysis is carried out, manual work is replaced to realize intelligent obstacle removal work, labor is saved, and the obstacle removal efficiency is improved.

Description

Technical Field

[0001] The present disclosure relates to the technical field of electric energy metering experiments, and in particular to an alarm device of an electric energy metering experiment scheduling management platform, an application method thereof, and an electronic device. Background Art

[0002] In the electric energy metering experiment, it is necessary to perform calibration tests on electric energy metering equipment, such as different types of electric energy meters.

[0003] As attached Figure 1 The electric energy metering experiment scheduling and management platform shown is a platform that can schedule and manage the electric energy metering calibration system including "six lines and one warehouse (six automated calibration lines and one intelligent storage system, which are systems for quality calibration and storage of electric energy metering equipment such as electricity meters)". The electric energy metering experiment scheduling and management platform is connected to the production data of "six lines and one warehouse (six automated calibration systems and one intelligent storage system)" to schedule and monitor the production situation of "six lines and one warehouse".

[0004] The electric energy metering experiment dispatching management platform, six automated calibration systems, intelligent storage system, and visual large-screen display system are networked and securely isolated from other application systems and public network channels using firewalls to ensure the information security of the system.

[0005] However, during the verification process of the electric energy metering experiment, it is necessary to monitor and analyze the experimental operation (platform) status data and fault alarm data in real time, and monitor the experimental operation quality of the automated assembly line (six lines and one library). Because the fault alarm in the operation status data of the verification assembly line corresponds to the verification result of the assembly line (that is, the experimental operation quality), if a large number of operation faults occur, the verification quality will be reduced, and even the verification results of the electric energy metering equipment will be affected, causing the experimental operation quality to be reduced. Therefore, it is necessary to find the specific fault of the verification assembly line, to cure it, and to improve the experimental operation quality of the verification assembly line.

[0006] The traditional technical solution mainly relies on manual analysis, on-site quality verification and inspection, and statistics on the operation and maintenance logs of the electricity metering experiment and the verification pass rate of the verification line, so as to conduct fault analysis and supervise the quality of experimental operation. This requires a lot of time to browse logs and statistical data, which is a huge workload and time-consuming, and it is impossible to quickly provide effective analysis results for the quality operation of the electricity metering experiment. Summary of the invention

[0007] In order to solve the above problems, the present application proposes an alarm device for an electric energy metering experiment scheduling management platform, an application method and an electronic device thereof.

[0008] On the one hand, the present application proposes an alarm device for an electric energy metering experiment scheduling management platform, comprising:

[0009] (1) The electric energy metering experiment dispatching management platform is used to read the verification data of each automated verification line in this metering verification, including the quality verification results and the corresponding operation and maintenance logs;

[0010] (2) A quality inspection system, which is used to read each piece of inspection data using a preset quality inspection model, and to determine in turn whether the quality inspection result in each piece of inspection data is qualified:

[0011] If qualified, give up;

[0012] If it fails, the operation and maintenance log A corresponding to the unqualified quality inspection result is imported into the fault intelligent detection system:

[0013] (3) A fault intelligent detection system, which is used to identify and analyze whether the operation and maintenance log A has fault log characteristics through a pre-deployed fault AI detection model:

[0014] If a fault log feature appears, corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log feature;

[0015] If no fault log feature appears, the operation and maintenance log A is sent to the management terminal of the unqualified automated verification line, and the on-site administrator is notified to conduct on-site investigation;

[0016] (4) an inspection system, which is used to construct a fault inspection and automatic maintenance plan for the current automated verification line according to the operation fault information, and send the fault inspection plan to the management terminal of the current automated verification line, so that the management terminal automatically executes the fault inspection and automatic maintenance plan;

[0017] (5) a management terminal, used to receive and display the operation and maintenance log A, or receive and execute the fault inspection and automatic maintenance plan;

[0018] The quality detection system, intelligent fault detection system and patrol inspection system are respectively deployed on the electric energy metering experiment scheduling management platform;

[0019] The electric energy metering experiment scheduling management platform is communicatively connected with the management terminal.

[0020] As an optional implementation scheme of the present application, optionally, the electric energy metering experiment scheduling management platform is further provided with:

[0021] Several high-speed network cables for communication connection with the management terminal of each automated verification line;

[0022] The electric energy metering experiment scheduling management platform and the management terminal of each automated verification line communicate data via a high-speed network cable.

[0023] As an optional implementation scheme of the present application, optionally, the electric energy metering experiment scheduling management platform is further provided with:

[0024] The Oracle database is used to store each verification data of each automated verification line in each metrological verification and save it as the verification big data of each automated verification line.

[0025] As an optional implementation scheme of the present application, optionally, the electric energy metering experiment scheduling management platform is further provided with:

[0026] The message middleware is used to realize the message distribution between the electric energy metering experiment scheduling management platform and the management terminal of each automated verification line according to the message queue mechanism.

[0027] As an optional implementation scheme of the present application, optionally, the method for generating the fault AI detection model includes:

[0028] Collecting the verification big data of each of the automated verification lines, including the operation and maintenance log of each automated verification line in each metrological verification;

[0029] Performing feature engineering on the verification big data, extracting fault log features corresponding to the operation and maintenance log, and forming a fault log feature set;

[0030] Importing the fault log feature set into a preset RNN model, performing feature learning, and training to generate a corresponding RNN fault recognition model;

[0031] Randomly collect a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verify the RNN fault identification model;

[0032] If the verification is qualified, the corresponding fault AI detection model is obtained, and the fault AI detection model is deployed in the fault intelligent detection system.

