A distributed meter identification method and system

Abstract

The present application relates to a kind of distributed meter identification method and system, belong to substation automation monitoring field, wherein method includes: using multiple intelligent camera nodes to collect meter image, using the lightweight local model carried out local reading identification to meter image, and train local model;Each intelligent camera node networking forms hierarchical aggregation network, hierarchical aggregation network according to the residual capacity and link signal strength of network in intelligent camera node Dynamic planning multi-hop routing, dynamically determine relay cluster head node, and through relay cluster head node, the hierarchical aggregation of local model uploaded by each intelligent camera node is carried out, and the forwarding of identification result, after hierarchical aggregation, cluster model and the identification result forwarded are uploaded to central server;Central server carries out aggregation to cluster model, updates global model, issues to each intelligent camera node, and summarizes meter reading identification result.Compared with prior art, the present application can improve the working efficiency and stability of system in complex environment.

Description

Technical Field

[0001] This invention relates to the field of substation automation monitoring, and in particular to a distributed meter identification method and system. Background Technology

[0002] With the continuous advancement of smart grid construction, substations, as a crucial component of the power system, are developing towards digitalization and intelligence. Traditional manual meter reading methods suffer from low efficiency, error-proneness, and high costs, and can no longer meet the needs of modern substations. Furthermore, as substations expand in scale, the workload of manual inspections increases significantly, leading to a rise in labor costs.

[0003] Currently, substation meter monitoring mainly relies on manual inspections, which suffers from low efficiency and high safety risks. Maintenance personnel need to periodically read meters on-site, resulting in time-consuming single-station inspections and risks associated with high-risk working environments and human error recording. Existing technologies have proposed methods to automate meter identification by combining artificial intelligence. For example, CN115171091A discloses a meter identification method for substation inspections, which includes the following steps: S1, acquiring meter images; S2, detecting feature points in the meter images and matching them with a meter template image; S3, calculating the meter pointer reading; S4, analyzing the pointer reading and outputting a structured output. This method employs an intelligent meter identification approach, enabling automatic and accurate meter reading identification during routine power grid inspections. It facilitates simple and convenient recording, statistics, and data analysis, reducing the manpower costs of manual meter reading and effectively improving efficiency and reliability. Furthermore, it utilizes deep learning to construct multilayer perceptron and graph neural network models, extracting high-dimensional vector representations of feature points. By introducing self-attention and cross-attention mechanisms, it better learns the internal and external connections and correspondences of feature points, enhancing their deeper information representation capabilities. While this method achieves high-precision automatic meter identification, it relies on centralized image processing, requiring the transmission of numerous images to a central server for processing. This results in high communication bandwidth pressure, poor real-time performance, and the single image recognition model struggles to adapt to complex and diverse environmental conditions (such as electromagnetic interference and lighting changes). Traditional automation solutions, on the other hand, use 4G / Wi-Fi to transmit high-definition image data, leading to high single-node power consumption requiring frequent battery replacements and poor communication stability in the complex electromagnetic environment of substations. Although low-power wide-area network technologies (such as LoRa) can support device networking, their one-way communication characteristics cannot meet the needs of distributed model collaboration.

[0004] In practical applications, remote monitoring can be achieved by deploying edge devices to improve recognition efficiency. However, centralized deep learning models deployed on edge devices suffer from bottlenecks such as insufficient computing resources and excessive power consumption. To address this, existing technologies have proposed lightweight models. For example, CN119027866A discloses a pointer-type meter recognition method and system for edge computing devices, including acquiring and preprocessing images of transmission lines; inputting the image data into a trained pointer-type meter recognition model to output pointer recognition results; and sending the recognized pointer results to a remote monitoring center after verifying a repeat verification command. This method uses the pointer-type meter recognition model for pointer recognition. It utilizes a target collaborative pruning strategy to maximally remove redundant parameters from the target detection model and then fine-tunes the model parameters to train a recognition model suitable for edge computing, thereby reducing the resource overhead of the automatic pointer-type recognition model while maintaining model accuracy. However, existing technologies primarily improve the recognition model from a lightweight perspective and do not consider optimizing the communication resources between the edge device and the central server. Federated learning has been proposed in some existing technologies to address this problem, but it faces the challenge of unstable network topology in substation scenarios.

