Looped network unit monitoring method and system based on Internet of Things

Abstract

The invention discloses a ring main unit monitoring method and system based on the Internet of Things, and the method comprises the steps: building a multi-source data collection network of a ring main unit, extracting the multi-dimensional time-frequency domain features of a signal through an edge computing device, generating a multi-source feature data set, and constructing a dynamic digital twinborn model, real-time mapping of the running state of the ring main unit is achieved through a physical-virtual synchronization algorithm, and twin data volumes are output; according to the node influence coefficient of the twin data volume, analyzing a topological structure of a ring network system by adopting a graph neural network, identifying a key fault propagation path and generating a priority monitoring node list; based on the priority monitoring node list and the twin data volume, dynamically generating an adaptive monitoring strategy through a reinforcement learning algorithm, executing the adaptive monitoring strategy and collecting feedback data in real time, and completing closed-loop iteration of the monitoring strategy through an online learning algorithm; through closed-loop control of multi-source perception-digital twinning-intelligent decision, high efficiency, intelligence and reliability of ring main unit monitoring are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution, and in particular to a ring network box monitoring method and system based on the Internet of Things. Background Art

[0002] With the rapid development of Internet of Things technology and the widespread application of smart grids, the ring network box monitoring system based on the Internet of Things has attracted increasing attention as an important part of power grid monitoring and management. The ring network box monitoring system based on the Internet of Things can collect and transmit the operation data of the ring network box in real time, and improve the monitoring accuracy and response speed of the power grid through intelligent analysis and processing, thereby enhancing the stability and reliability of the power grid. In addition, the application of Internet of Things technology makes the monitoring of the ring network box more flexible and efficient, which helps to realize the intelligent management and optimization of the power grid.

[0003] The existing technologies mainly rely on traditional monitoring equipment and data processing methods for IoT-based ring network box monitoring. Usually, the monitoring strategies are relatively fixed, and the real-time data collection and intelligent analysis capabilities of IoT devices are not fully utilized, and there is a lack of dynamic adjustment and optimization mechanisms. In addition, the existing technologies are insufficient in processing the distributed characteristics of ring network boxes and multi-source data fusion, and cannot achieve efficient monitoring and management in complex power grid environments. This results in limited response speed and accuracy of the monitoring system when the power grid responds to emergencies and load fluctuations, making it difficult to meet the requirements of modern power grids for high reliability and high efficiency.

[0004] In view of this, it is necessary to improve the existing IoT-based ring network box monitoring system to achieve more efficient, intelligent and dynamic monitoring management. Summary of the invention

[0005] The purpose of the present invention is to provide a ring network box monitoring method and system based on the Internet of Things to solve the above technical problems.

[0006] To achieve this object, the present invention adopts the following technical solutions: A ring network box monitoring method based on the Internet of Things, comprising: Establish a multi-source data acquisition network for the ring network box, collect current, voltage, temperature, vibration and electromagnetic radiation signals in real time through the sensor system deployed in the ring network box, and use edge computing devices to extract the multi-dimensional time and frequency domain features of the signals to generate a multi-source feature data set; A dynamic digital twin model is constructed based on the multi-source feature data set, and real-time mapping of the operation status of the ring network box is achieved through a physical-virtual synchronization algorithm, and a twin data body including equipment health evaluation and potential failure mode prediction is output; According to the node influence coefficient of the twin data body, a graph neural network is used to analyze the topology of the ring network system, identify key fault propagation paths and generate a priority monitoring node list; Based on the priority monitoring node list and the twin data body, an adaptive monitoring strategy is dynamically generated through a reinforcement learning algorithm, wherein the strategy includes a differentiated sampling frequency, a dynamic alarm threshold, and a resource allocation weight; The adaptive monitoring strategy is executed and feedback data is collected in real time. The digital twin model parameters are optimized through online learning algorithms, and predictive maintenance work orders are generated synchronously to complete the closed-loop iteration of the monitoring strategy.

[0007] Optionally, the multi-source data acquisition network of the ring network box is established, and the current, voltage, temperature, vibration and electromagnetic radiation signals are collected in real time by the sensor system deployed in the ring network box, and the multi-dimensional time-frequency domain features of the signals are extracted by edge computing equipment to generate a multi-source feature data set, which specifically includes: Deploy multi-type IoT sensor systems at key nodes of the ring network box, including current sensors, voltage sensors, temperature sensors, vibration sensors, and electromagnetic radiation sensors, and connect the sensors to the IoT data collection network through the low-power wide area network protocol; The raw signals collected by the sensor system are preprocessed using edge computing equipment, including denoising and filtering of current / voltage signals, time-frequency domain decomposition of vibration signals, and spectrum analysis of electromagnetic radiation signals, to generate a preliminarily processed multi-source signal data set; The feature extraction is performed on the multi-source signal data set after preliminary processing. The time-frequency domain features of the vibration signal are extracted by wavelet transform. The spectrum features of the electromagnetic radiation signal are analyzed by combining convolutional neural network to generate a multi-dimensional feature vector. The multi-dimensional feature vector is associated with the device identification information and geographic location information of the ring network box to construct a multi-source feature data set containing timestamp, device ID, feature vector and environmental label; The multi-source feature data set is compressed and encrypted through edge computing devices and uploaded to the cloud data warehouse to generate a standardized multi-source feature data set.