[0033] As an optional implementation scheme of the present application, optionally, in the step of collecting the verification big data of each of the automated verification lines, including the operation and maintenance log of each automated verification line in each metrological verification, the following model is adopted:

[0034] Assume that the data set is D = {d1, d2, ..., d n}, where each d i Indicates the i-th record, containing the following information: Quality inspection result q i (pass / fail) and operation and maintenance logsi ;

[0035] Where n represents the total number of records; d i represents the i-th record, including the quality inspection result and the corresponding operation and maintenance log; q i Indicates the quality inspection result of the i-th record; l i Represents the operation and maintenance log of the i-th record;

[0036] In the step of performing feature engineering on the verification big data, extracting fault log features of the corresponding operation and maintenance logs, and forming a fault log feature set, the following model is used:

[0037] Assume that the feature set is F = {f1,f2,…,f m}, where f j represents the jth feature;

[0038] Where m represents the total number of features; f j represents the jth feature, which may include time features, frequency features, and abnormal patterns;

[0039] In the step of importing the fault log feature set into the preset RNN model, performing feature learning, and training to generate the corresponding RNN fault recognition model, the following model is used:

[0040] Assume that the training set is T = {(f1,y1),(f2,y2)…,(f k ,y k )}, where y i Indicates the corresponding fault label;

[0041] Where k represents the number of samples; y i represents the fault label of the i-th sample (e.g., normal, fault type A, fault type B);

[0042] In the step of randomly collecting a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verifying the RNN fault identification model, the output of the RNN can be expressed as:

[0043] h t =σ(W h h t-1 +W x h x +b h );

[0044] y t =W h h t +b y ;

[0045] Among them, h trepresents the hidden state at time step t; x t Input feature vector; W h , W x , W y represents the weight matrix; b h , b y represents the bias term; σ represents the activation function (such as tanh or ReLU); t represents the time step; h t Represents the current hidden state, capturing sequence information; y t Represents the output of the current time step, indicating the prediction of the fault type.

[0046] As an optional implementation scheme of the present application, optionally, the management terminal is also used for the on-site administrator to feedback the on-site investigation results to the electric energy metering experiment scheduling management platform after the on-site investigation; or after the fault inspection and automatic maintenance plan are executed, the execution results are fed back to the electric energy metering experiment scheduling management platform.

[0047] On the other hand, the present application proposes an application method of an alarm device of an electric energy metering experiment scheduling management platform, comprising the following steps:

[0048] The electric energy metering experiment dispatching management platform reads each verification data of each automated verification line in this metering verification, including quality verification results and corresponding operation and maintenance logs;

[0049] The quality inspection system reads each piece of inspection data by using a preset quality inspection model, and determines in turn whether the quality inspection result in each piece of inspection data is qualified:

[0050] If qualified, give up;

[0051] If it fails, the operation and maintenance log A corresponding to the unqualified quality inspection result is imported into the fault intelligent detection system:

[0052] The intelligent fault detection system uses the pre-deployed fault AI detection model to identify and analyze whether the operation and maintenance log A has the following fault log characteristics:

[0053] If a fault log feature appears, corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log feature;

[0054] If no fault log feature appears, the operation and maintenance log A is sent to the management terminal of the unqualified automated verification line, and the on-site administrator is notified to conduct on-site investigation; after the on-site investigation, the on-site administrator feeds back the on-site investigation results to the electric energy metering experiment scheduling management platform through the management terminal;

[0055] The inspection system constructs a fault inspection and automatic maintenance plan for the current automated calibration pipeline based on the operating fault information, and sends the fault inspection plan to the management terminal of the current automated calibration pipeline, which is then automatically executed by the management terminal. After the fault inspection and automatic maintenance plan are executed, the management terminal feeds back the execution results to the electric energy metering experiment scheduling management platform.

[0056] In another aspect, the present application further provides an electronic device, comprising:

[0057] processor;

[0058] a memory for storing processor-executable instructions;

[0059] Wherein, the processor is configured to implement the application method when executing the executable instructions.

[0060] Technical effects of the present invention:

[0061] The present invention can replace manual analysis, and perform statistics on the operation and maintenance logs of the electric energy metering experiment and the verification pass rate of the verification pipeline, so as to perform fault analysis and automatic alarm supervision of the experimental operation quality, avoid spending a lot of time browsing logs and statistical data, reduce the huge workload and time-consuming, and quickly provide effective analysis results for the quality operation of the electric energy metering experiment, thereby improving the verification efficiency. The present invention can intelligently identify whether the operation and maintenance log A has a fault AI detection model with fault log characteristics, so as to perform fault analysis, thereby replacing manual intelligent troubleshooting, saving labor, and improving troubleshooting efficiency.

[0062] Further features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate exemplary embodiments, features, and aspects of the disclosure and, together with the description, serve to explain the principles of the disclosure.

[0064] Figure 1 The following is a schematic diagram of the application architecture of the electric energy metering experiment scheduling management platform of the present invention;

[0065] Figure 2 It is a schematic diagram showing the structure of the device of the present invention;

[0066] Figure 3 It is a schematic diagram showing another device structure of the present invention;

[0067] Figure 4 Shown is a schematic diagram of the network structure of the RNN of the present invention;

[0068] Figure 5 It is a schematic diagram showing the application of the electronic device of the present invention. DETAILED DESCRIPTION

[0069] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0070] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0071] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present disclosure can also be implemented without certain specific details. In some examples, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the main purpose of the present disclosure.

[0072] Example 1

[0073] like Figure 1 The figure shows the application architecture of the electric energy metering experiment scheduling management platform. The electric energy metering experiment scheduling management platform system is deployed at the first level of the electric energy metering company. It integrates with the marketing system, data middle platform, six-line and one-database system, standard device and laboratory equipment system, intelligent building system, and video surveillance system through application integration, data integration, and interface integration.

[0074] From the management perspective, the electric energy metering experiment scheduling management platform mainly includes four levels of content. The first is the basic layer, which realizes interconnection with the existing management system through the interface program, including data integration with the marketing platform, "six lines and one warehouse" system data integration, safety environment system, video system and other data integration; the second is the data layer, which provides data storage; the third is the application layer, including the management of job scheduling, intelligent warehousing, internal logistics and automatic calibration of the production site, replacing the manual operation mode, realizing automated production and intelligent scheduling, as well as the preparation and tracking of production plans, tracking of production operation processes, the circulation of asset experiment business, integrated management and control of second, third and fourth level warehouses, and other applications to achieve standardized and lean management; the fourth is the support layer, which summarizes the basic business data generated on site to form analytical data to provide scientific data support for decision-making; the fifth is the display layer, which forms a visual and graphical comprehensive display through the analysis results of the metering data.

[0075] The electric energy metering experiment dispatching management platform is based on data interaction with six lines and one warehouse. On the one hand, it realizes the metering data integration system, experimental scheduling services, metering experiment calibration business applications, statistical analysis and auxiliary decision-making, and realizes the effective dispatch and comprehensive integration of the internal calibration lines and three-dimensional warehouses of the electric energy metering company; on the other hand, it realizes the second, third and fourth level warehouse management functions, thus laying a good foundation for the refined management of the entire life cycle of metering equipment assets.

[0076] The electric energy metering experiment scheduling management platform can be connected to each automated calibration line (also called "calibration line") through a communication interface such as a high-speed network cable, and specifically communicated with the management terminal of each calibration line.

