An integrated AI diagnosis ring network cabinet state remote early warning maintenance method and system

Abstract

The application relates to an integrated AI diagnosis ring network cabinet state remote early warning maintenance method and system, and belongs to the technical field of intelligent operation and maintenance and monitoring control of power equipment. The method comprises the following steps: collecting distribution branch load data, constructing a temperature-current dual-mode grid to simulate load testing of a ring network cabinet, and generating ring network cabinet distribution branch state data; an intelligent ring network cabinet monitoring and diagnosis model is constructed, voltage-current curves under normal and abnormal loads are analyzed, and a ring network cabinet remote maintenance scheme is generated; the ring network cabinet remote maintenance scheme is simulated by a digital simulation control method, the insulation degradation period of the ring network cabinet is verified based on an electromagnetic interference simulation method, and ring network cabinet state early warning information is obtained; a multi-level risk early warning value is set based on the ring network cabinet remote maintenance scheme, a maintenance period and an inspection strategy are dynamically optimized, and ring network cabinet remote maintenance instructions and ring network cabinet state early warning information are pushed through a cloud monitoring platform.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance and monitoring and control technology of power equipment, specifically involving a remote early warning and maintenance method and system for ring main unit status integrated with AI diagnostics. Background Technology

[0002] Ring main units (RMS) are crucial equipment in power distribution systems, widely used in urban, industrial, and residential power distribution. Their primary functions include segmented protection of power lines, load regulation, and fault isolation to ensure the safe and stable operation of the power distribution system. However, with increasing electricity demand and aging equipment, traditional RMS units face numerous challenges, including insufficient equipment condition monitoring, difficulties in fault prediction, and low maintenance efficiency.

[0003] Currently, monitoring of ring main units typically relies on traditional manual inspections and periodic maintenance, mostly employing localized physical testing methods such as infrared imaging and temperature monitoring. However, these methods often fail to provide a comprehensive understanding of the equipment's operating status and are limited by the time window and spatial scope of the inspection, making real-time monitoring of the equipment's health impossible. Furthermore, traditional ring main unit monitoring lacks intelligent early warning mechanisms, exhibiting weak predictive capabilities for potential equipment failure risks and insulation degradation. This can easily lead to the failure to detect abnormal equipment operation in a timely manner, or even delays in maintenance.

[0004] In recent years, with the development of artificial intelligence (AI), digital twins, and the Internet of Things (IoT) technologies, intelligent operation and maintenance (O&M) technologies have been increasingly applied in power equipment management. Utilizing AI algorithms to analyze equipment operation data in real time, combined with digital twin technology to build virtual equipment models, has become an important means to improve the efficiency and accuracy of equipment O&M, enabling comprehensive monitoring of equipment operating status and fault prediction. However, intelligent monitoring and remote maintenance solutions for power equipment such as ring main units still face technical challenges, particularly in load testing, anomaly monitoring, and insulation condition assessment. A more efficient and accurate solution is urgently needed to meet the requirements of modern power grids for the safe and reliable operation of equipment.

[0005] Therefore, this invention proposes a remote early warning and maintenance method for ring main units based on AI diagnosis. Through precise data collection and intelligent model diagnosis, it solves the shortcomings of traditional ring main unit monitoring methods and can effectively realize real-time monitoring of equipment status, fault prediction and remote maintenance. Summary of the Invention

[0006] To address the aforementioned problems in existing technologies, this invention provides a remote early warning and maintenance method for ring main unit status that integrates AI diagnostics.

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

[0008] S1: Collect power distribution branch load data. The power distribution branch load data is based on the pulse current method when the ring main unit is partially discharged. The instantaneous pulse current signal is analyzed by the intelligent load analysis algorithm, and the temperature-current dual-mode grid is constructed according to the instantaneous pulse current signal to simulate the load test of the ring main unit and generate the power distribution branch status data of the ring main unit.