[0008] Optionally, a dynamic digital twin model is constructed based on the multi-source feature data set, and real-time mapping of the ring network box operation status is achieved through a physical-virtual synchronization algorithm, and a twin data body containing equipment health assessment and potential failure mode prediction is output, specifically including: Based on the equipment identification information and geographic location information in the multi-source feature data set, the initial three-dimensional geometric model of the ring network box is constructed, and the physical parameters of the model, including thermal conductivity and electromagnetic characteristics, are initialized through the finite element analysis method; Map the multi-dimensional feature vectors in the multi-source feature data set to the initial three-dimensional geometric model, and achieve real-time data synchronization between the physical device and the virtual model through the time series alignment algorithm to generate a preliminary synchronized digital twin; The Kalman filter algorithm is used to perform state estimation and error correction on the initially synchronized digital twin, optimize the real-time mapping accuracy of the model, and output a twin data body containing equipment operating status, environmental interference, and potential failure modes.

[0009] Optionally, the output includes a twin data body of equipment operating status, environmental interference and potential failure modes, and then further includes: Based on the equipment operation status data in the twin data body, the real-time health index RHI of the equipment is calculated through the health assessment algorithm, and the fault prediction model is trained in combination with historical fault data to generate equipment health assessment reports and potential failure mode prediction results; The equipment health assessment report and potential failure mode prediction results are fed back to the digital twin, and the physical parameters and fault prediction weights of the model are dynamically adjusted through the reinforcement learning algorithm.

[0010] Optionally, the node influence coefficient of the twin data body is used to analyze the topology of the ring network system using a graph neural network, identify key fault propagation paths and generate a priority monitoring node list, specifically including: Based on the equipment operation status data and topological connection relationship in the twin data body, a graph structure model of the ring network system is constructed, where the nodes of the graph structure model represent the ring network box equipment, the edges represent the electrical connection relationship between the equipment, and the attribute vectors of the nodes are initialized; Calculate the node influence coefficient NIF of each node, including electrical load weight, fault propagation probability and topological centrality index, and generate the node influence coefficient matrix; The node influence coefficient matrix is ​​input into the graph neural network (GNN) model, and the global and local features of the node are extracted through multi-layer graph convolution operations to output the potential fault propagation risk score of the node.

[0011] Optionally, extracting global features and local features of nodes through multi-layer graph convolution operations and outputting potential fault propagation risk scores of nodes, further includes: Based on the potential fault propagation risk score of nodes, a community detection algorithm is used to identify the key fault propagation paths in the ring network system and generate a fault propagation subgraph containing high-priority nodes. According to the node priority ranking in the fault propagation subgraph, combined with the real-time health index RHI of the equipment and historical fault data, a priority monitoring node list is generated, and the monitoring weight of each node is marked.

[0012] Optionally, based on the priority monitoring node list and the twin data body, an adaptive monitoring strategy is dynamically generated through a reinforcement learning algorithm, specifically including: Based on the node monitoring weights in the priority monitoring node list and the device operation status data in the twin data body, a reinforcement learning environment is constructed to define the state space, action space and reward function; The proximal policy optimization algorithm is used to train the reinforcement learning model, input state space data and generate action decisions through the policy network, and output the initial monitoring strategy including differentiated sampling frequency, dynamic alarm threshold and resource allocation weight; Apply the initial monitoring strategy to the ring network box monitoring system, collect feedback data in real time and evaluate the strategy effect through the value network, iteratively optimize the parameters of the strategy network based on the evaluation results, and generate an optimized adaptive monitoring strategy; The optimized adaptive monitoring strategy is fed back to the digital twin, and the strategy parameters are dynamically adjusted through the online learning algorithm.

[0013] Optionally, executing the adaptive monitoring strategy and collecting feedback data in real time, optimizing the digital twin model parameters through an online learning algorithm, synchronously generating a predictive maintenance work order, and completing the closed-loop iteration of the monitoring strategy specifically include: Execute adaptive monitoring strategies, collect the operation data of the ring network box in real time through the sensor system, and use edge computing devices to pre-process the data to generate real-time feedback data sets; Input the real-time feedback data set into the digital twin, update the model's operating status data through the physical-virtual synchronization algorithm, and use the Kalman filter algorithm to optimize the state estimation accuracy to generate an updated twin data body; Based on the updated twin data body, the fault prediction model is trained through the online random forest algorithm, and the historical fault data is combined with the real-time health index RHI to generate predictive maintenance work orders, marking the maintenance time, spare parts requirements and priority.