[0077] When connecting, because the electric energy metering experiment scheduling management platform needs to manage multiple automated verification lines, an industrial gateway is needed to route and transmit data for each verification line. Figure 2 As shown, the electric energy metering experiment scheduling management platform communicates with the management terminals of each calibration line (such as the industrial computers or control hosts of each calibration line, which can execute various inspection tasks and production tasks issued by the host computer (platform), collect operation and maintenance logs and share them with the background, and also upload the calibration data results of the electric energy meter equipment to the background) through an industrial gateway to realize various data transmission and control.

[0078] The management backend uses the industrial gateway to achieve centralized control of each terminal, ensuring smooth data interaction and command execution, thereby optimizing overall operational efficiency and monitoring capabilities. Through this integrated system, administrators can obtain real-time operating status information of each terminal, including but not limited to equipment performance parameters, production process data, and potential fault warnings. Using the powerful data processing capabilities of the industrial gateway, this information is quickly parsed and converted into easy-to-understand charts or reports, providing solid data support for decision-making.

[0079] At the control level, the management backend provides an intuitive operation interface, allowing administrators to remotely adjust the operating parameters of terminal equipment, such as speed, temperature, pressure, etc., through simple instructions or preset strategies to meet different production needs or respond to emergencies. At the same time, the system also supports batch operation functions, which greatly improves management efficiency.

[0080] In addition, the management backend also integrates a powerful fault diagnosis and recovery mechanism. When an abnormality is detected in a terminal device, the system can automatically trigger the diagnosis process, quickly locate the problem, and give corresponding solution suggestions or automatically perform recovery operations to reduce downtime and ensure production continuity.

[0081] In general, the management background's control over each terminal through the industrial gateway not only improves the intelligence level of the production process, but also enhances the flexibility and reliability of the system, laying a solid foundation for the sustainable development of the enterprise.

[0082] The technical contents of the present invention will be described below.

[0083] As attached Figure 2 As shown, on one hand, the present application proposes an alarm device for an electric energy metering experiment scheduling management platform, comprising:

[0084] (1) The electric energy metering experiment dispatching management platform (referred to as the "backstage") is used to read the various verification data of each automated verification line (referred to as the "verification line") in this metering verification, including the quality verification results and the corresponding operation and maintenance logs;

[0085] (2) A quality inspection system, which is used to read each piece of inspection data using a preset quality inspection model, and to determine in turn whether the quality inspection result in each piece of inspection data is qualified:

[0086] If qualified, give up;

[0087] If it fails, the operation and maintenance log A corresponding to the unqualified quality inspection result is imported into the fault intelligent detection system:

[0088] (3) A fault intelligent detection system, which is used to identify and analyze whether the operation and maintenance log A has fault log characteristics through a pre-deployed fault AI detection model:

[0089] If a fault log feature appears, corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log feature;

[0090] If no fault log feature appears, the operation and maintenance log A is sent to the management terminal of the unqualified automated verification line, and the on-site administrator is notified to conduct on-site investigation;

[0091] (4) an inspection system, which is used to construct a fault inspection and automatic maintenance plan for the current automated verification line according to the operation fault information, and send the fault inspection plan to the management terminal of the current automated verification line, so that the management terminal automatically executes the fault inspection and automatic maintenance plan;

[0092] (5) a management terminal (hereinafter referred to as the “terminal”), which is used to receive and display the operation and maintenance log A, or receive and execute the fault inspection and automatic maintenance plan;

[0093] The quality detection system, intelligent fault detection system and patrol inspection system are respectively deployed on the electric energy metering experiment scheduling management platform;

[0094] The electric energy metering experiment scheduling management platform is communicatively connected with the management terminal.

[0095] Each calibration line is equipped with its own industrial computer or control host (management end, control end or management terminal), which enables the on-site administrator to manage each calibration line based on the management terminal, including quality inspection of the electric energy meters tested on each calibration line and recording of operation and maintenance logs.

[0096] The management terminal can record the quality inspection results of each electric energy metering device and share (read) them to the background. At the same time, it can record the operation and maintenance logs of each inspection line, including the recording and preservation of the operation and maintenance parameters of each device, each inspection sensor, inspection system, inspection component (determined in combination with the composition system of the inspection line), etc. in the inspection line, and generate corresponding inspection operation and maintenance logs.

[0097] Therefore, during the real-time calibration process, the management terminal can record the calibration data of each calibration line it controls, and can record and save the quality calibration results of the electric energy metering equipment and the corresponding operation and maintenance logs, and then share them to the background. In this process, the background will actively read the calibration data in the management terminal of each production line, perform alarm monitoring, identify and locate the faults on each calibration line, and generate corresponding on-site troubleshooting notifications or corresponding on-site inspection plans to troubleshoot the calibration line.

[0098] If an on-site inspection plan is generated, the generated automatic fault maintenance plan can be sent to the management terminal of the corresponding calibration line, so that the management terminal of the corresponding calibration line can automatically execute the maintenance plan to handle operation and maintenance failures, thereby realizing automated inspection and maintenance and automatic maintenance.

[0099] For different types of faults, the background can learn to generate corresponding fault inspection and automatic maintenance plans, and can send the corresponding fault maintenance code to the management terminal of the vehicle where the faulty inspection line is located, so that the terminal can execute the corresponding fault maintenance code and program for automatic repair. For details, you can refer to the existing automatic fault maintenance, automatic patching, automatic troubleshooting and other programs generated by the background of smart terminals or PCs.

[0100] In order to realize the generation of corresponding automatic fault maintenance procedures based on the fault log characteristics, a systematic process can be designed in the background to parse the log, identify the fault mode, and automatically generate the corresponding maintenance scripts or instructions. The steps are as follows:

[0101] First, a fault log collection and analysis system needs to be established, which can collect log files from various system components in real time or at regular intervals. These log files record important information such as system operation status, error information, and abnormal events.

[0102] Next, the collected fault logs are preprocessed, including log cleaning (removing duplicate, irrelevant, or malformed log entries), log parsing (converting log text into structured data, such as key-value pairs), and log aggregation (merging logs of the same or similar types together).

[0103] The pre-processed log data is then analyzed using machine learning or data mining techniques to identify common failure modes and abnormal behaviors. This typically involves feature extraction (extracting features from log data that help distinguish normal from abnormal states), model training (training classification or clustering models using historical failure data), and pattern recognition (applying trained models to classify or cluster new log data to identify potential failures).

