A power terminal edge control method, system, device and medium

Abstract

The application belongs to the field of smart grid edge control, and discloses a power terminal edge control method, system, device and medium. By deploying a control substation with a built-in edge light intelligent agent, an intelligent intermediary layer between the master station and the power terminal is constructed. The uplink and downlink data and instructions are identified and classified in real time. The effective information is transmitted and counted, and the invalid information is recorded to the local problem library. New data and new instructions are continuously accumulated to the local knowledge base, and the conventional information is intelligently identified according to the record frequency. The local problem library and the knowledge base are uploaded to the cloud to drive incremental learning and generate learning results, and then the edge light intelligent agent is updated through the master station to realize the intelligent calibration of the substation. The substation can also generate a collection rule fine tuning request based on the data statistical result and the preset rule, and the master station issues personalized collection instructions accordingly to realize the customized strategy adaptation of each terminal. The method effectively improves the intelligent processing level of the terminal side and the flexibility of the master station control strategy.

Description

Technical Field

[0001] This invention belongs to the field of smart grid edge control technology, and particularly relates to a power terminal edge control method, system, device and medium. Background Technology

[0002] Smart grid edge control is a core technology in the intelligent operation and maintenance management of power grids, and is widely used in scenarios such as power grid data acquisition, equipment condition monitoring, and operational status management. Currently, traditional smart grid edge control systems generally adopt an operation mode of local processing at the terminal and centralized management by the master station. After completing basic local calculations, the edge terminal directly uploads various power grid information, such as power data and equipment status, to the master station. The master station achieves overall acquisition of power grid operation data and subsequent business processing through unified monitoring and data collection of all edge terminals. This architecture is the mainstream technical solution in the current field of power grid edge management.

[0003] Existing traditional edge control technologies have significant technical limitations in practical applications, making them difficult to adapt to the high-level operation and maintenance requirements of modern smart grids. On the one hand, the built-in computing logic of existing edge terminals is relatively simple, mostly employing fixed threshold judgment mechanisms. They can only complete basic data statistics and simple judgment tasks, lacking the ability to intelligently analyze, logically judge, and make real-time autonomous decisions on complex and massive amounts of grid data. The level of intelligent processing at the terminals is severely insufficient. On the other hand, the edge terminals connected to the grid master station are diverse and numerous, generally adopting a rigid control mode driven by fixed rules. The control strategy has extremely low flexibility and cannot adapt to the dynamically changing operating conditions of the grid and the complex and ever-changing environmental information on site. These dual technical deficiencies result in extremely poor adaptability of traditional edge control solutions in high-precision, high-real-time, and complex and ever-changing grid application scenarios. This directly leads to a series of operation and maintenance problems such as low grid data acquisition efficiency, delayed equipment fault monitoring and identification, and untimely response to abnormal operating conditions, severely restricting the accuracy and efficiency of edge-side management and control of smart grids.

[0004] It is evident that existing smart grid traditional edge control technologies suffer from weak terminal intelligent processing capabilities and insufficient flexibility in master station control strategies, failing to meet the power grid edge management and control requirements under high-precision, high-real-time, and complex dynamic scenarios. Summary of the Invention

[0005] This invention provides a power terminal edge control method, system, device, and medium. This method can solve the problems of weak terminal intelligent processing capability and insufficient flexibility of master station control strategy in existing smart grid traditional edge control technologies, and can meet the power grid edge management and control requirements in high-precision, high-real-time, and complex dynamic scenarios.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: An edge control method for a power terminal is applied to a control substation, which is deployed between a master station and the power terminal, and the control substation has a built-in edge lightweight agent; the control method includes: The system identifies the types of data transmitted from power terminals, uploads the identified good data to the main station and counts it, and records the identified bad data in the local problem database. The system identifies the type of commands issued by the main station, sends valid commands to the power terminals and counts them, and records invalid commands in the local problem database. Among them, the newly received data and instructions are recorded in the local knowledge base and counted. Intelligent analysis is carried out based on the recording frequency of the new data and instructions. When the preset conditions are met, the new data and instructions are judged as regular data and regular instructions. Upload the local question bank and local knowledge bank to the main site so that the cloud can perform incremental learning based on the local question bank and local knowledge bank and output the learning results; The edge lightweight agent receives and loads the updated edge agent through the main station to achieve its own intelligent calibration; the edge lightweight agent updates based on the learning results output from the cloud. Based on the statistical results of the data and instructions, and combined with the preset request rules, the request is judged, a terminal collection rule fine-tuning request is generated and uploaded to the main station; The system receives a data collection instruction from the master station. This instruction indicates that the corresponding power terminal is allowed to adopt the fine-tuned data collection rules, so that each terminal corresponds to a specific strategy. The data collection instruction is generated by edge-side calculation and analysis after the master station receives the fine-tuning request.