[0005] Therefore, there is an urgent need for a low-power, stable, and scalable solution to automate the identification and reporting of readings from traditional meters in substations, providing strong technical support for the intelligent management of substations. Summary of the Invention

[0006] The purpose of this invention is to provide a distributed meter identification method and system. This method deploys intelligent camera nodes equipped with lightweight local models to perform meter detection and reading analysis at the edge. A dynamic clustered network topology is constructed using a low-power Bluetooth network, and a hierarchical federated learning mechanism is used to aggregate local model parameters through relay cluster head nodes, reducing network load. A central server is responsible for global model aggregation updates and identification result management. By employing a joint optimization strategy of communication and computation, low-power automated identification and analysis of traditional meter readings in substations is achieved, reducing the manual maintenance costs of substations.

[0007] The objective of this invention can be achieved through the following technical solutions:

[0008] According to a first aspect of the present invention, a distributed meter identification method is provided, the method comprising the following steps:

[0009] Multiple smart camera nodes are used to collect meter images. A lightweight local model mounted on the smart camera nodes is used to perform local reading recognition on the meter images. A local dataset is constructed and the local model is trained.

[0010] Each smart camera node is networked to form a hierarchical aggregation network. The hierarchical aggregation network dynamically plans multi-hop routes and dynamically determines relay cluster head nodes based on the remaining power and link signal strength of the smart camera nodes in the network.

[0011] The hierarchical aggregation network uses relay cluster head nodes to perform hierarchical aggregation of local models uploaded by each smart camera node and forward the recognition results, and then uploads the hierarchically aggregated cluster model and the forwarded recognition results to the central server.

[0012] The central server aggregates and updates the global model of the cluster model, distributes the updates to each smart camera node to update the local model, and summarizes the meter reading recognition results.

[0013] As a preferred technical solution, in the hierarchical aggregation network, the state S of node i at time t is... i The definition is as follows:

[0014]

[0015] Among them, RSSI i and RSSI avg E represents the signal strength at node i and the global average signal strength, respectively. i and E avg Let represent the remaining power of node i and the global average remaining power, respectively, and α be the scaling factor.

[0016] As a preferred technical solution, the hierarchical aggregation network prioritizes nodes with good status as relay cluster head nodes based on a dynamic routing mechanism, forwarding data from other weaker adjacent nodes to form a multi-cluster topology, resulting in multiple relay cluster head nodes and their corresponding clusters. Each relay cluster head node performs weighted aggregation of local models uploaded by multiple smart camera nodes within a cluster based on the amount of data held locally by each node, obtaining a cluster model. For cluster c, its cluster model w... c The aggregation method is as follows:

[0017]

[0018] Among them, |D i | represents the data size of the local dataset corresponding to the i-th local model in cluster c, M is the number of local models in cluster c, and w i Let i be the i-th local model.

[0019] As a preferred technical solution, the central server receives cluster models uploaded from all hierarchical aggregation networks and updates the global model. Specifically, for time t+1, the global model is updated as follows:

[0020]

[0021] Among them, w g Let represent the global model at time t, η be the learning rate, and K be the number of relay cluster head nodes. w represents the total amount of data within cluster c. c It is a cluster model.

[0022] As a preferred technical solution, the lightweight local model mounted in the smart camera node employs different recognition algorithms depending on the type of meter.

[0023] For pointer-type meters, the local model performs the following steps: target detection is performed on the meter image to obtain the dial area, image segmentation is performed on the dial area to extract the pointer area, and the angular position of the pointer is calculated by combining geometric analysis methods. An angle-reading mapping relationship is established based on the range and scale distribution of the dial, and the meter reading is determined based on the calculated angle and angle-reading mapping relationship.

[0024] For digital meters, the local model performs the following steps: locating the digital region of the meter image, using a character recognition model to extract the digital information on the meter, and obtaining the meter reading.