[0014] Optionally, the historical fault data is combined with the real-time health index RHI to generate a predictive maintenance work order, marking the maintenance time, spare parts requirements and priority, and then also includes: Feedback predictive maintenance work orders to the monitoring system, and dynamically adjust the parameters of the adaptive monitoring strategy, including sampling frequency, alarm threshold, and resource allocation weight, through reinforcement learning algorithms; Regularly reconstruct the physical parameters and fault prediction model of the digital twin based on real-time feedback data sets and predictive maintenance work orders, generate optimized digital twins, and support continuous iteration of monitoring strategies; Synchronize the optimized digital twin and adaptive monitoring strategy to the cloud data warehouse to complete the closed-loop iteration of the monitoring strategy.

[0015] The present invention also provides a ring network box monitoring system based on the Internet of Things, which is implemented by the ring network box monitoring method based on the Internet of Things as described above, and the monitoring system includes: The sensing system includes a current sensor, a voltage sensor, a temperature sensor, a vibration sensor and an electromagnetic radiation sensor, which are arranged at corresponding positions of the ring network box respectively; Digital twin module, used to output twin data bodies including equipment health assessment and potential failure mode prediction; Topology analysis module, used to generate a priority monitoring node list and optimize monitoring resource allocation; The adaptive strategy generation module is used to dynamically generate adaptive monitoring strategies based on the priority monitoring node list and the twin data body through the reinforcement learning algorithm; Closed-loop iterative optimization module, which is used to execute monitoring strategies and collect feedback data in real time, optimize digital twin model parameters, and generate predictive maintenance work orders to support closed-loop iteration of monitoring strategies; The communication module is used to realize data interaction between modules through the Internet of Things communication protocol.

[0016] Compared with the prior art, the present invention has the following beneficial effects: first, IoT sensors are deployed to collect current, voltage, temperature, vibration and electromagnetic radiation signals in real time, and edge computing devices are used to extract multi-dimensional time-frequency domain features to generate a multi-source feature data set; a dynamic digital twin model is constructed based on the data set, and real-time mapping of the operating status of the ring network box is achieved through a physical-virtual synchronization algorithm, and a twin data body containing equipment health assessment and potential fault mode prediction is output; then a graph neural network is used to analyze the topological structure of the ring network system, and the key fault propagation path is identified according to the node influence coefficient and a priority monitoring node list is generated; based on the priority monitoring node list and the twin data body, an adaptive monitoring strategy containing differentiated sampling frequency, dynamic alarm threshold and resource allocation weight is dynamically generated through a reinforcement learning algorithm; finally, the adaptive monitoring strategy is executed and feedback data is collected in real time, the digital twin model parameters are optimized through an online learning algorithm, and a predictive maintenance work order is synchronously generated to complete the closed-loop iteration of the monitoring strategy; this method realizes the high efficiency, intelligence and reliability of ring network box monitoring through the closed-loop control of multi-source perception-digital twin-intelligent decision-making, and provides a solution for the stable operation of the smart grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0018] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0019] Figure 1 This is one of the flow charts of the ring network box monitoring method based on the Internet of Things in the first embodiment of the present invention.

[0020] Figure 2 This is the second flow chart of the ring network box monitoring method based on the Internet of Things of the first embodiment. DETAILED DESCRIPTION

[0021] In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] In the description of the present invention, it should be understood that the terms "upper", "lower", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. It should be noted that when a component is considered to be "connected" to another component, it may be directly connected to the other component or there may be a centrally arranged component at the same time.

[0023] The technical solution of the present invention is further described below with reference to the accompanying drawings and through specific implementation methods.

[0024] Embodiment 1: Combination Figure 1 and Figure 2 As shown, an embodiment of the present invention provides a ring network box monitoring method based on the Internet of Things, including: S1, establish a multi-source data acquisition network for the ring network box, collect current, voltage, temperature, vibration and electromagnetic radiation signals in real time through the sensor system deployed in the ring network box, and use edge computing devices to extract the multi-dimensional time and frequency domain features of the signals to generate a multi-source feature data set; Through multimodal data fusion and dynamic digital twin modeling, the comprehensiveness and real-time performance of ring network box monitoring have been significantly improved, and it can capture mechanical vibration and electromagnetic radiation anomalies that are difficult to detect with traditional methods.

[0025] S2, builds a dynamic digital twin model based on multi-source feature data sets, realizes real-time mapping of the ring network box operation status through the physical-virtual synchronization algorithm, and outputs a twin data body containing equipment health assessment and potential failure mode prediction.

[0026] S3, based on the node influence coefficient of the twin data body, uses graph neural network to analyze the topological structure of the ring network system, identify key fault propagation paths and generate a priority monitoring node list; based on the topological analysis technology of graph neural network, it effectively identifies key fault propagation paths, optimizes the allocation of monitoring resources, and shortens the fault detection response time to less than 200ms.

[0027] S4, based on the priority monitoring node list and twin data body, dynamically generates an adaptive monitoring strategy through reinforcement learning algorithm. The strategy includes differentiated sampling frequency, dynamic alarm threshold and resource allocation weight; the adaptive monitoring strategy driven by reinforcement learning dynamically adjusts the sampling frequency and alarm threshold to reduce the false alarm rate.

[0028] S5 executes the adaptive monitoring strategy and collects feedback data in real time, optimizes the digital twin model parameters through online learning algorithms, and simultaneously generates predictive maintenance work orders to complete the closed-loop iteration of the monitoring strategy.