[0007] Furthermore, it also includes: Based on the retrieved local problem database and pre-defined fault rules, fault diagnosis is performed, fault alarm information is generated and uploaded to the main station, including: The control substation pre-acquires threshold standard values ​​and historical fault data. During terminal polling, it queries terminal fault data and sequentially determines whether the communication failure rate, data acquisition failure rate, null value rate, and post-cleaning abnormal data rate exceed the corresponding thresholds. If the communication failure rate exceeds the threshold, a terminal fault flag is set; if the data acquisition failure rate exceeds the threshold, a terminal instability flag is set; if the null value rate exceeds the threshold, a terminal unreliable flag is set; if the post-cleaning abnormal data rate exceeds the threshold, a terminal unreliable flag is set; if none of the rates exceed the threshold, a terminal normal flag is set. After merging the terminal working status, a comprehensive judgment is made based on historical and current data. If the judgment result exceeds the preset judgment threshold, a fault alarm message is generated and uploaded to the main station.

[0008] Furthermore, the update process of the edge lightweight agent includes: The cloud prepares updated edge lightweight agent data and sorts it according to the main station address range and control substation address range. The edge lightweight agent data is distributed to the corresponding main station according to the main station address table. The main station saves the edge lightweight agent data and sends back a successful reception message. The system periodically checks the reception status and retransmits the edge lightweight agent data to the main station that has not responded. The master station distributes data to the corresponding control substations according to the control substation address table. The control substations save the data and send back a message indicating successful reception. The master station periodically checks the reception status and retransmits the data to control substations that do not respond until the control substations have completed the loading of the edge lightweight agent.

[0009] Furthermore, the process of identifying the type of instructions issued by the master station, issuing valid instructions to the power terminal and counting them, and recording invalid instructions in the local problem database includes: After receiving an instruction from the master station, the control substation first checks whether the instruction has been acknowledged. If not, it checks whether the instruction is being received for the first time. If it is, the instruction is stored in the local new instruction library and the counter is updated. If it is not, it checks whether the instruction conforms to the preset validity rules. If the instruction conforms to the validity rules, it is determined to be a valid instruction and sent to the power terminal; if the instruction does not conform to the validity rules, or the counter count of the instruction in the new instruction library does not exceed the preset threshold, it is determined to be an invalid instruction and the instruction information is recorded in the local problem library; if the counter count of the instruction in the new instruction library exceeds the preset threshold, it is determined to be a valid instruction and sent to the power terminal.

[0010] Furthermore, the process of identifying the type of data transmitted by the power terminal, uploading the identified good data to the main station and counting it, and recording the identified bad data in the local problem database includes: The control substation pre-acquires bad data characteristics, good data characteristics, and data cleaning rules; After the power terminal uploads data, it compares the similarity of the data with the characteristics of bad data. If the data matches the characteristics of bad data, it is determined to be bad data, stored in the local problem database, and the problem counter is updated. If the data does not match the characteristics of bad data, it compares the similarity of the data with the characteristics of good data. If the data matches the characteristics of good data, it is determined to be good data and uploaded to the main station. If the data does not match the characteristics of good data, it is determined whether the number of data cleaning times exceeds the preset threshold. If the number of cleaning attempts exceeds the threshold, the data is identified as bad data and stored in the local problem database; if the number of cleaning attempts does not exceed the threshold, the data is cleaned according to the preset cleaning rules, the problem counter is updated, and the data type is re-identified.

[0011] Furthermore, the intelligent analysis based on the recording frequency of new data and new instructions includes: The control substation will store the newly received data and instructions into its local knowledge base and count them accordingly. When the frequency of recording new data or new instructions is lower than the preset frequency threshold, only counting is performed, and no data is forwarded or instructions are issued. When the frequency of recording new data or instructions exceeds a preset frequency threshold, the data is forwarded or instructions are issued, and the proportion of bad data subsequently transmitted by power terminals is counted. If the proportion of bad data is higher than a preset bad data threshold, the new data or instructions are identified as non-compliant data or instructions and recorded in the local problem database. If the proportion of bad data is lower than a preset bad data threshold, it is determined whether it conforms to a preset usage paradigm. If it does, it is identified as regular data or instructions and the local knowledge base is updated. If it does not, it is identified as non-compliant data or instructions and recorded in the local problem database.

[0012] Furthermore, the step of determining the request based on the statistical results of data and instructions, combined with preset request rules, generating a terminal acquisition rule fine-tuning request, and uploading it to the main station also includes: The main site's processing steps include: After receiving the fine-tuning request uploaded by the control substation, the main station performs edge-side calculations and analysis to determine whether intervention is needed. If no intervention is needed, the main station directly issues a collection command allowing fine-tuning to the control substation and records the fine-tuning case in the main station's knowledge base. If intervention is needed, the main station generates the revised collection rules and sends them to the control substation, and records the revised rules in the main station's knowledge base. After receiving the acquisition command or the revised acquisition rule, the control substation sends it to the corresponding power terminal so that the power terminal can operate according to the fine-tuned acquisition rule.