[0025] As a preferred technical solution, the central server dynamically adjusts the working parameters of the smart camera nodes based on monitoring data to optimize system performance and energy consumption. The monitoring data includes the network load and ambient lighting conditions of the smart camera nodes. The working parameters include shooting frequency, exposure time, and data transmission interval. When the ambient lighting conditions are weaker than preset conditions, the shooting frequency of the smart camera nodes is reduced or the exposure time is increased. When the network load is greater than a preset threshold, the data transmission interval of the smart camera nodes is extended.

[0026] As a preferred technical solution, the central server performs version management on the global model, including a rollback mechanism and a canary release strategy. Specifically, the rollback mechanism is as follows: when the performance of the central server's validation set drops beyond a preset threshold for a consecutive preset number of rounds, it automatically rolls back to the historically optimal global model. The canary release strategy is as follows: the updated global model is verified through canary release and first distributed to a preset proportion of smart camera nodes for trial operation. If there are no abnormal confidence alarms within a preset time interval, it is then fully pushed to all smart camera nodes.

[0027] As a preferred technical solution, the method further includes:

[0028] While using a local model to identify meter readings, the confidence level of the identification result is determined. If the confidence level is greater than a preset threshold, the identification result is forwarded to the central server through a hierarchical aggregation network. Otherwise, the meter image is compressed and forwarded to the central server through the hierarchical aggregation network. The central server calls a high-precision identification model for identification and submits it for manual review.

[0029] As a preferred technical solution, the method further includes:

[0030] When a new smart camera node is added, the hierarchical aggregation network automatically detects it and assigns it a unique network address.

[0031] The new smart camera node establishes a connection with existing nodes by broadcasting messages and downloads the necessary configuration files and initial models from the central server as the initial local model.

[0032] According to a second aspect of the present invention, a distributed meter identification system is provided for implementing the above method, the system comprising:

[0033] Intelligent camera node: As a front-end sensing and edge computing unit, it is deployed in the meter monitoring area of ​​the substation. It supports wireless communication and is powered by a battery. It is equipped with a high-resolution camera and an edge computing module to collect meter images and perform local reading recognition and federated learning local model uploading. It also has local storage function to cache data that was not successfully transmitted.

[0034] The hierarchical aggregation network is formed by multiple smart camera nodes. The nodes communicate with each other through the Bluetooth Low Energy Mesh network protocol. The hierarchical aggregation network dynamically plans multi-hop routes based on the remaining power and link signal strength of the smart camera nodes in the network, dynamically determines the relay cluster head node, and performs hierarchical aggregation of the local models uploaded by each smart camera node and forwards the recognition results through the relay cluster head node. The hierarchically aggregated cluster model and the forwarded recognition results are then uploaded to the central server.

[0035] Central server: Used to summarize the reading recognition results forwarded by the hierarchical aggregation network, and execute the top-level aggregation algorithm of federated learning to aggregate the cluster model to update the global model, and distribute the updated global model to each smart camera node to update the local model.

[0036] According to a third aspect of the present invention, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the program to implement the method described thereon.

[0037] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method described thereon.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) This invention improves the efficiency and accuracy of large-scale meter identification deployment and working scope by automating meter identification and monitoring, reducing the need for manual operation and lowering maintenance costs.

[0040] (2) This invention uses federated learning and lightweight edge algorithm model deployment, which eliminates the need to transmit a large amount of image resources to the central server for processing. Most of the reading recognition can be completed locally. Only when the confidence of the local recognition result does not meet the requirements does the image need to be uploaded to the central server, which greatly reduces the communication bandwidth pressure, has good real-time performance, and meets the requirements of distributed model collaboration.

[0041] (3) This invention combines the low power consumption characteristics of Bluetooth Mesh network with a hierarchical aggregation network with dynamic routing mechanism. It dynamically plans multi-hop routes based on the remaining power of smart camera nodes in the network and the link signal strength, reduces the network link load caused by model uploading, and continuously optimizes the algorithm accuracy to overcome recognition errors in complex environments and ensures that the system can operate reliably in resource-constrained complex environments.

[0042] (4) This invention effectively overcomes the problem of unstable network topology in existing federated learning technologies. The unstable network topology in existing technologies is mainly manifested in communication interruptions between nodes, high latency, frequent disconnections, and unexpected mounting of old nodes. This invention, however, introduces a node state index S... i Taking into account both the signal strength (RSSI) and remaining power (E) of the nodes, the stability and availability of the nodes are dynamically evaluated through a weighted function. This ensures that the selected relay cluster head nodes in the network have strong communication quality and endurance, thereby reducing problems such as communication interruptions, high latency, frequent disconnections, and unexpected attachment of old nodes caused by network topology instability.