[0029] Through the synergy of online learning algorithms and predictive maintenance work orders, the accuracy of equipment health assessment is improved and unplanned downtime is significantly reduced.

[0030] The working principle of the present invention is as follows: first, IoT sensors are deployed to collect current, voltage, temperature, vibration and electromagnetic radiation signals in real time, and edge computing devices are used to extract multi-dimensional time-frequency domain features to generate a multi-source feature data set; a dynamic digital twin model is constructed based on the data set, and real-time mapping of the operating status of the ring network box is achieved through a physical-virtual synchronization algorithm, and a twin data body containing equipment health assessment and potential fault mode prediction is output; then, a graph neural network is used to analyze the topological structure of the ring network system, and the key fault propagation path is identified according to the node influence coefficient and a priority monitoring node list is generated; based on the priority monitoring node list and the twin data body, an adaptive monitoring strategy containing differentiated sampling frequency, dynamic alarm threshold and resource allocation weight is dynamically generated through a reinforcement learning algorithm; finally, the adaptive monitoring strategy is executed and feedback data is collected in real time, the digital twin model parameters are optimized through an online learning algorithm, and a predictive maintenance work order is synchronously generated to complete the closed-loop iteration of the monitoring strategy; this method realizes the high efficiency, intelligence and reliability of ring network box monitoring through the closed-loop control of multi-source perception-digital twin-intelligent decision-making, and provides a solution for the stable operation of the smart grid.

[0031] In this embodiment, it is specifically explained that step S1 specifically includes: S11, deploy multiple types of IoT sensor systems at key nodes of the ring network box. The sensor systems include current sensors, voltage sensors, temperature sensors, vibration sensors, and electromagnetic radiation sensors. The sensors are connected to the IoT data collection network through the low-power wide area network protocol. Deploy multiple types of IoT sensors (current, voltage, temperature, vibration and electromagnetic radiation sensors) at key nodes of the ring network box, and connect the sensors to the IoT data collection network through the low-power wide area network protocol (such as LoRaWAN). The low-power wide area network protocol has the characteristics of low power consumption and long-distance transmission, and is suitable for the distributed deployment environment of the ring network box.

[0032] Realize real-time collection of multimodal data and expand monitoring dimensions (such as vibration signal detection of mechanical failures and electromagnetic radiation signal evaluation of insulation degradation); low-power wide area network protocol ensures the stability and coverage of data transmission and is suitable for complex power grid environments.

[0033] S12, using edge computing equipment to pre-process the original signals collected by the sensor system, including denoising and filtering of current / voltage signals, time-frequency domain decomposition of vibration signals, and spectrum analysis of electromagnetic radiation signals, to generate a preliminarily processed multi-source signal data set; Edge computing devices are used to pre-process the raw signals collected by the sensors: Current / voltage signal: Kalman filtering is used to remove noise and improve signal quality; Vibration signal: Use short-time Fourier transform to decompose in the time-frequency domain and extract vibration characteristics; Electromagnetic radiation signal: Analyze the spectrum characteristics through fast Fourier transform (FFT) to identify abnormal harmonic components.

[0034] Edge computing devices complete preprocessing at the source of data, reducing the computing load on the cloud; signal denoising and feature extraction improve data quality and provide reliable input for subsequent analysis.

[0035] S13, extracting features from the preliminarily processed multi-source signal data set, extracting time-frequency domain features of the vibration signal through wavelet transform, analyzing the spectrum features of the electromagnetic radiation signal in combination with a convolutional neural network, and generating a multi-dimensional feature vector; Feature extraction is performed on the multi-source signal data set after preliminary processing: the vibration signal is extracted through wavelet transform to extract time-frequency domain features (such as energy entropy and spectrum peak); the electromagnetic radiation signal is analyzed by convolutional neural network (CNN) to analyze the spectrum features (such as harmonic components and spectrum distortion rate) to generate high-dimensional feature vectors.

[0036] The wavelet transform is combined with CNN to realize the quantitative analysis of unstructured data (such as vibration waveforms and electromagnetic spectra). The generated multi-dimensional feature vectors provide high-dimensional input for subsequent modeling and analysis.

[0037] S14, associating the multidimensional feature vector with the device identification information and geographic location information of the ring network box, and constructing a multi-source feature data set including a timestamp, a device ID, a feature vector and an environmental label.

[0038] S15, compress and encrypt the multi-source feature data set through the edge computing device, and upload it to the cloud data warehouse to generate a standardized multi-source feature data set.

[0039] Data compression reduces transmission bandwidth usage and improves transmission efficiency, while data encryption ensures the security of transmission and storage; standardized data sets support efficient processing of subsequent modeling and analysis.

[0040] In this embodiment, it is specifically described that step S2 specifically includes: S21, based on the equipment identification information and geographic location information in the multi-source feature data set, construct an initial three-dimensional geometric model of the ring network box, and initialize the physical parameters of the model, including thermal conductivity and electromagnetic characteristics, through a finite element analysis method; Based on the equipment identification information and geographic location information in the multi-source feature data set, the initial three-dimensional geometric model of the ring network box is constructed, and the physical parameters of the model, including thermal conductivity, mechanical stiffness and electromagnetic properties, are initialized through the finite element analysis (FEA) method. Finite element analysis converts complex physical field problems into computable mathematical models through discretization methods to ensure the accuracy and reliability of the model.