[0104] Once the failure modes are identified, corresponding automated fault maintenance procedures can be generated based on the characteristics of these modes.

[0105] This usually involves the following steps:

[0106] 1. Determine the scope and severity of the fault: Evaluate the impact of the fault on system performance and business operations by analyzing fault logs and system status information.

[0107] 2. Develop a troubleshooting strategy: Select appropriate troubleshooting measures based on the nature and severity of the fault, such as restarting services, restoring backups, adjusting configuration parameters, etc.

[0108] 3. Write automated scripts or instructions: Convert the fault handling strategy into an executable automated script or instruction set so that the corresponding processing operations can be automatically performed when a fault occurs.

[0109] 4. Deployment and testing: Deploy the written automated fault maintenance program to the production environment and conduct sufficient testing to ensure that it can run correctly under various fault scenarios.

[0110] 5. Continuous optimization and improvement: Based on actual application results and user feedback, continuously optimize and improve the automated fault maintenance procedures to improve their accuracy and efficiency.

[0111] Through the above steps, the goal of automatically generating corresponding automated fault maintenance programs according to the fault log characteristics can be achieved in the background, thereby improving the automation level of system operation and maintenance and the fault response speed.

[0112] In the quality inspection system, a quality inspection model is deployed, which can make qualified judgments on the quality inspection results of each inspection data read in the background.

[0113] The quality inspection model learns various quality inspection rules (set by the administrator) through a deep learning model to generate a corresponding quality inspection AI model for intelligent qualification judgment of various inspection data (can be combined with CNN and RNN models for model training).

[0114] The quality inspection model works by analyzing and learning a series of predefined quality inspection rules to evaluate the eligibility of given data. These rules may cover data integrity, accuracy, consistency, format requirements, logical constraints, etc. When the model receives the data to be inspected, it compares and verifies the data one by one with the learned rules.

[0115] During the validation process, the model checks whether the data meets the requirements of each rule. If all data items pass the test of all rules, the model will judge the data as qualified; if any one or more data items fail to meet the requirements of the rules, the model will judge the data as unqualified, and may further point out which specific data items or rules have failed the validation so that corrections or improvements can be made.

[0116] It is worth noting that the accuracy and effectiveness of the quality detection model largely depends on the comprehensiveness and accuracy of the rules it learns. Therefore, when designing and implementing the quality detection model, it is necessary to ensure that the rules formulated can fully cover all aspects of data quality, and these rules themselves should be clear, unambiguous, and unambiguous. At the same time, as the business develops and changes, the quality detection model also needs to regularly update and adjust its rule set to adapt to new data quality requirements and challenges. When continuing to explore the application and optimization of the quality detection model, according to the automation and intelligent characteristics of the model, the quality detection model can not only automatically execute the detection process, reduce manual intervention, and improve detection efficiency, but also continuously optimize itself through machine learning algorithms to improve detection accuracy and adaptability.

[0117] As data continues to accumulate and process, the quality detection model can learn and identify patterns, anomalies, and trends in the data, and then fine-tune existing detection rules or propose new rules. This self-evolutionary ability enables the model to more accurately identify data quality issues and provide strong support for data governance.

[0118] In addition, in order to ensure the effectiveness and reliability of the quality detection model, it is also necessary to regularly verify and evaluate the model. This includes testing the model using a data set with known quality results to verify the accuracy of its judgment results; at the same time, it is also necessary to pay attention to the performance of the model when processing new types of data or abnormal data, so as to discover and solve problems in a timely manner.

[0119] In addition to the technical aspects mentioned above, the successful application of the quality detection model also requires support and cooperation from within the organization. This includes clarifying data quality goals and standards, establishing data quality management processes and mechanisms, and cultivating employees' attention and awareness of data quality. Only when the entire organization regards data quality as a key success factor can the quality detection model truly play its value and promote the continuous improvement of the organization's data governance level. The administrator will specifically formulate the corresponding rules and conduct in-depth model training.

[0120] For unqualified calibration data, it is necessary to further determine whether the operation and maintenance log A has fault log characteristics, and further check the calibration line that fails the test (the default calibration line is normal. If an unqualified test result occurs, the calibration line may have a calibration fault, such as changes in detection components, sensors, detection working parameters, environmental parameters, program logic, etc., or a calibration unit fails). Therefore, further fault detection and maintenance are required.

[0121] Therefore, we use a fault AI detection model that can intelligently identify whether the operation and maintenance log A contains fault log features to perform fault analysis, thereby replacing manual intelligent troubleshooting, saving manpower and improving troubleshooting efficiency.

[0122] For details, see the generation principle of the fault AI detection model below.

[0123] As an optional implementation scheme of the present application, optionally, the electric energy metering experiment scheduling management platform is further provided with:

[0124] Several high-speed network cables for communication connection with the management terminal of each automated verification line;

[0125] The electric energy metering experiment scheduling management platform and the management terminal of each automated verification line communicate data via a high-speed network cable.

[0126] The management terminal of each automated verification line communicates data via a high-speed network cable. Specifically, the management terminal on the verification line side can connect to the high-speed network cable and perform terminal communication control with the background server.

[0127] In terminal control, high-speed network cable communication has shown many significant advantages, which make it an important means of communication between modern computer systems and devices. The following is a detailed description of the advantages of high-speed network cable communication in terminal control:

[0128] First of all, high-speed network cable communication has the ability to transmit high-speed data. With the continuous evolution of high-speed network cable standards, from 1.5Mbps and 12Mbps of high-speed network cable 1.1, to 480Mbps of high-speed network cable 2.0, and then to the significant improvement of high-speed network cable 3.0 and higher versions, high-speed network cable interfaces can support high-speed data transmission and meet the high requirements for data transmission rate in terminal control. This high speed makes high-speed network cables perform well in applications such as real-time data transmission and video streaming transmission.

[0129] Secondly, high-speed network cable communication has a high degree of flexibility and freedom. The high-speed network cable interface supports a variety of topologies, allowing users to expand the number of high-speed network cables by connecting high-speed network cable hubs, and then connect more peripheral devices. This flexible connection method enables high-speed network cables to easily cope with complex device connection requirements in terminal control. At the same time, high-speed network cables also support hot-swap functions. Users can connect or disconnect high-speed network cable devices at any time without shutting down the system or restarting the computer, which greatly improves the ease of use and flexibility of the device.