[0043] (5) This invention supports the dynamic addition and configuration of new nodes, which can improve the scalability of the system.

[0044] (6) The central server of the present invention dynamically adjusts the working parameters of the camera nodes by monitoring data to cope with the complex working conditions of the substation, thereby enhancing the environmental adaptability of the system. Attached Figure Description

[0045] Figure 1 This is a schematic diagram of the system structure of the present invention;

[0046] Figure 2 This is a flowchart of the method of the present invention;

[0047] Figure 3This is a schematic diagram of the layered polymerization process of the present invention;

[0048] Figure 4 This is a schematic diagram of the model update process of the present invention. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0050] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0051] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0052] Example 1

[0053] This embodiment relates to the interdisciplinary field of substation automation monitoring and edge artificial intelligence, specifically to a distributed meter identification system based on edge computing and federated learning. It combines target detection algorithms, Bluetooth Low Energy communication networks, and a federated learning hierarchical aggregation method to achieve automated reading identification and continuous learning of large-scale meters in substations under complex environments. Figure 1 As shown, the distributed meter identification system includes:

[0054] (1) Smart Camera Node: As a front-end sensing and edge computing unit, it is deployed in the meter monitoring area of ​​the substation, supports wireless communication, and is powered by a battery. The smart camera node is equipped with a high-resolution camera and an edge computing module (i.e., a local model equipped with a lightweight YOLO image recognition algorithm) to acquire meter images at the edge and perform local reading recognition and federated learning of the local model upload. Each smart camera node has local storage function to cache data that has not been successfully transmitted, ensuring that data is not lost when the network is unstable or interrupted. In a preferred embodiment, each smart camera node first uses the locally mounted lightweight detection algorithm to identify the meter reading and gives a confidence level. If the confidence level is lower than the threshold, the image is compressed and sent to the central server, which uses a more accurate algorithm model to identify the reading and submits it for manual review.

[0055] (2) Hierarchical Aggregation Network: This network consists of multiple smart camera nodes that communicate with each other via the Bluetooth Low Energy Mesh network protocol. It supports large-scale multi-hop communication and dynamic routing, making it a low-power, highly scalable wireless network architecture. Each node in the hierarchical aggregation network can act as both a data sender and a relay node, forming a multi-hop self-organizing network structure. The hierarchical aggregation network dynamically plans multi-hop routes based on the remaining battery power and link signal strength of the smart camera nodes within the network, dynamically determining the relay cluster head node to avoid single-point overload problems caused by fixed routes. The relay cluster head node performs hierarchical aggregation of the local models uploaded by each smart camera node and forwards the recognition results. The hierarchically aggregated cluster model and the forwarded recognition results are then uploaded to the central server, further reducing network load and achieving low-power, scalable data transmission. In other words, to reduce the communication load in the network, the local models uploaded by the smart camera nodes will first undergo hierarchical aggregation at the relay cluster head according to the self-organizing network structure, and then the aggregated cluster model will be sent to the central server. In addition, when a node's signal strength is weak or the network is congested, the dynamic routing mechanism will actively rebuild the link and select a node with a stronger signal and lower load for data transmission.

[0056] (3) Central Server: The core of the entire system's control and management, used to aggregate the reading recognition results forwarded by the hierarchical aggregation network, execute the federated learning top-level aggregation algorithm to aggregate the cluster model to update the global model, distribute the updated global model to each smart camera node to update its local model, and store and analyze the data. In addition, the central server is also responsible for managing the firmware upgrades of the smart camera nodes. The central server also has a built-in high-performance computing module capable of processing image data with low confidence and further optimizing the recognition results through deep learning models. In a preferred embodiment, the central server can dynamically adjust the operating parameters of the smart camera nodes based on monitoring data to optimize system power consumption performance.