[0041] The initial three-dimensional geometric model provides the spatial structure and physical properties basis for the digital twin. Finite element analysis ensures the accuracy of the thermal-mechanical-electromagnetic multi-physics field coupling of the model and supports high-precision state mapping.

[0042] S22, mapping the multi-dimensional feature vectors in the multi-source feature data set to the initial three-dimensional geometric model, achieving real-time data synchronization between the physical device and the virtual model through a time series alignment algorithm, and generating a preliminarily synchronized digital twin; The multi-dimensional feature vectors in the multi-source feature data set are mapped to the initial three-dimensional geometric model, and the real-time data synchronization between the physical device and the virtual model is achieved through the time series alignment algorithm DTW. The DTW algorithm eliminates the time delay between the physical device and the virtual model through dynamic time warping to ensure synchronization accuracy.

[0043] Multi-dimensional feature vector mapping realizes real-time state synchronization between physical devices and virtual models. The time series alignment algorithm ensures that the synchronization error is less than 1ms, supporting high-precision fault prediction and state deduction.

[0044] S23, uses the Kalman filter algorithm to perform state estimation and error correction on the initially synchronized digital twin, optimizes the real-time mapping accuracy of the model, and outputs a twin data body containing the equipment operating status, environmental interference, and potential failure modes.

[0045] The Kalman filter algorithm is used to perform state estimation and error correction on the initially synchronized digital twin. The Kalman filter optimizes the state estimation accuracy and reduces noise interference by combining the system dynamics model with real-time observation data through a prediction-update cycle.

[0046] S24, based on the equipment operation status data in the twin data body, the real-time health index RHI of the equipment is calculated through the health assessment algorithm, and the fault prediction model is trained in combination with the historical fault data to generate the equipment health assessment report and potential failure mode prediction results; Based on the equipment operation status data in the twin data body, the real-time health index (RHI) of the equipment is calculated through the health assessment algorithm, and the fault prediction model (such as the LSTM model) is trained in combination with historical fault data. The RHI index integrates equipment operation parameters and environmental interference to quantify the health status of the equipment.

[0047] The health assessment algorithm generates a real-time health index (RHI), supports quantitative assessment of equipment status, and the fault prediction model (LSTM) has an accuracy rate of >85%, significantly improving fault prediction capabilities.

[0048] S25, feeds back the equipment health assessment report and potential failure mode prediction results to the digital twin, and dynamically adjusts the physical parameters and fault prediction weights of the model through the reinforcement learning algorithm.

[0049] Feedback the equipment health assessment report and potential failure mode prediction results to the digital twin, and dynamically adjust the model's physical parameters and failure prediction weights through reinforcement learning algorithms (such as the PPO algorithm). Reinforcement learning optimizes strategies through reward functions to improve the model's adaptability and prediction accuracy.

[0050] In this embodiment, it is specifically described that step S3 specifically includes: S31, based on the equipment operation status data and topological connection relationship in the twin data body, a graph structure model of the ring network system is constructed, wherein the nodes of the graph structure model represent the ring network box equipment, the edges represent the electrical connection relationship between the equipment, and the attribute vector of the node is initialized; Based on the equipment operation status data and topological connection relationship in the twin data body, a graph structure model of the ring network system is constructed. The nodes in the graph represent the ring network box equipment, the edges represent the electrical connection relationship between the equipment, and the attribute vector of the node is initialized, including multi-dimensional data such as electrical load, fault history, and environmental interference.

[0051] The graph structure model intuitively reflects the topological relationship of the ring network system and supports fault propagation path analysis; the node attribute vector contains multi-dimensional data, providing comprehensive input for subsequent node influence calculations.

[0052] S32, calculating the node influence coefficient NIF of each node, including the electrical load weight, the fault propagation probability and the topological centrality index, and generating a node influence coefficient matrix; The node influence coefficient NIF quantifies the importance of the device in fault propagation and supports the identification of key nodes; the node influence coefficient matrix provides high-dimensional input for the graph neural network to improve the accuracy of fault propagation risk scoring.

[0053] S33, input the node influence coefficient matrix into the graph neural network GNN model, extract the global and local features of the node through multi-layer graph convolution operations, and output the potential fault propagation risk score of the node.

[0054] Graph convolution operations capture the global and local features of nodes, improving the accuracy of fault propagation risk scoring; potential fault propagation risk scoring supports accurate identification of key fault propagation paths.

[0055] S34, based on the potential fault propagation risk score of the nodes, a community detection algorithm is used to identify the key fault propagation paths in the ring network system, and a fault propagation subgraph containing high priority nodes is generated; The community detection algorithm identifies key fault propagation paths, supports the optimal allocation of monitoring resources, and marks high-priority nodes in the fault propagation sub-map, providing a basis for subsequent monitoring strategy generation.