[0013] A power terminal edge control system, used in the aforementioned power terminal edge control method, comprising: The first identification module is used to identify the type of data sent by the power terminal, upload the identified good data to the main station and count it, and record the identified bad data to the local problem database. The second identification module is used to identify the type of instructions issued by the main station, send the identified valid instructions to the power terminal and count them, and record the identified invalid instructions to the local problem database. The intelligent analysis module is used to record and count the new data and instructions received in the local knowledge base. It performs intelligent analysis based on the recording frequency of new data and instructions, and determines the new data and instructions as regular data and regular instructions when the preset conditions are met. The data upload module is used to upload the local question bank and local knowledge bank to the main site, so that the cloud can carry out incremental learning based on the local question bank and local knowledge bank and output the learning results. The update module is used to receive and load the updated edge light agent through the main station to achieve its own intelligent calibration; the edge light agent is updated based on the learning results output from the cloud. The request judgment module is used to judge requests based on the statistical results of data and instructions, combined with preset request rules, generate a terminal collection rule fine-tuning request, and upload it to the main station. The instruction receiving module is used to receive the acquisition instructions issued by the master station. The acquisition instructions are used to indicate that the corresponding power terminal is allowed to adopt the fine-tuned acquisition rules so that each terminal corresponds to a countermeasure. The acquisition instructions are generated by the master station after receiving the fine-tuning request and performing edge-side calculations and analysis.

[0014] A power terminal edge control device, comprising: Memory, used to store computer programs; A processor is used to implement the aforementioned power terminal edge control method when executing the computer program.

[0015] A computer-readable storage medium storing a computer program, which, when executed by a processor, is used to implement the aforementioned power terminal edge control method.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a power terminal edge control method. By deploying a control substation with a built-in edge lightweight agent, an intelligent intermediary layer is constructed between the master station and the power terminal. This layer performs real-time identification and classification of uplink and downlink data and commands, transmitting and counting valid information while recording invalid information in a local problem database. Simultaneously, new data and commands are continuously accumulated in a local knowledge base, and routine information is intelligently identified based on recording frequency. The system uploads the local problem database and knowledge base to the cloud to drive incremental learning and generate learning results. The master station then updates the edge lightweight agent to achieve intelligent calibration of the substation. The substation can also generate fine-tuning requests for acquisition rules based on data statistics and preset rules. The master station then issues personalized acquisition commands accordingly, achieving customized strategy adaptation for each terminal. Through layered processing and cloud collaboration, rigid control driven by fixed rules is transformed into flexible intelligent control with dynamic learning and optimization. Real-time identification, accumulation, and feedback at the edge are combined with incremental learning and model iteration in the cloud to form a closed-loop optimization mechanism. This method effectively improves the intelligent processing level of the terminal side and the flexibility of the master station control strategy, enhances the system's adaptability to complex dynamic scenarios of the power grid, and thus significantly improves data acquisition efficiency, fault monitoring timeliness and abnormal operating condition response speed, fully supporting the precise and efficient management and control needs of the smart grid edge side. Attached Figure Description

[0017] Figure 1An implementation framework diagram of a power terminal edge control method provided in an embodiment of the present invention; Figure 2 A flowchart illustrating the process of a power terminal edge control method provided in an embodiment of the present invention; Figure 3 A data flow diagram of a power terminal edge control method provided in an embodiment of the present invention; Figure 4 A flowchart illustrating the data distribution to the controller provided in this embodiment of the invention; Figure 5 A flowchart for instruction identification provided in an embodiment of the present invention; Figure 6 A data processing flowchart provided for embodiments of the present invention; Figure 7 A fault determination flowchart provided for embodiments of the present invention; Figure 8 This is a core flowchart of a power terminal edge control method provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the structure of a power terminal edge control system provided in an embodiment of the present invention. Detailed Implementation

[0018] To further understand the content of this invention, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0019] The technical terms involved in this invention are explained as follows: AI stands for Artificial Intelligence.

[0020] As described in the background section, in the edge control process of traditional smart grids, edge terminals typically perform local calculations and then directly access the master station to upload collected data (electricity data and operating status). The master station then monitors the edge terminals to obtain grid data and performs corresponding processing. However, traditional control methods lack the ability to intelligently analyze complex data and make real-time decisions due to the simple computational logic or fixed thresholds on the terminal side. Furthermore, the master station mainly performs rule-based rigid control due to the large variety of terminals it supports, making it difficult to cope with dynamically changing working conditions or complex and ever-changing environmental information. These limitations make traditional edge control methods inadequate when facing high-precision, high-real-time, and complex and ever-changing application scenarios, resulting in low data acquisition efficiency and delayed fault monitoring response.

[0021] To address the aforementioned issues, this embodiment provides a power terminal edge control method. This method adds a control substation equipped with a lightweight edge intelligence agent between the master station and the terminal, achieving effective isolation and intelligent connection between the master station and the terminal. It transforms the traditional control mode, which relies solely on local computation at the edge terminal, into a control mode where terminal computation combined with substation-assisted decision-making. This effectively avoids the problems of insufficient intelligent analysis and real-time decision-making capabilities on the terminal side. Simultaneously, this method utilizes lightweight model technology, dynamic model deployment and hot-loading technology, incremental learning and knowledge accumulation technology to integrate the traditional substation into an intelligent control node with "thinking" capabilities. This enables timely realization of experience, flexible control rules, and localized control decisions. By enhancing decision-making capabilities and strengthening localized decision-making, overall data acquisition efficiency is improved, and fault response time is reduced.