[0057] Compared with existing technologies, this system achieves localized data processing through edge computing, continuously optimizes model performance through hierarchical federated learning, and ensures low power consumption and high reliability by combining Bluetooth Mesh networks, thereby improving the system's working efficiency and stability in complex environments, while significantly reducing the need for manual maintenance.

[0058] Example 2

[0059] This embodiment, based on Embodiment 1, provides a distributed meter identification method, such as... Figure 2 As shown, the method includes the following steps:

[0060] S1 uses multiple smart camera nodes to collect meter images, uses a lightweight local model mounted on the smart camera nodes to perform local reading recognition on the meter images, builds a local dataset, and trains the local model.

[0061] In this embodiment, the lightweight local model mounted in the smart camera node has been specially optimized to achieve efficient meter reading recognition with limited computing resources. The algorithm employs specialized feature extraction models designed for both analog and digital meters, utilizing techniques such as image segmentation, edge detection, object detection, key point detection, and character recognition to achieve rapid meter reading recognition.

[0062] In one preferred embodiment, different identification algorithms are used depending on the type of meter, wherein,

[0063] For pointer-type meters, the local model performs the following steps: target detection is performed on the meter image to obtain the dial area; image segmentation is performed on the dial area to extract the pointer area; the angular position of the pointer is calculated using geometric analysis methods; an angle-reading mapping relationship is established based on the dial's range and scale distribution; and the meter reading is determined based on the calculated angle and angle-reading mapping relationship.

[0064] For digital meters, the local model performs the following steps: locate the digital area of ​​the meter image, use a character recognition model to extract the digital information on the meter, and obtain the meter reading.

[0065] Each smart camera node i in the local dataset D i Train the local model w i Then its objective function F is:

[0066]

[0067] Where l(·) is the loss function, and x and y are the local dataset D, respectively. i Data and labels in the text.

[0068] While using a local model to identify meter readings, the confidence level of the identification results is determined. If the confidence level is greater than a preset threshold, the identification results are forwarded to the central server through a hierarchical aggregation network. Otherwise, the meter images are compressed and forwarded to the central server through the hierarchical aggregation network. The central server calls a high-precision identification model for identification and submits the data for manual review to ensure the accuracy of the final data.

[0069] S2, each smart camera node is networked to form a hierarchical aggregation network. The hierarchical aggregation network dynamically plans multi-hop routes based on the remaining power of the smart camera nodes and the link signal strength within the network, and dynamically determines the relay cluster head node.

[0070] like Figure 3 As shown, this embodiment introduces federated learning to transfer model parameters between the edge devices and the central server. After the smart camera nodes complete local model training, they upload the models to the hierarchical aggregation network for cluster model aggregation. Then, the hierarchical aggregation network uploads the cluster models to the central server for global model aggregation. Finally, the central server redistributes the updated global model to the smart camera nodes to update their local models.

[0071] The hierarchical aggregation network supports a dynamic routing mechanism that dynamically establishes the network topology based on the load of smart camera nodes. It then performs hierarchical aggregation of local models uploaded by smart camera nodes at cluster relay nodes, thereby avoiding single-point overload and network congestion, optimizing network resource utilization efficiency, and preventing excessive load on certain links from impacting overall performance. This mechanism dynamically adjusts routing and network topology strategies by monitoring the resource utilization and network link load of each node. When a smart camera node has high resource utilization or excessive link load, the dynamic routing mechanism will reallocate the link to other nodes with lower loads, ensuring balanced resource utilization across smart camera nodes throughout the network.

[0072] Specifically, this dynamic routing mechanism is based on the state of the nodes, specifically the state S of node i at time t. i The definition is as follows:

[0073]

[0074] Among them, RSSI i and RSSI avg E represents the signal strength at node i and the global average signal strength, respectively. i and E avg Let represent the remaining power of node i and the global average remaining power, respectively, and α be the scaling factor.

[0075] S3, the hierarchical aggregation network performs hierarchical aggregation of the local models uploaded by each smart camera node and forwards the recognition results through the relay cluster head node, and uploads the hierarchically aggregated cluster model and the forwarded recognition results to the central server.