[0056] S35, according to the node priority ranking in the fault propagation subgraph, combined with the real-time health index RHI of the equipment and the historical fault data, a priority monitoring node list is generated, and the monitoring weight of each node is marked.

[0057] The priority monitoring node list optimizes monitoring resource allocation and improves monitoring efficiency. The monitoring weight annotation supports the generation of differentiated monitoring strategies to ensure real-time monitoring of key nodes.

[0058] In this embodiment, it is specifically described that step S4 specifically includes: S41, based on the node monitoring weights in the priority monitoring node list and the equipment operation status data in the twin data body, build a reinforcement learning environment, define the state space, action space and reward function; the state space includes the node health index, fault risk score and resource occupancy rate, the action space includes sampling frequency adjustment, alarm threshold setting and resource allocation weight, and the reward function comprehensively considers the response speed, false alarm rate and energy consumption optimization.

[0059] The state space comprehensively reflects the operating status of the equipment and resource usage, providing multi-dimensional input for strategy generation; the action space covers sampling frequency, alarm threshold and resource allocation weight, supporting the generation of differentiated monitoring strategies; the reward function balances response speed, false alarm rate and energy consumption optimization to ensure the global optimality of the strategy.

[0060] S42, uses the proximal policy optimization algorithm to train the reinforcement learning model, inputs the state space data and generates action decisions through the policy network, and outputs the initial monitoring strategy including differentiated sampling frequency, dynamic alarm threshold and resource allocation weight; The proximal policy optimization algorithm generates the initial monitoring strategy through the policy network, supporting the setting of differentiated sampling frequencies and dynamic alarm thresholds; the initial monitoring strategy optimizes the resource allocation weight to improve monitoring efficiency and fault response speed.

[0061] S43, applying the initial monitoring strategy to the ring network box monitoring system, collecting feedback data in real time and evaluating the strategy effect through the value network, iteratively optimizing the parameters of the strategy network based on the evaluation results, and generating an optimized adaptive monitoring strategy.

[0062] S44, feeds back the optimized adaptive monitoring strategy to the digital twin, and dynamically adjusts the strategy parameters through the online learning algorithm to ensure the continuous optimization and adaptability of the monitoring strategy.

[0063] In this embodiment, it is specifically described that step S5 specifically includes: S51 executes an adaptive monitoring strategy, collects the operating data of the ring network box in real time through the sensor system, and uses edge computing equipment to pre-process the data to generate a real-time feedback data set; the operating data includes current, voltage, temperature, vibration and electromagnetic radiation signals.

[0064] Execute adaptive monitoring strategies, collect the operation data of the ring network box (including current, voltage, temperature, vibration and electromagnetic radiation signals) in real time through the sensor system, and use edge computing devices to pre-process the data (such as denoising filtering and time-frequency domain decomposition) to generate a real-time feedback data set.

[0065] Collect multimodal data in real time to ensure the comprehensiveness and real-time nature of the monitoring strategy; edge computing devices complete preprocessing at the data source, reducing cloud computing load and improving data processing efficiency.

[0066] S52, input the real-time feedback data set into the digital twin, update the model's operating status data through the physical-virtual synchronization algorithm, and use the Kalman filter algorithm to optimize the state estimation accuracy to generate an updated twin data body; the physical-virtual synchronization algorithm ensures the real-time consistency between the digital twin and the physical device.

[0067] S53, based on the updated twin data body, trains the fault prediction model through the online random forest algorithm, combines historical fault data with the real-time health index RHI, generates predictive maintenance work orders, and marks maintenance time, spare parts requirements and priority. Predictive maintenance work orders optimize maintenance resource allocation and reduce unplanned downtime.

[0068] S54, feeds back the predictive maintenance work order to the monitoring system, and dynamically adjusts the parameters of the adaptive monitoring strategy, including sampling frequency, alarm threshold, and resource allocation weight, through a reinforcement learning algorithm; The reinforcement learning algorithm dynamically adjusts the monitoring strategy parameters to improve monitoring efficiency and fault response speed; the feedback mechanism supports the continuous optimization of the monitoring strategy to ensure the adaptability and accuracy of the strategy.

[0069] S55, regularly reconstructs the physical parameters and fault prediction model of the digital twin based on real-time feedback data sets and predictive maintenance work orders, generates optimized digital twins, and supports continuous iteration of monitoring strategies; The regular reconstruction mechanism ensures the long-term accuracy of the digital twin, and the optimized digital twin supports the continuous optimization and closed-loop iteration of the monitoring strategy.

[0070] S56, synchronizes the optimized digital twin and adaptive monitoring strategy to the cloud data warehouse to complete the closed-loop iteration of the monitoring strategy.

[0071] The cloud data warehouse supports efficient storage and sharing of data, and the closed-loop iteration mechanism ensures continuous optimization and long-term reliability of monitoring strategies.

[0072] Embodiment 2: The present invention also provides a ring network box monitoring system based on the Internet of Things, which is implemented by the ring network box monitoring method based on the Internet of Things as in the first embodiment. The monitoring system includes: The sensing system, including a current sensor, a voltage sensor, a temperature sensor, a vibration sensor and an electromagnetic radiation sensor, is respectively arranged at corresponding positions of the ring network box.