[0130] In addition, high-speed network cable communication also has the feature of built-in power supply. The high-speed network cable bus can provide 5V voltage and a maximum current supply of 500mA (high-speed network cable 2.0) or higher (such as 900mA for high-speed network cable 3.0) for the devices connected to it. This feature allows high-speed network cable devices to work normally without additional power supply, reducing the cost and complexity of equipment use. At the same time, for high-power devices, high-speed network cables also support power supply through hubs or external power supplies to meet their higher power requirements.

[0131] Furthermore, high-speed network cable communication supports multiple transmission modes, including control transmission, interrupt transmission, synchronous transmission, and batch transmission. These transmission modes have their own characteristics and can meet the data transmission requirements in different application scenarios. For example, control transmission is suitable for operations such as device configuration and status query; interrupt transmission is suitable for devices that need to send small amounts of data regularly; synchronous transmission is suitable for real-time data transmission such as audio and video; and batch transmission is suitable for the transmission of large amounts of data. These diverse transmission modes enable high-speed network cables to flexibly respond to various data transmission requirements in terminal control.

[0132] Finally, high-speed network cable communication also has the advantages of unified standards and wide support. As a widely used interface standard, high-speed network cable has been supported by many hardware and software manufacturers. This makes high-speed network cable equipment very popular in the market, and users can easily find high-speed network cable equipment and accessories that suit their needs. At the same time, the uniformity of high-speed network cable standards also reduces compatibility issues between devices, allowing high-speed network cable equipment produced by different manufacturers to be compatible and communicate with each other.

[0133] In summary, high-speed network cable communication has shown significant advantages in terminal control, such as high-speed data transmission, high-degree-of-freedom connection, built-in power supply, support for multiple transmission modes, unified standards and wide support. These advantages make high-speed network cables an important tool for communication between modern computer systems and devices.

[0134] As an optional implementation scheme of the present application, optionally, the electric energy metering experiment scheduling management platform is further provided with:

[0135] The Oracle database is used to store each verification data of each automated verification line in each metrological verification and save it as the verification big data of each automated verification line.

[0136] Implementing dynamic data storage in an Oracle database mainly involves table design, data insertion, update, query, and deletion operations. Dynamic data storage means that data can be added, modified, or deleted in real time according to application requirements without the need to predefine a fixed data structure. The following is a simplified description of how to handle dynamic data storage in MySQL:

[0137] First, when designing the table structure, you need to consider the possible addition of fields or data changes in the future. A common approach is to use "extensible fields" (such as JSON type fields or TEXT type fields for storing serialized data), which allows you to store additional information without modifying the table structure. For example, you can create a table with basic information and add a JSON type field to store additional data that may change frequently.

[0138] When inserting data, you can directly insert dynamic data into the corresponding field as a JSON string, or insert basic data in the traditional column value method, and save the dynamic data in other ways (such as serializing and storing it in a TEXT field).

[0139] For data updates, you can use standard UPDATE statements for basic data fields. For dynamic data stored in JSON or TEXT fields, you may need to use the JSON functions provided by MySQL (if the data is stored in a JSON field) or deserialize the data first, modify it, and then serialize it back to the database.

[0140] When querying dynamic data, you can use MySQL's JSON functions (if applicable) to extract and process data stored in JSON fields, or by writing more complex query logic to process serialized data stored in TEXT fields.

[0141] Finally, when you need to delete data, you can use the standard DELETE statement to delete the entire row of data from the table, including the dynamic data.

[0142] In the Oracle database, each calibration data of each automated calibration line in each metrological calibration will be saved independently, so calibration big data can be generated for each calibration line.

[0143] After long-term data storage, each calibration line will record corresponding calibration big data, which can be used for big data analysis and data learning training, providing big data support for the fault AI detection model.

[0144] As an optional implementation scheme of the present application, optionally, the electric energy metering experiment scheduling management platform is further provided with:

[0145] The message middleware is used to realize the message distribution between the electric energy metering experiment scheduling management platform and the management terminal of each automated verification line according to the message queue mechanism.

[0146] The message queue mechanism is an efficient, scalable and reliable distributed message publishing / subscribing system. It is mainly used to process large data streams, especially in real-time data pipelines and stream processing applications. You can understand it in conjunction with the message queue mechanism. The specific message sending and receiving mechanism or the corresponding middleware can be deployed by the user.

[0147] As an optional implementation scheme of the present application, optionally, the method for generating the fault AI detection model includes:

[0148] Collecting the verification big data of each of the automated verification lines, including the operation and maintenance log of each automated verification line in each metrological verification;

[0149] Performing feature engineering on the verification big data, extracting fault log features corresponding to the operation and maintenance log, and forming a fault log feature set;

[0150] Importing the fault log feature set into a preset RNN model, performing feature learning, and training to generate a corresponding RNN fault recognition model;

[0151] Randomly collect a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verify the RNN fault identification model;

[0152] If the verification is qualified, the corresponding fault AI detection model is obtained, and the fault AI detection model is deployed in the fault intelligent detection system;

[0153] Otherwise, repeat the above steps.

[0154] As a further implementation of the optional implementation scheme of the above-mentioned fault AI detection model generation method, in the step of collecting the verification big data of each of the automated verification lines, including the operation and maintenance log of each automated verification line in each metrological verification, the following model is adopted:

[0155] Assume that the data set is D = {d1, d2, ..., d n}, where each d i Indicates the i-th record, containing the following information: Quality inspection result q i (pass / fail) and operation and maintenance logs i ;

[0156] Where n represents the total number of records; d i represents the i-th record, including the quality inspection result and the corresponding operation and maintenance log; q i Indicates the quality inspection result of the i-th record; l i Represents the operation and maintenance log of the i-th record;

[0157] This step is to collect calibration big data from each automated calibration line, including operation and maintenance logs and quality calibration results in each metrological calibration; this process provides basic data for subsequent feature extraction and model training, ensuring the integrity and accuracy of the data.

[0158] In the step of performing feature engineering on the verification big data, extracting fault log features of the corresponding operation and maintenance logs, and forming a fault log feature set, the following model is used:

[0159] Assume that the feature set is F = {f1,f2,…,f m}, where f j represents the jth feature;

[0160] Where m represents the total number of features; f j represents the jth feature, which may include time features, frequency features, and abnormal patterns;

[0161] This step is to extract features from the collected operation and maintenance logs and construct a fault log feature set. The result of feature engineering will be used as the input data set X for subsequent model training to ensure that the model can learn effective features.