[0009] S2: Construct an intelligent ring main unit monitoring and diagnostic model. Input the status data of the power distribution branch of the ring main unit as input parameters into the intelligent ring main unit monitoring and diagnostic model. Use deep neural networks to predict and analyze the voltage-current curves under normal and abnormal loads. Combine the changes in the voltage-current curves to set the ring main unit maintenance information. Based on intelligent drive, simulate the operating status of the ring main unit and generate a remote maintenance plan for the ring main unit.

[0010] S3: Simulate the remote maintenance scheme of the ring main unit by digital simulation control method, analyze the operating status data of the ring main unit, extract the response delay, harmonics and failure rate data in the ring main unit operation log through convolutional neural network for the simulated load test, and verify the insulation degradation cycle of the ring main unit based on electromagnetic countermeasure interference network to obtain the ring main unit status early warning information.

[0011] S4: Based on the remote maintenance scheme of the ring main unit, set multi-level risk warning values. When the current power distribution branch data of the ring main unit exceeds the risk warning value, dynamically optimize the maintenance cycle and inspection strategy, and push the remote maintenance command and ring main unit status warning information to the maintenance personnel through the cloud intelligent reasoning monitoring platform.

[0012] Specifically, the method for collecting the load data of the distribution branch is as follows: analyze the operating parameters during the operation of the ring main unit, including current data, temperature data, and operation log data, perform time-series mode fusion on the operating parameters, and combine the switching instructions and feedback signals in the operation log data to obtain the load data of the distribution branch.

[0013] Specifically, the method for generating the status data of the ring main unit's power distribution branches is as follows:

[0014] Based on the temperature-current dual-mode grid, segmented loads are applied to the ring main unit, and corresponding current response curves are collected at different load levels. The time when the ring main unit switching command is issued and the start time of the actual voltage and current curve response are recorded to obtain the interference-free response range during the power supply switching operation of the ring main unit's distribution branch.

[0015] The response delay parameters are structured and stored according to the interference-free interval, and the ring main unit power distribution dataset is generated by combining the current response curve during the power supply load application process. The ring main unit health dataset is used as the basic input parameter of the intelligent ring main unit monitoring and diagnosis model, and the current response curve is reconstructed by polynomial fitting to generate the ring main unit power distribution branch status data.

[0016] Specifically, the temperature-current dual-mode mesh is divided into mesh nodes based on the load stroke of the ring main unit. The temperature data and current data are paired in dual modes, and the corresponding current characteristic parameters and temperature characteristic parameters are recorded on each node. The mesh is interpolated and densified according to the node characteristic parameters to simulate the power distribution load of the ring main unit.

[0017] Specifically, the method for constructing the intelligent ring network cabinet monitoring and diagnostic model is as follows:

[0018] Based on the current characteristics extracted during the operation of the ring main unit from the simulated load test, the unstructured power distribution operation status data is transformed into a set of extracted response delay and fault parameters. Combined with the response delay in the ring main unit operation log and the status of the ring main unit power distribution branch, a remote maintenance feature matrix of the ring main unit is generated.

[0019] Based on the remote maintenance feature matrix of the ring main unit, the response delay and fault parameter set are received, the ring main unit maintenance monitoring features are extracted and processed through feature vectors, and power distribution maintenance weights are allocated according to the ring main unit maintenance monitoring features.

[0020] Based on the power distribution maintenance weight analysis of the ring main unit operation execution data and feedback results, and under the constraint of the temperature-current dual-modal grid, the parameter weights of the intelligent ring main unit monitoring and diagnosis model for monitoring the status data of the ring main unit power distribution branches are dynamically corrected by gradient descent, thereby constructing the intelligent ring main unit monitoring and diagnosis model.

[0021] Specifically, the simulated load test simulates the micro-load execution process of the ring main unit by synchronously sampling the switching response of the interference-free ring main unit, and verifies the current curve scheme based on the test results.

[0022] Specifically, the verification method for the insulation degradation cycle of the ring main unit is as follows: obtain the failure rate data of the insulation parts during the operation of the ring main unit, conduct accelerated aging tests on the insulation components of the ring main unit through electromagnetic interference simulation, determine the degradation rate index based on the changes in characteristic parameters before and after accelerated aging, and perform curve fitting between the degradation rate index and the operation data of the remote maintenance scheme of the ring main unit to calculate the insulation degradation cycle of the ring main unit.