[0022] For example, such as Figure 8 As shown, this embodiment provides a power terminal edge control method applied to a control substation. An edge lightweight intelligent agent is installed within the control substation. The specific edge control method includes: The system identifies the types of data transmitted from power terminals, uploads the identified good data to the main station and counts it, and records the identified bad data in the local problem database. The system identifies the type of commands issued by the main station, sends valid commands to the power terminals and counts them, and records invalid commands in the local problem database. Among them, the newly received data and instructions are recorded in the local knowledge base and counted. Intelligent analysis is carried out based on the recording frequency of the new data and instructions. When the preset conditions are met, the new data and instructions are judged as regular data and regular instructions. Upload the local question bank and local knowledge bank to the main site so that the cloud can perform incremental learning based on the local question bank and local knowledge bank and output the learning results; The edge lightweight agent receives and loads the updated edge agent through the main station to achieve its own intelligent calibration; the edge lightweight agent updates based on the learning results output from the cloud. Based on the statistical results of the data and instructions, and combined with the preset request rules, the request is judged, a terminal collection rule fine-tuning request is generated and uploaded to the main station; The system receives a data collection instruction from the master station. This instruction indicates that the corresponding power terminal is allowed to adopt the fine-tuned data collection rules, so that each terminal corresponds to a specific strategy. The data collection instruction is generated by edge-side calculation and analysis after the master station receives the fine-tuning request.

[0023] The edge control method provided in this embodiment will be further described below with reference to the accompanying drawings: like Figure 1The diagram shown illustrates the overall framework of edge control. The entire architecture mainly consists of three parts: the main station, the control substation, and the terminal. Figure 2 As shown, this method implements edge control based on artificial intelligence, and the specific steps are as follows: A control substation with built-in AI is added between the traditional data acquisition front-end (main station) and the terminal; Based on the number of terminals and the types of power grid business functions they serve, the control substations are deployed in a distributed manner. Pre-install mature edge lightweight agents in the control substation; After the edge lightweight intelligent agent is iterated and upgraded, it is downloaded from the cloud to the control substation via the main station; When the terminal sends data, the control substation identifies "bad" data, "good" data is intelligently cleaned and then uploaded to the main station and counted, and "bad" data is not uploaded but only recorded in the local problem database; The control substation improves the data quality of the main station by isolating "bad" data, while reducing the unnecessary use of the main station's communication bandwidth. When the master station issues a command, the control substation identifies "invalid" commands, performs intelligent correction on "valid" commands, issues them to the terminal and counts them, and does not issue "invalid" commands but only records them in the local problem database. By isolating "invalid" commands, the control substation improves the timeliness of terminal data collection and reduces the possibility of terminal misoperation. When the control substation receives new data or instructions that are "unrecognizable", it does not forward them, but instead records the data or instructions in the local knowledge base and counts them. The control substation performs intelligent analysis based on the frequency and statistics of new data and instructions. If the conditions are met, the data and instructions are then processed as "regular" data and instructions. The control substation regularly uploads local question and knowledge bases to the cloud through the main station to assist the cloud in incremental learning and achieve "knowledge sharing". The updated model and knowledge base are downloaded from the main station to the control substation via the cloud to assist the control substation in self-updating and achieve "intelligent calibration". The control substation utilizes a local problem database and, with the help of traditional rules and intelligent judgment, can proactively generate fault alarm information and send it to the main station to assist in fault diagnosis. Based on the statistics of data and instructions, the control substation can proactively generate and send the terminal data collection rule fine-tuning request to the main station by combining traditional rules with intelligent judgment. The master station performs edge-side calculations and analyses based on the requests from the control substation, sends instructions to the control substation, and allows individual terminals to use fine-tuned collection rule instructions to achieve "one policy per terminal"; Traditionally, rules are issued uniformly by the main station. Due to the wide impact, people are hesitant to change the rules. This embodiment adopts a "one terminal, one policy" approach, which effectively solves this problem and improves the flexibility of terminal control.

[0024] As can be seen, this method first deploys a control substation for each type of terminal or terminal with similar functions based on the power grid service function type of the corresponding terminal under the main station. When the number of terminals of a certain type exceeds the control limit of N units, another control substation is added, and so on, to achieve distributed deployment of control substations. The control substation isolates the terminal from direct access to the main station. The control substation filters requests from illegitimate terminals through a terminal IP (Internet Protocol) whitelist and security authentication. At the same time, the control substation forwards communication requests bidirectionally after "intelligent thinking," filtering out "blocking" irrelevant access requests, thereby reducing bidirectional interference and invalid bandwidth consumption, and realizing intelligent connection.

[0025] Combination Figures 3 to 7 As shown, the edge control method provided in this embodiment will be further explained: Furthermore, the cloud trains smaller models based on large models for specific application scenarios in the power sector, and adds a knowledge base adapted to the application scenarios to form edge lightweight agents. These edge lightweight agents are pre-installed in the control substations at the factory. The cloud periodically adjusts the smaller models and optimizes the edge lightweight agents based on information feedback (data, problems, logs, etc.) from the master station. The cloud then distributes the iterated edge lightweight agents to the control substations through the master station according to the master station's address table.