[0162] In the step of importing the fault log feature set into the preset RNN model, performing feature learning, and training to generate the corresponding RNN fault recognition model, the following model is used:

[0163] Assume that the training set is T = {(f1,y1),(f2,y2),…,(f k ,y k )}, where y i Indicates the corresponding fault label;

[0164] Where k represents the number of samples; y i represents the fault label of the i-th sample (e.g., normal, fault type A, fault type B);

[0165] This step associates the extracted features with the corresponding fault labels to form the training set required for supervised learning; the training set T is the basis for subsequent model training to ensure that the model can predict the fault type based on the features.

[0166] In the step of randomly collecting a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verifying the RNN fault identification model, the output of the RNN can be expressed as:

[0167] h t =σ(W h h t-1 +W x h x +b h );

[0168] y t =W h h t +b y ;

[0169] Among them, h t represents the hidden state at time step t; x t Input feature vector; W h , W x , W y represents the weight matrix; b h , b y represents the bias term; σ represents the activation function (such as tanh or ReLU); t represents the time step; h t Represents the current hidden state, capturing sequence information; y t Represents the output of the current time step, indicating the prediction of the fault type.

[0170] This step imports the fault log feature set into the preset recursive neural network (RNN) model for feature learning and model training; through training, the RNN model can learn the features and patterns in the time series data to provide support for subsequent fault identification.

[0171] In the step of randomly collecting a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verifying the RNN fault identification model, the following model is used:

[0172] 1. Construction of validation set:

[0173] Let the validation set be V = {(x1,y1),(x2,y2),…,(x m ,y m )},in:

[0174] x i The feature vector representing the i-th operation and maintenance log; y i Indicates the actual fault label of the i-th operation and maintenance log.

[0175] 2. Output of RNN model:

[0176] For each input feature vector x i , the RNN model generates the prediction output

[0177]

[0178] in:

[0179] f is the trained RNN model function.

[0180] θ represents the parameters (weights and biases) of the model.

[0181] 3. Loss function:

[0182] The cross entropy loss function is used to evaluate the prediction performance of the model:

[0183]

[0184] in:

[0185] L(θ) represents the loss value; y i is the true label (1 for failure, 0 for normal; represents the predicted probability of the model.

[0186] 4. Evaluation indicators:

[0187] Calculate the model's accuracy, precision, recall, and F1-score to comprehensively evaluate the model performance.

[0188] Accuracy:

[0189]

[0190] Accuracy:

[0191]

[0192] Recall:

[0193]

[0194] F1-score:

[0195]

[0196] in,

[0197] TP: number of true positives (number of faults correctly predicted by the model);

[0198] TN: number of true negative examples (number of examples correctly predicted by the model as normal);

[0199] FP: number of false positives (number of faults incorrectly predicted by the model);

[0200] FN: The number of false negatives (number of examples incorrectly predicted as normal by the model).

[0201] The validation set V is the basis for evaluating model performance. By calculating the loss function L(θ) and the evaluation index, we can understand the performance of the model on unseen data and then judge the generalization ability of the model. If the model performs well on the validation set, the training process can be considered successful and the fault AI detection model can be deployed in the fault intelligent detection system for practical application.

[0202] The following is another implementation of the fault AI detection model.

[0203] The fault AI detection model includes data collection, data preprocessing, feature engineering, model selection and training, model evaluation and optimization, and model deployment. The specific construction process includes:

[0204] 1. Data Collection

[0205] First, you need to collect data sets related to quality inspection. These data sets should contain various quality indicators of the product, test results, and the final quality judgment results (qualified or unqualified). Data sources can include historical records, user feedback, test results, sensor data, and other channels. Ensure the diversity and sufficiency of the data so that the model can learn the quality characteristics in different scenarios.

[0206] Because the fault AI detection model is needed to identify and analyze whether the operation and maintenance log A has fault log characteristics:

[0207] If a fault log feature appears, corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log feature;

[0208] If no fault log feature appears, the operation and maintenance log A is sent to the management terminal of the unqualified automated verification line to notify the on-site administrator to conduct on-site inspection.

[0209] Therefore, the collected verification big data of each automated verification pipeline needs to include operation and maintenance logs of fault logs. The collected operation and maintenance logs need to be screened to avoid excessive training due to excessive data set volume, which affects the training progress of the model.

[0210] In order to screen the collected verification big data of each automated verification line, this department adopts an automated data screening and cleaning method for cleaning:

[0211] Through LLM, based on the input fault log screening prompt words, the operation and maintenance logs containing fault logs are screened from the operation and maintenance log library. Specifically:

[0212] The administrator first constructs keywords and filtering logic for filtering operation and maintenance logs containing fault logs.

[0213] Generate corresponding fault log screening prompt words based on keywords and screening logic;

[0214] Log in to the third party through the LLM API interface deployed on the platform, request to call LLM, and enter the fault log screening prompt word into LLM;

[0215] Generate data collection instructions, activate LLM, and let LLM filter and output operation and maintenance logs containing fault logs from the log library based on prompts.

[0216] Therefore, through the above method, the data collection efficiency can be improved.

[0217] The principle of LLM large language model data retrieval: LLM uses deep learning technology to train on large amounts of text data, master the rules of word association, and thus be able to predict and generate reasonable language output.

[0218] Working principle:

[0219] Word vector representation: LLM converts the words in the input text into a long list of word vectors. These vectors are represented in a "word space", where words with similar meanings are closer in space.

[0220] Transformer architecture: This architecture processes the input content through self-attention mechanisms, captures implicit correlation information, and predicts the probability of the next element appearing.

[0221] Data retrieval: Based on the trained model and the input prompt words, LLM can generate text output that is relevant to the prompt words and semantically coherent, realizing data retrieval function.

[0222] LLM's data retrieval capability relies on its powerful language processing capabilities and learning of massive text data. Through deep learning technology and complex neural network structures, LLM can capture deep and abstract feature representations in the data, thereby achieving efficient data retrieval.

[0223] 2. Data Preprocessing

[0224] The collected data often needs to be preprocessed to meet the requirements of model training. The preprocessing steps include:

[0225] 1. Clean data: handle missing values, outliers and other issues to ensure data integrity and accuracy.

[0226] 2. Convert data: Convert non-numeric data to numeric data so that the model can process it. For example, convert the quality level described in text into a numeric label.

[0227] 3. Standardization or normalization: Make sure the data is on the same scale so that the model can fairly compare the importance of different features.