[0076] Hierarchical aggregation networks, based on dynamic routing mechanisms, prioritize nodes in good condition as relay cluster heads, forwarding data from adjacent weaker nodes to form a multi-cluster topology, resulting in K relay cluster head nodes, and subsequently, clusters under these K relay cluster head nodes. For example... Figure 4 As shown, the relay cluster head node performs weighted aggregation on the local models uploaded by the M smart camera nodes within the cluster, based on the amount of data held locally by each node, to obtain the cluster model. This aggregation is performed before reaching the central server for global model aggregation, forming a hierarchical aggregation structure. Specifically, for cluster c, its cluster model w... c The aggregation method is as follows:

[0077]

[0078] Among them, |D i | represents the data size of the local dataset corresponding to the i-th local model in cluster c, M is the number of local models in cluster c, and w i Let i be the i-th local model.

[0079] The relay cluster head node will connect the cluster model w c The data is then sent to the central server. In one embodiment, if the network is complex enough, there may be multiple nested clusters. In this case, the current cluster model will need to be aggregated again at the next layer of relay cluster head nodes until it reaches the central server.

[0080] S4, the central server aggregates and updates the global model of the cluster model, distributes it to each smart camera node to update the local model, and summarizes the meter reading recognition results.

[0081] like Figure 4As shown, the central server receives cluster models uploaded from all hierarchical aggregation networks and updates the global model.

[0082] For time t+1, the global model is updated as follows:

[0083]

[0084] Among them, w g Let represent the global model at time t, η be the learning rate, and K be the number of relay cluster head nodes. w represents the total amount of data within cluster c. c It is a cluster model.

[0085] Each time the central server updates the global model, it generates a model file and pushes it to each smart camera node through a hierarchical aggregation network. Upon receiving the updated data, the smart camera nodes merge it with the corresponding parameters of their local models to generate the updated model. After aggregation, the smart camera nodes perform inference tests on the updated model to ensure its performance and stability.

[0086] In a preferred embodiment, the central server dynamically adjusts the operating parameters of the smart camera nodes based on monitoring data to optimize system performance and energy consumption. The monitoring data includes the network load and ambient lighting conditions of the smart camera nodes, while the operating parameters include shooting frequency, exposure time, and data transmission interval. When ambient lighting conditions are weaker than preset conditions, the shooting frequency of the smart camera nodes is reduced or the exposure time is increased to obtain clearer images. When the network load exceeds a preset threshold, the data transmission interval of the smart camera nodes is extended to reduce congestion in the hierarchical aggregation network.

[0087] In one preferred embodiment, the central server performs version management on the global model, including a rollback mechanism and a canary release strategy. Specifically, the rollback mechanism is as follows: when the performance of the central server's validation set drops beyond a preset threshold for a consecutive preset number of rounds, it automatically rolls back to the historical best global model. The canary release strategy is as follows: the updated global model is verified through canary release and first distributed to 10% of the smart camera nodes for trial operation. If there are no abnormal confidence alarms within 24 hours, it is then fully pushed to all smart camera nodes.

[0088] In this embodiment, the method also includes the dynamic addition and configuration of new smart camera nodes without manual intervention, improving system scalability and reducing maintenance costs. Specifically, when a new smart camera node is added, the hierarchical aggregation network automatically detects it and assigns it a unique network address; the new smart camera node establishes a connection with existing nodes through broadcast messages and downloads the necessary configuration files and initial models from the central server as its initial local model. The entire process requires no manual intervention, simplifying system maintenance and making it suitable for situations where substations are expanded or new equipment is added.

[0089] Example 3

[0090] Based on Example 2, this embodiment provides an application process for a distributed meter identification method, including the following steps:

[0091] Step 1: Deployment Phase. First, multiple smart camera nodes are deployed within the substation according to data acquisition requirements, ensuring coverage of all traditional meters that need monitoring. Each node's camera is aimed at the target meter, and the focus and angle are adjusted to obtain a clear image. The nodes have a pre-trained YOLO algorithm model and a Bluetooth Mesh protocol stack built-in.

[0092] Step 2: During the networking phase, after the smart camera node is powered on, it automatically broadcasts a network access request, prioritizes establishing a connection with the neighbor node with the best signal strength (RSSI > -70dBm), receives the network key and application key issued by the central server, and then builds a neighbor node table containing signal strength (RSSI) and remaining battery power, which is dynamically updated every 30 minutes to optimize the communication path.