[0073] The digital twin module is used to output a twin data body that includes equipment health assessment and potential failure mode prediction.

[0074] The topology analysis module is used to generate a priority monitoring node list and optimize the allocation of monitoring resources.

[0075] The adaptive strategy generation module is used to dynamically generate adaptive monitoring strategies based on the priority monitoring node list and the twin data body through the reinforcement learning algorithm.

[0076] The closed-loop iterative optimization module is used to execute monitoring strategies and collect feedback data in real time, optimize digital twin model parameters, and generate predictive maintenance work orders to support closed-loop iteration of monitoring strategies.

[0077] The communication module is used to realize data interaction between modules through the Internet of Things communication protocol.

[0078] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A ring network box monitoring method based on the Internet of Things, characterized in that: include: Establish a multi-source data acquisition network for the ring network box, collect current, voltage, temperature, vibration and electromagnetic radiation signals in real time through the sensor system deployed in the ring network box, and use edge computing devices to extract the multi-dimensional time and frequency domain features of the signals to generate a multi-source feature data set; A dynamic digital twin model is constructed based on the multi-source feature data set, and real-time mapping of the operation status of the ring network box is achieved through a physical-virtual synchronization algorithm, and a twin data body including equipment health evaluation and potential failure mode prediction is output; According to the node influence coefficient of the twin data body, a graph neural network is used to analyze the topology of the ring network system, identify key fault propagation paths and generate a priority monitoring node list; Based on the priority monitoring node list and the twin data body, an adaptive monitoring strategy is dynamically generated through a reinforcement learning algorithm, wherein the strategy includes a differentiated sampling frequency, a dynamic alarm threshold, and a resource allocation weight; The adaptive monitoring strategy is executed and feedback data is collected in real time. The digital twin model parameters are optimized through online learning algorithms, and predictive maintenance work orders are generated synchronously to complete the closed-loop iteration of the monitoring strategy.

2. The ring network box monitoring method based on the Internet of Things according to claim 1 is characterized in that: The multi-source data acquisition network of the ring network box is established, and the current, voltage, temperature, vibration and electromagnetic radiation signals are collected in real time by the sensor system deployed in the ring network box, and the multi-dimensional time-frequency domain features of the signals are extracted by edge computing equipment to generate a multi-source feature data set, which specifically includes: Deploy multi-type IoT sensor systems at key nodes of the ring network box, including current sensors, voltage sensors, temperature sensors, vibration sensors, and electromagnetic radiation sensors, and connect the sensors to the IoT data collection network through the low-power wide area network protocol; The raw signals collected by the sensor system are preprocessed using edge computing equipment, including denoising and filtering of current / voltage signals, time-frequency domain decomposition of vibration signals, and spectrum analysis of electromagnetic radiation signals, to generate a preliminarily processed multi-source signal data set; The feature extraction is performed on the multi-source signal data set after preliminary processing. The time-frequency domain features of the vibration signal are extracted by wavelet transform. The spectrum features of the electromagnetic radiation signal are analyzed by combining convolutional neural network to generate a multi-dimensional feature vector. The multi-dimensional feature vector is associated with the device identification information and geographic location information of the ring network box to construct a multi-source feature data set containing timestamp, device ID, feature vector and environmental label; The multi-source feature data set is compressed and encrypted through edge computing devices and uploaded to the cloud data warehouse to generate a standardized multi-source feature data set.

3. The ring network box monitoring method based on the Internet of Things according to claim 2 is characterized in that: A dynamic digital twin model is constructed based on the multi-source feature data set, and the real-time mapping of the ring network box operation status is realized through the physical-virtual synchronization algorithm. The twin data body containing equipment health evaluation and potential failure mode prediction is output, which specifically includes: Based on the equipment identification information and geographic location information in the multi-source feature data set, the initial three-dimensional geometric model of the ring network box is constructed, and the physical parameters of the model, including thermal conductivity and electromagnetic characteristics, are initialized through the finite element analysis method; Map the multi-dimensional feature vectors in the multi-source feature data set to the initial three-dimensional geometric model, and achieve real-time data synchronization between the physical device and the virtual model through the time series alignment algorithm to generate a preliminary synchronized digital twin; The Kalman filter algorithm is used to perform state estimation and error correction on the initially synchronized digital twin, optimize the real-time mapping accuracy of the model, and output a twin data body containing equipment operating status, environmental interference, and potential failure modes.

4. The ring network box monitoring method based on the Internet of Things according to claim 3 is characterized in that: The output includes a twin data volume of equipment operating status, environmental interference and potential failure modes, followed by: Based on the equipment operation status data in the twin data body, the real-time health index RHI of the equipment is calculated through the health assessment algorithm, and the fault prediction model is trained in combination with historical fault data to generate equipment health assessment reports and potential failure mode prediction results; The equipment health assessment report and potential failure mode prediction results are fed back to the digital twin, and the physical parameters and fault prediction weights of the model are dynamically adjusted through the reinforcement learning algorithm.