[0023] Specifically, the ring main unit status early warning information is constructed by a recurrent neural network to form a time-series feature matrix, and the health score of each distribution branch is calculated based on the time-series feature matrix. When the health score is lower than the risk warning value, the ring main unit status early warning information corresponding to the risk level is output through the intelligent ring main unit monitoring and diagnosis model.

[0024] Specifically, the process of pushing remote maintenance instructions for the ring main unit is as follows: the maintenance instructions generated in the cloud are sent to the maintenance terminal platform based on the multi-stream communication transmission protocol, and the on-site execution data of the ring main unit is cached and verified according to the edge computing node. After receiving the maintenance instructions, the maintenance terminal platform automatically generates an execution confirmation signal and verifies the maintenance instructions in combination with the feedback information of the execution confirmation signal.

[0025] Specifically, the voltage-current curve is based on the power distribution branch of the ring main unit. By extracting the peak value, phase difference, and response delay parameters of the curve during load changes and switching command triggering, the abnormal fluctuation characteristics of the ring main unit under different load conditions are identified.

[0026] Specifically, the simulation process of the digital simulation control method is as follows: the control commands in the remote maintenance scheme are mapped to the virtual space to perform load switching and insulation status simulation modeling, and the simulation modeling results are combined with the differences between the simulation modeling results and the abnormal fluctuation characteristics of the voltage-current response curve measured by the actual object to generate the ring main unit simulation maintenance feedback results.

[0027] Specifically, a remote early warning and maintenance system for ring main unit status integrating AI diagnostics is characterized by comprising:

[0028] Distribution data acquisition module: collects distribution branch load data. The distribution branch load data is based on the pulse current method. When the ring main unit is partially discharged, the instantaneous pulse current signal is analyzed by the intelligent load analysis algorithm. The instantaneous pulse current signal is used to construct a temperature-current dual-modal grid to simulate the load test of the ring main unit and generate the distribution branch status data of the ring main unit.

[0029] Intelligent detection and diagnosis module: Constructs an intelligent ring main unit monitoring and diagnosis model, inputs the status data of the power distribution branch of the ring main unit as input parameters into the intelligent ring main unit monitoring and diagnosis model, predicts and analyzes the voltage-current curves under normal and abnormal loads through deep neural networks, sets ring main unit maintenance information based on the changes in the voltage-current curves, simulates the operating status of the ring main unit based on intelligent drive, and generates a remote maintenance plan for the ring main unit.

[0030] Digital simulation control module: Simulates the execution of the remote maintenance scheme of the ring main unit through digital simulation control method, analyzes the operating status data of the ring main unit, extracts response delay, harmonic and failure rate data from the ring main unit operation log through convolutional neural network for the simulated load test, and verifies the insulation degradation cycle of the ring main unit based on electromagnetic countermeasure interference network to obtain ring main unit status early warning information;

[0031] Maintenance push early warning module: Based on the remote maintenance scheme of the ring main unit, multi-level risk early warning values ​​are set. When the current power distribution branch data of the ring main unit exceeds the risk early warning value, the maintenance cycle and inspection strategy are dynamically optimized, and remote maintenance instructions and ring main unit status early warning information are pushed to the maintenance personnel through the cloud intelligent reasoning monitoring platform.

[0032] The beneficial effects of this invention are as follows:

[0033] This invention provides a remote early warning and maintenance method for ring main unit status that integrates AI diagnostics. By introducing artificial intelligence, digital twin simulation technology and intelligent monitoring and diagnostic models, it solves many shortcomings in existing ring main unit monitoring technologies and has significant technical advantages.