[0093] Step 3: The smart camera nodes capture meter images at set time intervals and perform preliminary image preprocessing, including real-time correction of barrel distortion caused by wide-angle lenses and supplementary lighting strategies for common low-light scenarios in substations. The smart camera nodes identify meter readings using local models and perform confidence checks. If identification fails or the confidence level is below a set threshold, the original image is compressed and uploaded to the central server via a hierarchical aggregation network for further processing.

[0094] Step 4: The smart camera nodes use the newly acquired images and labels during operation as local datasets to further train their built-in local models. Each node's local model is then uploaded to the hierarchical aggregation network according to a preset reporting cycle.

[0095] Step 5: Based on the current status of each smart camera node and the network status (such as signal strength and network load), the hierarchical aggregation network plans the optimal multi-hop topology for transmission to the central server based on the dynamic routing algorithm, and forms relay cluster head nodes in the network.

[0096] Step 6: The local models uploaded by each smart camera node will be transmitted via the route planned by the hierarchical aggregation network. After receiving the local models relayed here, the cluster head relay node will first perform aggregation. The cluster models after hierarchical aggregation will then be sent to the central server for receiving and executing global model aggregation.

[0097] Step 7: The central server receives the cluster models and performs global model aggregation. The central server's global model version management includes a rollback mechanism and a canary release strategy. When the server-side validation set declines beyond a preset threshold for three consecutive rounds, the system automatically reverts to the historically best model (retaining the five most recent versions). The new model is validated through canary release: it is first deployed to 10% of nodes for trial operation; if no abnormal confidence alerts are received within 24 hours, the model update is pushed to all nodes.

[0098] The electronic device of this invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM may also store various programs and data required for device operation. The CPU, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0099] Multiple components in the device are connected to the I / O interface, including: input units such as keyboards and mice; output units such as various types of displays and speakers; storage units such as disks and optical discs; and communication units such as network interface cards (NICs), modems, and wireless transceivers. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0100] The processing unit executes the various methods and processes described above, such as methods S1 to S4. For example, in some embodiments, methods S1 to S4 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM and / or a communication unit. When the computer program is loaded into RAM and executed by the CPU, one or more steps of methods S1 to S4 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S4 by any other suitable means (e.g., by means of firmware).

[0101] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0102] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0103] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0104] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A distributed meter identification method, characterized in that, The method includes the following steps: Multiple smart camera nodes are used to collect meter images. A lightweight local model mounted on the smart camera nodes is used to perform local reading recognition on the meter images. A local dataset is constructed and the local model is trained. Each smart camera node is networked to form a hierarchical aggregation network. The hierarchical aggregation network dynamically plans multi-hop routes and dynamically determines relay cluster head nodes based on the remaining power and link signal strength of the smart camera nodes in the network. The hierarchical aggregation network uses relay cluster head nodes to perform hierarchical aggregation of local models uploaded by each smart camera node and forward the recognition results, and then uploads the hierarchically aggregated cluster model and the forwarded recognition results to the central server. The central server aggregates and updates the global model of the cluster model, distributes the updates to each smart camera node to update the local model, and summarizes the meter reading recognition results.

2. The distributed meter identification method according to claim 1, characterized in that, In the hierarchical aggregation network described above, the state S of node i at time t is... i The definition is as follows: Among them, RSSI i and RSSI avg E represents the signal strength at node i and the global average signal strength, respectively. i and E avg Let represent the remaining power of node i and the global average remaining power, respectively, and α be the scaling factor.

3. The distributed meter identification method according to claim 1, characterized in that, The hierarchical aggregation network, based on a dynamic routing mechanism, prioritizes nodes with good status as relay cluster head nodes, forwarding data from adjacent weaker nodes to form a multi-cluster topology. This results in multiple relay cluster head nodes and their corresponding clusters. Each relay cluster head node weights and aggregates the local models uploaded by multiple smart camera nodes within its cluster based on the amount of data held locally by each node, obtaining a cluster model. For cluster c, its cluster model w... c The aggregation method is as follows: Among them, |D i | represents the data size of the local dataset corresponding to the i-th local model in cluster c, M is the number of local models in cluster c, and w i Let i be the i-th local model.