[0228] 3. Feature Engineering

[0229] Feature engineering is one of the key steps in building a quality inspection model. Features that are highly correlated with the evaluation target are screened out through statistical analysis, correlation analysis, etc. These features should be able to accurately reflect the quality characteristics of the product and be easy to be learned and understood by the model.

[0230] IV. Model Selection and Training

[0231] Choose the appropriate model type based on the characteristics and requirements of the problem. Common models include linear regression, decision tree, random forest, support vector machine (SVM), neural network, etc. For complex data and tasks, you can consider using deep learning models such as convolutional neural network (CNN) or recurrent neural network (RNN).

[0232] As attached Figure 4 The figure shows the network structure of RNN. The selected RNN model and feature set are used for training. During the training process, the RNN model parameters, such as learning rate, number of iterations, etc., need to be adjusted continuously to optimize the performance of the RNN model. At the same time, methods such as cross-validation can be used to evaluate the generalization ability of the model to ensure that the model performs well on unseen data.

[0233] 5. Model Evaluation and Optimization

[0234] Use evaluation indicators (such as accuracy, precision, recall, F1 score, AUC value, MSE, MAE, etc.) to evaluate the performance of the model. If the model performance is not ideal, you can return to the feature engineering or model selection stage for adjustment and optimization. For example, you can try different feature combinations or choose different model types for training.

[0235] 6. Model Deployment

[0236] Deploy the trained model to actual application scenarios for quality testing. In actual applications, new data needs to be collected regularly to update the model to maintain its prediction accuracy and effectiveness. At the same time, the model also needs to be monitored and maintained to ensure that it runs stably and meets business needs.

[0237] The above training data set can be read from the background database. When performing feature engineering, some neural networks can be used for feature extraction, and can be combined with manual annotation to perform feature recognition. The specific work can be completed by the administrator.

[0238] The training scheme of the above-mentioned model using RNN should be understood in conjunction with the neural network structure and application of RNN, and will not be described in detail in this embodiment.

[0239] As an optional implementation scheme of the present application, optionally, the management terminal is also used for the on-site administrator to feedback the on-site investigation results to the electric energy metering experiment scheduling management platform after the on-site investigation; or after the fault inspection and automatic maintenance plan are executed, the execution results are fed back to the electric energy metering experiment scheduling management platform.

[0240] After the administrator conducts on-site inspection, he or she can provide feedback to the backend through the management terminal.

[0241] If the management terminal executes the automatic troubleshooting and troubleshooting procedures generated by the background, it can report to the background after completing the inspection and automatic fault clearance procedures. The specific inspection plan can be automatically generated by the inspection system according to the identified fault feature type (the background is pre-configured with the corresponding type of inspection plan and automatic troubleshooting program, that is, fault inspection and automatic maintenance plan), and the corresponding fault inspection and automatic maintenance plan is retrieved and sent to the management terminal for execution, so as to complete the automated fault inspection, upgrade and troubleshooting work.

[0242] Obviously, those skilled in the art should understand that the implementation of all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. Those skilled in the art can understand that the implementation of all or part of the processes in the above-mentioned embodiments can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated as: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0243] Example 2

[0244] Based on the implementation principle of Example 1, on the other hand, the present application proposes an application method of an alarm device of an electric energy metering experiment scheduling management platform, comprising the following steps:

[0245] The electric energy metering experiment dispatching management platform reads each verification data of each automated verification line in this metering verification, including quality verification results and corresponding operation and maintenance logs;

[0246] The quality inspection system reads each piece of inspection data by using a preset quality inspection model, and determines in turn whether the quality inspection result in each piece of inspection data is qualified:

[0247] If qualified, give up;

[0248] If it fails, the operation and maintenance log A corresponding to the unqualified quality inspection result is imported into the fault intelligent detection system:

[0249] The intelligent fault detection system uses the pre-deployed fault AI detection model to identify and analyze whether the operation and maintenance log A has the following fault log characteristics:

[0250] If a fault log feature appears, corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log feature;

[0251] If no fault log feature appears, the operation and maintenance log A is sent to the management terminal of the unqualified automated verification line, and the on-site administrator is notified to conduct on-site investigation; after the on-site investigation, the on-site administrator feeds back the on-site investigation results to the electric energy metering experiment scheduling management platform through the management terminal;

[0252] The inspection system constructs a fault inspection and automatic maintenance plan for the current automated calibration pipeline based on the operating fault information, and sends the fault inspection plan to the management terminal of the current automated calibration pipeline, which is then automatically executed by the management terminal. After the fault inspection and automatic maintenance plan are executed, the management terminal feeds back the execution results to the electric energy metering experiment scheduling management platform.

[0253] Please understand and implement the above steps in conjunction with Example 1.

[0254] The modules or steps of the present invention described above can be implemented by a general-purpose computing system, they can be concentrated on a single computing system, or distributed on a network composed of multiple computing systems, and optionally, they can be implemented by a program code executable by a computing system, so that they can be stored in a storage system and executed by the computing system, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0255] Example 3

[0256] like Figure 5 As shown, further, in another aspect, the present application also proposes an electronic device, including:

[0257] processor;

[0258] a memory for storing processor-executable instructions;

[0259] Wherein, the processor is configured to implement the application method described in Example 2 when executing the executable instructions.

[0260] The electronic device according to the embodiment of the present disclosure includes a processor and a memory for storing instructions executable by the processor, wherein the processor is configured to implement any of the above-mentioned application methods when executing the executable instructions.

[0261] Here, it should be noted that the number of processors can be one or more. At the same time, the electronic device of the embodiment of the present disclosure may also include an input system and an output system. Among them, the processor, memory, input system and output system may be connected through a bus or in other ways, which are not specifically limited here.

[0262] The memory is a computer-readable storage medium that can be used to store software programs, computer executable programs, and various modules, such as the program or module corresponding to the application method of the embodiment of the present disclosure. The processor executes various functional applications and data processing of the electronic device by running the software programs or modules stored in the memory.

[0263] The input system can be used to receive input numbers or signals. The signal can be a key signal related to user settings and function control of the device / terminal / server. The output system can include display devices such as display screens.