[0026] Furthermore, the cloud categorizes problematic data from historical data collected across different application scenarios. Utilizing data feature value approximation analysis combined with manually intervened business data rules, it extracts reference data judgment rules (good data, bad data) and cleaning rules, accumulating a data sample library for specific scenarios. Through model training, these judgment rules and the sample database are integrated into the edge lightweight agent, forming intelligent judgment capabilities. When terminal data is uploaded, the control substation uses its own capabilities to determine the "goodness" of the data. "Good" data that meets the rules is intelligently cleaned before being uploaded to the main station, and logs are recorded simultaneously. "Bad" data that does not meet the rules is discarded directly after recording detailed information (terminal information, time, data) in the local problem database. In extreme cases, if the terminal fails to upload data within the agreed-upon time, the control substation can temporarily send a terminal "wake-up" command and automatically record a problem log.

[0027] Furthermore, the cloud categorizes and organizes instructions from different application scenarios into instruction sets. Simultaneously, it refines the usage conditions and methods of atomic instructions to form instruction paradigms. Through model training, these instruction sets and usage paradigms are integrated into the edge lightweight intelligent agent, forming intelligent judgment capabilities. When the master station issues an instruction, the control substation uses its own capabilities to determine the "validity" of the instruction. "Valid" instructions belonging to the instruction set and meeting the usage paradigm requirements are issued to the terminal and logged simultaneously. "Invalid" instructions that do not meet the usage paradigm requirements are discarded directly after recording detailed information (instruction information, time) in the local problem database. In extreme cases, when short-term, excessively frequent instruction issuance causes communication bandwidth "congestion," the control substation can temporarily close the communication connection to actively "silence" itself and automatically log the problem. In special circumstances, for new instructions or data outside the instruction set, the control substation neither readily accepts nor rashly discards the instructions. It performs a preliminary assessment of the frequency of such instructions or data occurrences. If the frequency is below a threshold, no instruction is issued, and the count is increased. If the frequency is above the threshold, an instruction is issued, and the count is increased. Simultaneously, the proportion of "bad" data subsequently transmitted by the terminal is recorded. If the proportion of "bad" data is above the threshold, the new instruction or data is judged as a "violation" command and recorded in the local problem database. If the proportion of "bad" data is below the threshold, a usage paradigm compliance assessment is performed. If the usage paradigm requirements are met, the new instruction or data is judged as a "normal" command; otherwise, it is judged as a "violation" command and recorded in the local problem database. The relevant process data for this assessment is saved to the local knowledge base.

[0028] Furthermore, when the amount of data in the local question base and knowledge base reaches a set threshold, or when the time interval since the last "feedback" of local knowledge accumulated in the cloud reaches a set threshold, the control substation uploads the local question base and knowledge base to the cloud through the main station. Based on the information "feedback" from numerous AI substations and terminals, the cloud reorganizes the questions or knowledge, classifying them according to device type, problem nature, severity of problem impact, invalid data grouping, and correlation between invalid data and device defects. This process further refines and combines the information, transforming static data into part of a dynamic cognitive framework, thereby providing incremental knowledge to the cloud, gradually enriching and improving the cloud model, and ultimately enhancing the AI's ability to "think" and solve real-world problems.

[0029] Furthermore, during periods of relatively low communication channel downtime, the cloud-based system downloads a lightweight intelligent agent—based on a finely tuned small model and a refined knowledge base—to the control substation via the main station. Utilizing dynamic deployment and hot-loading technologies, the control substation automatically upgrades. After the upgrade, the control substation performs self-debugging by sending test data and commands upwards and downwards. Simultaneously, it optimizes local parameters using its local knowledge base, achieving a "one-site-one-policy" strategy configuration, ultimately completing the substation's self-calibration. Further, the control substation focuses on monitoring "suspected" problem terminals based on its local problem database. By sending targeted diagnostic commands, it obtains feedback information from "suspected" terminals. Based on the accumulated feedback information, it combines traditional rules and AI analysis to perform local assisted diagnosis. If the diagnosis leans towards a false positive, it simply records the result in the local database without blocking data uploads or command issuance. If the diagnosis leans towards a confirmed positive, it records the result in the local database, blocks data uploads and command issuance, and simultaneously uploads the result to the main station, achieving proactive early warning.

[0030] Furthermore, the control substation makes adaptive fine-tuning of the local data collection rules based on the quality of local data collection (validity rules) and the busy / idle status of the local communication channel. On the one hand, it records the fine-tuning information in the local knowledge base (prerequisites and fine-tuning scheme), and on the other hand, it sends the fine-tuning request to the main station for record-keeping to avoid misjudgment by the main station. The fine-tuning scheme can be selected from the preset schemes, or it can be added or subtracted from the preset schemes within a threshold range. Through fine-tuning, the control substation achieves the intelligent control purpose of "thinking" for itself.

[0031] Furthermore, after receiving the fine-tuning request from the control substation, the master station performs calculations and analysis to determine the necessity of intervention. If no intervention is required, it directly sends a permission instruction to the control substation and records the fine-tuning case in the knowledge base. If intervention is required, it sends the revised collection rules to the control substation, records the revised case in the knowledge base, and then the control substation issues the collection rules to the relevant terminals, achieving "one policy per terminal" and realizing the goal of flexible control.