5. The ring network box monitoring method based on the Internet of Things according to claim 1 is characterized in that: The node influence coefficient of the twin data body is used to analyze the topology of the ring network system using a graph neural network, identify key fault propagation paths and generate a priority monitoring node list, specifically including: Based on the equipment operation status data and topological connection relationship in the twin data body, a graph structure model of the ring network system is constructed, where the nodes of the graph structure model represent the ring network box equipment, the edges represent the electrical connection relationship between the equipment, and the attribute vectors of the nodes are initialized; Calculate the node influence coefficient NIF of each node, including electrical load weight, fault propagation probability and topological centrality index, and generate the node influence coefficient matrix; The node influence coefficient matrix is ​​input into the graph neural network (GNN) model, and the global and local features of the node are extracted through multi-layer graph convolution operations to output the potential fault propagation risk score of the node.

6. The ring network box monitoring method based on the Internet of Things according to claim 5 is characterized in that: The method extracts the global and local features of the node through multi-layer graph convolution operations, outputs the potential fault propagation risk score of the node, and then includes: Based on the potential fault propagation risk score of nodes, a community detection algorithm is used to identify the key fault propagation paths in the ring network system and generate a fault propagation subgraph containing high-priority nodes. According to the node priority ranking in the fault propagation subgraph, combined with the real-time health index RHI of the equipment and historical fault data, a priority monitoring node list is generated, and the monitoring weight of each node is marked.

7. The ring network box monitoring method based on the Internet of Things according to claim 1 is characterized in that: Based on the priority monitoring node list and the twin data body, an adaptive monitoring strategy is dynamically generated through a reinforcement learning algorithm, specifically including: Based on the node monitoring weights in the priority monitoring node list and the device operation status data in the twin data body, a reinforcement learning environment is constructed to define the state space, action space and reward function; The proximal policy optimization algorithm is used to train the reinforcement learning model, input state space data and generate action decisions through the policy network, and output the initial monitoring strategy including differentiated sampling frequency, dynamic alarm threshold and resource allocation weight; Apply the initial monitoring strategy to the ring network box monitoring system, collect feedback data in real time and evaluate the strategy effect through the value network, iteratively optimize the parameters of the strategy network based on the evaluation results, and generate an optimized adaptive monitoring strategy; The optimized adaptive monitoring strategy is fed back to the digital twin, and the strategy parameters are dynamically adjusted through the online learning algorithm.

8. The ring network box monitoring method based on the Internet of Things according to claim 1 is characterized in that: The said executing the said adaptive monitoring strategy and collecting feedback data in real time, optimizing the parameters of the digital twin model through the online learning algorithm, synchronously generating the predictive maintenance work order, and completing the closed-loop iteration of the monitoring strategy specifically include: Execute adaptive monitoring strategies, collect the operation data of the ring network box in real time through the sensor system, and use edge computing devices to pre-process the data to generate real-time feedback data sets; The real-time feedback data set is input into the digital twin, the operation status data of the model is updated through the physical-virtual synchronization algorithm, and the Kalman filter algorithm is used to optimize the state estimation accuracy to generate an updated twin data body; Based on the updated twin data body, the fault prediction model is trained through the online random forest algorithm, and the historical fault data is combined with the real-time health index RHI to generate predictive maintenance work orders, marking the maintenance time, spare parts requirements and priority.

9. The ring network box monitoring method based on the Internet of Things according to claim 8 is characterized in that: The above method combines historical fault data with the real-time health index (RHI) to generate a predictive maintenance work order, marking the maintenance time, spare parts requirements and priority, and then includes: Feedback predictive maintenance work orders to the monitoring system, and dynamically adjust the parameters of the adaptive monitoring strategy, including sampling frequency, alarm threshold, and resource allocation weight, through reinforcement learning algorithms; Regularly reconstruct the physical parameters and fault prediction model of the digital twin based on real-time feedback data sets and predictive maintenance work orders, generate optimized digital twins, and support continuous iteration of monitoring strategies; Synchronize the optimized digital twin and adaptive monitoring strategy to the cloud data warehouse to complete the closed-loop iteration of the monitoring strategy.

10. A ring network box monitoring system based on the Internet of Things, characterized in that: The ring network box monitoring method based on the Internet of Things is implemented by any one of claims 1 to 9, wherein the monitoring system comprises: The sensing system includes a current sensor, a voltage sensor, a temperature sensor, a vibration sensor and an electromagnetic radiation sensor, which are arranged at corresponding positions of the ring network box respectively; Digital twin module, used to output twin data bodies including equipment health assessment and potential failure mode prediction; Topology analysis module, used to generate a priority monitoring node list and optimize monitoring resource allocation; The adaptive strategy generation module is used to dynamically generate adaptive monitoring strategies based on the priority monitoring node list and the twin data body through the reinforcement learning algorithm; Closed-loop iterative optimization module, which is used to execute monitoring strategies and collect feedback data in real time, optimize digital twin model parameters, and generate predictive maintenance work orders to support closed-loop iteration of monitoring strategies; The communication module is used to realize data interaction between modules through the Internet of Things communication protocol.

Images