4. The distributed meter identification method according to claim 1, characterized in that, The central server receives cluster models uploaded from all hierarchical aggregation networks and updates the global model. Specifically, for time t+1, the global model is updated as follows: Among them, w g Let represent the global model at time t, η be the learning rate, and K be the number of relay cluster head nodes. w represents the total amount of data within cluster c. c It is a cluster model.

5. The distributed meter identification method according to claim 1, characterized in that, The lightweight local model mounted in the smart camera node employs different recognition algorithms depending on the type of meter. For pointer-type meters, the local model performs the following steps: target detection is performed on the meter image to obtain the dial area, image segmentation is performed on the dial area to extract the pointer area, and the angular position of the pointer is calculated by combining geometric analysis methods. An angle-reading mapping relationship is established based on the range and scale distribution of the dial, and the meter reading is determined based on the calculated angle and angle-reading mapping relationship. For digital meters, the local model performs the following steps: locating the digital region of the meter image, using a character recognition model to extract the digital information on the meter, and obtaining the meter reading.

6. The distributed meter identification method according to claim 1, characterized in that, The central server dynamically adjusts the operating parameters of the smart camera nodes based on monitoring data to optimize system performance and energy consumption. The monitoring data includes the network load and ambient lighting conditions of the smart camera nodes. The operating parameters include shooting frequency, exposure time, and data transmission interval. When the ambient lighting conditions are weaker than preset conditions, the shooting frequency of the smart camera nodes is reduced or the exposure time is increased. When the network load is greater than a preset threshold, the data transmission interval of the smart camera nodes is extended.

7. The distributed meter identification method according to claim 1, characterized in that, The central server manages the global model in versions, including a rollback mechanism and a canary release strategy. Specifically, the rollback mechanism is as follows: when the performance of the central server's validation set drops beyond a preset threshold for a preset number of consecutive rounds, it automatically rolls back to the historical best global model. The canary release strategy is as follows: the updated global model is verified through canary release and first distributed to a preset proportion of smart camera nodes for trial operation. If there are no abnormal confidence alarms within a preset time interval, it is then fully pushed to all smart camera nodes.

8. The distributed meter identification method according to claim 1, characterized in that, The method further includes: While using a local model to identify meter readings, the confidence level of the identification result is determined. If the confidence level is greater than a preset threshold, the identification result is forwarded to the central server through a hierarchical aggregation network. Otherwise, the meter image is compressed and forwarded to the central server through the hierarchical aggregation network. The central server calls a high-precision identification model for identification and submits it for manual review.

9. The distributed meter identification method according to claim 1, characterized in that, The method further includes: When a new smart camera node is added, the hierarchical aggregation network automatically detects it and assigns it a unique network address. The new smart camera node establishes a connection with existing nodes by broadcasting messages and downloads the necessary configuration files and initial models from the central server as the initial local model.

10. A distributed meter identification system, characterized in that, The system for implementing the method as described in any one of claims 1-9 includes: Intelligent camera node: As a front-end sensing and edge computing unit, it is deployed in the meter monitoring area of ​​the substation. It supports wireless communication and is powered by a battery. It is equipped with a high-resolution camera and an edge computing module to collect meter images and perform local reading recognition and federated learning local model uploading. It also has local storage function to cache data that was not successfully transmitted. The hierarchical aggregation network is formed by multiple smart camera nodes. The nodes communicate with each other through the Bluetooth Low Energy Mesh network protocol. The hierarchical aggregation network dynamically plans multi-hop routes based on the remaining power and link signal strength of the smart camera nodes in the network, dynamically determines the relay cluster head node, and performs hierarchical aggregation of the local models uploaded by each smart camera node and forwards the recognition results through the relay cluster head node. The hierarchically aggregated cluster model and the forwarded recognition results are then uploaded to the central server. Central server: Used to summarize the reading recognition results forwarded by the hierarchical aggregation network, and execute the top-level aggregation algorithm of federated learning to aggregate the cluster model to update the global model, and distribute the updated global model to each smart camera node to update the local model.

Images