[0032] like Figure 9 As shown, this embodiment also provides a power terminal edge control system, including: a first identification module, used to identify the type of data sent by the power terminal, upload the identified good data to the main station and count it, and record the identified bad data to the local problem database; The second identification module is used to identify the type of instructions issued by the main station, send the identified valid instructions to the power terminal and count them, and record the identified invalid instructions to the local problem database. The intelligent analysis module is used to record and count the new data and instructions received in the local knowledge base. It performs intelligent analysis based on the recording frequency of new data and instructions, and determines the new data and instructions as regular data and regular instructions when the preset conditions are met. The data upload module is used to upload the local question bank and local knowledge bank to the main site, so that the cloud can carry out incremental learning based on the local question bank and local knowledge bank and output the learning results. The update module is used to receive and load the updated edge light agent through the main station to achieve its own intelligent calibration; the edge light agent is updated based on the learning results output from the cloud. The request judgment module is used to judge requests based on the statistical results of data and instructions, combined with preset request rules, generate a terminal collection rule fine-tuning request, and upload it to the main station. The instruction receiving module is used to receive the acquisition instructions issued by the master station. The acquisition instructions are used to indicate that the corresponding power terminal is allowed to adopt the fine-tuned acquisition rules so that each terminal corresponds to a countermeasure. The acquisition instructions are generated by the master station after receiving the fine-tuning request and performing edge-side calculations and analysis.

[0033] The present invention also provides a power terminal edge control device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the power terminal edge control method.

[0034] The present invention also provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the power terminal edge control method.

[0035] When the processor executes the computer program, it implements the above-mentioned power terminal edge control steps, such as: identifying the type of data sent by the power terminal, uploading the identified good data to the main station and counting it, and recording the identified bad data to the local problem database. The system identifies the type of commands issued by the main station, sends valid commands to the power terminals and counts them, and records invalid commands in the local problem database. Among them, the newly received data and instructions are recorded in the local knowledge base and counted. Intelligent analysis is carried out based on the recording frequency of the new data and instructions. When the preset conditions are met, the new data and instructions are judged as regular data and regular instructions. Upload the local question bank and local knowledge bank to the main site so that the cloud can perform incremental learning based on the local question bank and local knowledge bank and output the learning results; The edge lightweight agent receives and loads the updated edge agent through the main station to achieve its own intelligent calibration; the edge lightweight agent updates based on the learning results output from the cloud. Based on the statistical results of the data and instructions, and combined with the preset request rules, the request is judged, a terminal collection rule fine-tuning request is generated and uploaded to the main station; The system receives a data collection instruction from the master station. This instruction indicates that the corresponding power terminal is allowed to adopt the fine-tuned data collection rules, so that each terminal corresponds to a specific strategy. The data collection instruction is generated by edge-side calculation and analysis after the master station receives the fine-tuning request.

[0036] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing preset functions, the instruction segments describing the execution process of the computer program in the power terminal edge control device. For example, the computer program can be divided into a first identification module, a second identification module, an intelligent analysis module, a data upload module, an update module, a request judgment module, and an instruction receiving module; the specific functions are as follows: The first identification module is used to identify the type of data sent by the power terminal, upload the identified good data to the main station and count it, and record the identified bad data in the local problem database; the second identification module is used to identify the type of instructions issued by the main station, send the identified valid instructions to the power terminal and count them, and record the identified invalid instructions in the local problem database; the intelligent analysis module is used to record the received new data and new instructions in the local knowledge base and count them, perform intelligent analysis based on the recording frequency of new data and new instructions, and determine the new data and new instructions as regular data or regular... The system comprises the following modules: a data upload module for uploading local problem and knowledge bases to the main station for incremental learning and output of learning results by the cloud; an update module for receiving and loading updated edge agents through the main station for intelligent calibration; the edge agents are updated based on the learning results output by the cloud; a request judgment module for judging requests based on statistical results of data and instructions, combined with preset request rules, generating terminal acquisition rule fine-tuning requests and uploading them to the main station; and an instruction receiving module for receiving acquisition instructions from the main station, which indicate that the corresponding power terminal is allowed to adopt the fine-tuned acquisition rules, so that each terminal corresponds to a specific strategy; the acquisition instructions are generated by edge-side calculation and analysis after the main station receives the fine-tuning request.

[0037] The power terminal edge control device can be a computing device such as a desktop computer, laptop, handheld computer, or cloud server. The power terminal edge control device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above are examples of power terminal edge control devices and do not constitute a limitation on them. They may include more components than described above, or combine certain components, or different components. For example, the power terminal edge control device may also include input / output devices, network access devices, buses, etc.

[0038] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor, or any conventional processor. This processor is the control center of the power terminal edge control system, connecting various parts of the power terminal edge control device via various interfaces and lines.

[0039] The memory can be used to store the computer program and / or module. The processor implements various functions of the power terminal edge control device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0040] The memory may primarily include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function (such as sound playback, image playback, etc.). The data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD cards), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0041] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the power terminal edge control method described above.