[0264] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. An alarm device for an electric energy metering experiment scheduling management platform, characterized in that: include: The electric energy metering experiment dispatching management platform is used to read the verification data of each automated verification line in this metering verification. The verification data includes the quality verification results and the corresponding operation and maintenance logs; The quality inspection system is used to read each verification data using a preset quality inspection model, and determine in turn whether the quality inspection result in each verification data is qualified: if qualified, then give up; If it fails, the operation and maintenance log corresponding to the unqualified quality inspection result will be imported into the fault intelligent detection system: The intelligent fault detection system is used to identify and analyze whether the operation and maintenance log has fault log features through a pre-deployed fault AI detection model: if the fault log features appear, the corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log features; if the fault log features do not appear, the operation and maintenance log is sent to the management terminal of the automated verification line corresponding to the unqualified verification data, and the on-site administrator is notified to conduct on-site investigation; An inspection system is used to construct a fault inspection and automatic maintenance plan for the current automated verification pipeline according to the operation fault information, and send the fault inspection plan to the management terminal of the current automated verification pipeline, so that the management terminal automatically executes the fault inspection and automatic maintenance plan; A management terminal, used to receive and display the operation and maintenance log, or receive and execute the fault inspection and automatic maintenance plan; The quality detection system, intelligent fault detection system and patrol inspection system are respectively deployed on the electric energy metering experiment scheduling management platform; the electric energy metering experiment scheduling management platform is communicatively connected with the management terminal.

2. The alarm device of the electric energy metering experiment scheduling management platform according to claim 1 is characterized in that: The electric energy metering experiment scheduling management platform is also provided with: Several high-speed network cables for communication connection with the management terminal of each automated verification line; The electric energy metering experiment scheduling management platform and the management terminal of each automated verification line communicate data via a high-speed network cable.

3. The alarm device of the electric energy metering experiment scheduling management platform according to claim 1 is characterized in that: The electric energy metering experiment scheduling management platform is also provided with: The Oracle database is used to store each verification data of each automated verification line in each metrological verification and save it as the verification big data of each automated verification line.

4. The alarm device of the electric energy metering experiment scheduling management platform according to claim 1 is characterized in that: The electric energy metering experiment dispatching management platform is also provided with: The message middleware is used to realize the message distribution between the electric energy metering experiment scheduling management platform and the management terminal of each automated verification line according to the message queue mechanism.

5. The alarm device of the electric energy metering experiment scheduling management platform according to claim 1 is characterized in that: The method for generating the fault AI detection model comprises: Collecting the verification big data of each of the automated verification lines, including the operation and maintenance log of each automated verification line in each metrological verification; Performing feature engineering on the verification big data, extracting fault log features corresponding to the operation and maintenance log, and forming a fault log feature set; Importing the fault log feature set into a preset RNN model, performing feature learning, and training to generate a corresponding RNN fault recognition model; Randomly collect a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verify the RNN fault identification model; If the verification is qualified, the corresponding fault AI detection model is obtained, and the fault AI detection model is deployed in the fault intelligent detection system.

6. The alarm device of the electric energy metering experiment scheduling management platform according to claim 5 is characterized in that: In the step of collecting the verification big data of each of the automated verification lines, including the operation and maintenance log of each automated verification line in each metrological verification, the following model is adopted: Assume that the data set is D = {d1, d2, ..., d n }, where each d i Indicates the i-th record, containing the following information: Quality inspection result q i (pass / fail) and operation and maintenance logs i ; Where n represents the total number of records; d i represents the i-th record, including the quality inspection result and the corresponding operation and maintenance log; q i Indicates the quality inspection result of the i-th record; l i Represents the operation and maintenance log of the i-th record; In the step of performing feature engineering on the verification big data, extracting fault log features of the corresponding operation and maintenance logs, and forming a fault log feature set, the following model is used: Assume that the feature set is F = {f1,f2,…,f m }, where f j represents the jth feature; Where m represents the total number of features; f j represents the jth feature, which may include time features, frequency features, and abnormal patterns; In the step of importing the fault log feature set into the preset RNN model, performing feature learning, and training to generate the corresponding RNN fault recognition model, the following model is used: Assume that the training set is T = {(f1,y1),(f2,y2),…,(f k ,y k )}, where y i Indicates the corresponding fault label; Where k represents the number of samples; y i represents the fault label of the i-th sample (e.g., normal, fault type A, fault type B); In the step of randomly collecting a number of operation and maintenance logs from a certain automated verification pipeline in real time as a verification set and verifying the RNN fault identification model, the output of the RNN can be expressed as: h t =σ(W h h t-1 +W x h x +b h ); y t =W h h t +b y ; Among them, h t represents the hidden state at time step t; x t Input feature vector; W h , W x , W y represents the weight matrix; b h , b y represents the bias term; σ represents the activation function (such as tanh or ReLU); t represents the time step; h t Represents the current hidden state, capturing sequence information; y t Represents the output of the current time step, indicating the prediction of the fault type.

7. The alarm device of the electric energy metering experiment scheduling management platform according to claim 1 is characterized in that: The management terminal is also used for the on-site administrator to feedback the on-site investigation results to the electric energy metering experiment scheduling management platform after the on-site investigation; or to feedback the execution results to the electric energy metering experiment scheduling management platform after the fault inspection and automatic maintenance plan are executed.

8. An application method of the alarm device of the electric energy metering experiment scheduling management platform according to any one of claims 1 to 7, characterized in that: The steps include: The electric energy metering experiment dispatching management platform reads each verification data of each automated verification line in this metering verification, including quality verification results and corresponding operation and maintenance logs; The quality inspection system reads each piece of inspection data by using a preset quality inspection model, and determines in turn whether the quality inspection result in each piece of inspection data is qualified: If qualified, give up; If it fails, the operation and maintenance log A corresponding to the unqualified quality inspection result is imported into the fault intelligent detection system: The intelligent fault detection system uses the pre-deployed fault AI detection model to identify and analyze whether the operation and maintenance log A has the following fault log characteristics: If a fault log feature appears, corresponding operation fault information is generated in real time, wherein the operation fault information includes the operation and maintenance fault type and location of the fault log feature; If no fault log feature appears, the operation and maintenance log A is sent to the management terminal of the unqualified automated verification line, and the on-site administrator is notified to conduct on-site investigation; after the on-site investigation, the on-site administrator feeds back the on-site investigation results to the electric energy metering experiment scheduling management platform through the management terminal; The inspection system constructs a fault inspection and automatic maintenance plan for the current automated calibration pipeline based on the operating fault information, and sends the fault inspection plan to the management terminal of the current automated calibration pipeline, which is then automatically executed by the management terminal. After the fault inspection and automatic maintenance plan are executed, the management terminal feeds back the execution results to the electric energy metering experiment scheduling management platform.

9. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the application method described in claim 8 when executing the executable instructions.

Images