[0042] If the modules / units integrated in the power terminal edge control system are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium.

[0043] Based on this understanding, the present invention can implement all or part of the processes in the above-described power terminal edge control method, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-described power terminal edge control method. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or a preset intermediate form, etc.

[0044] The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0045] It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0046] Compared with existing control measures, this invention provides a power terminal edge control method, system, device, and medium. This method can effectively solve the problems of insufficient local analysis and decision-making capabilities of current power acquisition systems, inability to rationally utilize historical problems and lessons learned, rigid rules being difficult to adapt to changes in scenarios, and untimely fault monitoring response. It has the following advantages: First, it enabled the deployment of AI capabilities closer to the front line, improving local analysis and decision-making capabilities; Secondly, it has achieved intelligent cleaning of local data, which has not only enriched the knowledge base of problem data, but also reduced the transmission of invalid data and purified the communication channel. Third, it achieves intelligent isolation of invalid commands, reduces the possibility of misoperation, improves the accuracy of data collection, and further purifies the communication channel; Fourth, it enables intelligent recognition of new instructions, improving the system's flexible control capability to adapt to changes in scenarios; Fifth, it enables the uploading of local knowledge to the cloud knowledge base, achieving relatively wide-area knowledge sharing and further improving the system's potential for handling accidents; Sixth, it enables the download of cloud-based knowledge to subsites, allowing for timely local updates, optimizations, and adjustments, thus preventing knowledge obsolescence and further enhancing the system's local processing capabilities; Seventh, relying on the local knowledge base and using traditional rules plus intelligent analysis and judgment, early fault warnings were achieved, improving the timeliness of fault response; Eighth, leveraging local data advantages and combining traditional rules with intelligent analysis and judgment, it has achieved fine-tuning of local data collection rules, providing a brand-new management model of "one policy per terminal"; Ninth, by adopting the "one terminal, one solution" model, the dilemma of "not daring to make adjustments due to the wide range of implications" has been resolved, effectively improving the flexibility of system control.

[0047] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the present invention.

Claims

1. A power terminal edge control method, characterized in that, The control method is applied to a control substation deployed between the main station and the power terminal, and the control substation has a built-in edge lightweight agent; the control method includes: The system identifies the types of data transmitted from power terminals, uploads the identified good data to the main station and counts it, and records the identified bad data in the local problem database. The system identifies the type of commands issued by the main station, sends valid commands to the power terminals and counts them, and records invalid commands in the local problem database. Among them, the newly received data and instructions are recorded in the local knowledge base and counted. Intelligent analysis is carried out based on the recording frequency of the new data and instructions. When the preset conditions are met, the new data and instructions are judged as regular data and regular instructions. Upload the local question bank and local knowledge bank to the main site so that the cloud can perform incremental learning based on the local question bank and local knowledge bank and output the learning results; The edge light agent receives and loads the updated edge agent through the main station to achieve its own intelligent calibration; the edge light agent updates based on the learning results output from the cloud. Based on the statistical results of the data and instructions, and combined with the preset request rules, the request is judged, a terminal collection rule fine-tuning request is generated and uploaded to the main station; The system receives collection instructions from the master station. These instructions represent the rules that are allowed for the corresponding power terminals to be fine-tuned, so that each terminal corresponds to a specific strategy. The collection instructions are generated by edge-side calculations and analysis after the master station receives the fine-tuning request.

2. The power terminal edge control method according to claim 1, characterized in that, Also includes: Based on the retrieved local problem database and pre-defined fault rules, fault diagnosis is performed, fault alarm information is generated and uploaded to the main station, including: The control substation pre-acquires threshold standard values ​​and historical fault data. During terminal polling, it queries terminal fault data and sequentially determines whether the communication failure rate, data acquisition failure rate, null value rate, and post-cleaning abnormal data rate exceed the corresponding thresholds. If the communication failure rate exceeds the threshold, a terminal fault flag is set; if the data acquisition failure rate exceeds the threshold, a terminal instability flag is set; if the null value rate exceeds the threshold, a terminal unreliable flag is set; if the post-cleaning abnormal data rate exceeds the threshold, a terminal unreliable flag is set; if none of the rates exceed the threshold, a terminal normal flag is set. After merging the terminal working status, a comprehensive judgment is made based on historical and current data. If the judgment result exceeds the preset judgment threshold, a fault alarm message is generated and uploaded to the main station.

3. The power terminal edge control method according to claim 1, characterized in that, The update process of the edge lightweight agent includes: The cloud prepares updated edge lightweight agent data and sorts it according to the main station address range and control substation address range. The edge lightweight agent data is distributed to the corresponding main station according to the main station address table. The main station saves the edge lightweight agent data and sends back a successful reception message. The system periodically checks the reception status and retransmits the edge lightweight agent data to the main station that has not responded. The master station distributes data to the corresponding control substations according to the control substation address table. The control substations save the data and send back a message indicating successful reception. The master station periodically checks the reception status and retransmits the data to control substations that do not respond until the control substations have completed the loading of the edge lightweight agent.

4. The power terminal edge control method according to claim 1, characterized in that, The process of identifying the type of instructions issued by the master station, issuing valid instructions to the power terminal and counting them, and recording invalid instructions in the local problem database includes: After receiving an instruction from the master station, the control substation first checks whether the instruction has been acknowledged. If not, it checks whether the instruction is being received for the first time. If it is, the instruction is stored in the local new instruction library and the counter is updated. If it is not, it checks whether the instruction conforms to the preset validity rules. If the instruction conforms to the validity rules, it is determined to be a valid instruction and sent to the power terminal; if the instruction does not conform to the validity rules, or the counter count of the instruction in the new instruction library does not exceed the preset threshold, it is determined to be an invalid instruction and the instruction information is recorded in the local problem library; if the counter count of the instruction in the new instruction library exceeds the preset threshold, it is determined to be a valid instruction and sent to the power terminal.

5. The power terminal edge control method according to claim 1, characterized in that, The process of identifying the data type transmitted by the power terminal, uploading the identified good data to the main station and counting it, and recording the identified bad data in the local problem database includes: The control substation pre-acquires bad data characteristics, good data characteristics, and data cleaning rules; After the power terminal uploads data, it compares the similarity of the data with the characteristics of bad data. If the data matches the characteristics of bad data, it is determined to be bad data, stored in the local problem database, and the problem counter is updated. If the data does not match the characteristics of bad data, it compares the similarity of the data with the characteristics of good data. If the data matches the characteristics of good data, it is determined to be good data and uploaded to the main station. If the data does not match the characteristics of good data, it is determined whether the number of data cleaning times exceeds the preset threshold. If the number of cleaning attempts exceeds the threshold, the data is identified as bad data and stored in the local problem database; if the number of cleaning attempts does not exceed the threshold, the data is cleaned according to the preset cleaning rules, the problem counter is updated, and the data type is re-identified.

6. The power terminal edge control method according to claim 1, characterized in that, The intelligent analysis based on the recording frequency of new data and new instructions includes: The control substation will store the newly received data and instructions into its local knowledge base and count them accordingly. When the frequency of recording new data or new instructions is lower than the preset frequency threshold, only counting is performed, and no data is forwarded or instructions are issued. When the frequency of recording new data or instructions exceeds a preset frequency threshold, the data is forwarded or instructions are issued, and the proportion of bad data subsequently transmitted by power terminals is counted. If the proportion of bad data is higher than a preset bad data threshold, the new data or instructions are identified as non-compliant data or instructions and recorded in the local problem database. If the proportion of bad data is lower than a preset bad data threshold, it is determined whether it conforms to a preset usage paradigm. If it does, it is identified as regular data or instructions and the local knowledge base is updated. If it does not, it is identified as non-compliant data or instructions and recorded in the local problem database.

7. The power terminal edge control method according to claim 1, characterized in that, The step of judging requests based on statistical results of data and instructions, combined with preset request rules, generating a terminal data collection rule fine-tuning request, and uploading it to the main station also includes: The main site's processing steps include: After receiving the fine-tuning request uploaded by the control substation, the main station performs edge-side calculations and analysis to determine whether intervention is needed. If no intervention is needed, the main station directly issues a collection command allowing fine-tuning to the control substation and records the fine-tuning case in the main station's knowledge base. If intervention is needed, the main station generates the revised collection rules and sends them to the control substation, and records the revised rules in the main station's knowledge base. After receiving the acquisition command or the revised acquisition rule, the control substation sends it to the corresponding power terminal so that the power terminal can operate according to the fine-tuned acquisition rule.

8. A power terminal edge control system for implementing the power terminal edge control method of claim 1, characterized in that, include: The first identification module is used to identify the type of data sent by the power terminal, upload the identified good data to the main station and count it, and record the identified bad data to the local problem database. The second identification module is used to identify the type of instructions issued by the main station, send the identified valid instructions to the power terminal and count them, and record the identified invalid instructions to the local problem database. The intelligent analysis module is used to record and count the new data and instructions received in the local knowledge base. It performs intelligent analysis based on the recording frequency of new data and instructions, and determines the new data and instructions as regular data and regular instructions when the preset conditions are met. The data upload module is used to upload the local question bank and local knowledge bank to the main site, so that the cloud can carry out incremental learning based on the local question bank and local knowledge bank and output the learning results. The update module is used to receive and load the updated edge lightweight agent through the main station to achieve its own intelligent calibration; The edge lightweight agent updates based on the learning results output from the cloud; The request judgment module is used to judge requests based on the statistical results of data and instructions, combined with preset request rules, generate a terminal collection rule fine-tuning request, and upload it to the main station. The instruction receiving module is used to receive the collection instructions issued by the master station. The collection instructions are used to indicate that the corresponding power terminal is allowed to adopt the fine-tuned collection rules, so that each terminal corresponds to a countermeasure. The acquisition command is generated by edge-side calculation and analysis after the main station receives the fine-tuning request.

9. A power terminal edge control device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the power terminal edge control method according to any one of claims 1-7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it is used to implement the power terminal edge control method according to any one of claims 1-7.

Images