[0034] First, this invention, based on the combination of pulse current method and temperature-current dual-mode grid, enables accurate simulated load testing of ring main unit (RMU) distribution branches. Traditional load monitoring methods often have limitations and cannot fully reflect the true performance of the RMU under different operating conditions. This invention, through precise instantaneous pulse current signal analysis and temperature data fusion, can obtain the voltage-current response curves of the RMU under normal and abnormal loads, comprehensively assessing the insulation performance and health status of the RMU, greatly improving the accuracy and reliability of the test.

[0035] Secondly, the construction of the intelligent ring main unit monitoring and diagnostic model makes the monitoring of the ring main unit's operating status more intelligent. By analyzing changes in the voltage-current curve through AI algorithms, the system can detect abnormal loads in the equipment in real time and automatically generate remote maintenance plans based on the diagnostic results, reducing the need for manual intervention and improving the efficiency of automated monitoring and maintenance of the ring main unit. In addition, the application of digital twin simulation control method can simulate the remote maintenance plan of the ring main unit in a virtual environment, and optimize and adjust it by analyzing the equipment status data in real time, effectively improving the accuracy and real-time performance of remote maintenance.

[0036] Finally, the remote early warning and maintenance push module of the present invention, by setting multi-level risk warning values, can dynamically optimize the maintenance cycle and inspection strategy when the data of the power distribution branch of the ring main unit exceeds the warning threshold, and push remote maintenance instructions to the maintenance personnel through the cloud platform, thereby realizing intelligent and refined management of ring main unit operation and maintenance, significantly improving the reliability and safety of equipment operation, and reducing the cost of manual inspection and maintenance.

[0037] In summary, this invention, through the comprehensive application of multiple advanced technologies, not only improves the monitoring accuracy and early warning capabilities of ring main units, but also optimizes remote maintenance schemes, reduces maintenance costs, and enhances the reliability and security of power systems, demonstrating significant technical advantages and broad application prospects. Attached Figure Description

[0038] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0039] Figure 1 This is a schematic diagram of the structure of a remote early warning and maintenance method and system for ring main unit status integrating AI diagnostics according to the present invention.

[0040] Figure 2 This is a schematic diagram illustrating the overall technical flow of a remote early warning and maintenance method and system for ring main unit status integrating AI diagnostics, as described in this invention. Detailed Implementation

[0041] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0042] Please see Figure 1 A remote early warning and maintenance method for ring main unit status integrating AI diagnostics:

[0043] S1: Collect power distribution branch load data. The power distribution branch load data is based on the pulse current method when the ring main unit is partially discharged. The instantaneous pulse current signal is analyzed by the intelligent load analysis algorithm, and the temperature-current dual-mode grid is constructed according to the instantaneous pulse current signal to simulate the load test of the ring main unit and generate the power distribution branch status data of the ring main unit.

[0044] S2: Construct an intelligent ring main unit monitoring and diagnostic model. Input the status data of the power distribution branch of the ring main unit as input parameters into the intelligent ring main unit monitoring and diagnostic model. Use deep neural networks to predict and analyze the voltage-current curves under normal and abnormal loads. Combine the changes in the voltage-current curves to set the ring main unit maintenance information. Based on intelligent drive, simulate the operating status of the ring main unit and generate a remote maintenance plan for the ring main unit.

[0045] S3: Simulate the remote maintenance scheme of the ring main unit by digital simulation control method, analyze the operating status data of the ring main unit, extract the response delay, harmonics and failure rate data in the ring main unit operation log through convolutional neural network for the simulated load test, and verify the insulation degradation cycle of the ring main unit based on electromagnetic countermeasure interference network to obtain the ring main unit status early warning information.

[0046] S4: Based on the remote maintenance scheme of the ring main unit, set multi-level risk warning values. When the current power distribution branch data of the ring main unit exceeds the risk warning value, dynamically optimize the maintenance cycle and inspection strategy, and push the remote maintenance command and ring main unit status warning information to the maintenance personnel through the cloud intelligent reasoning monitoring platform.

[0047] Specifically, the method for collecting the load data of the distribution branch is as follows: analyze the operating parameters during the operation of the ring main unit, including current data, temperature data, and operation log data, perform time-series mode fusion on the operating parameters, and combine the switching instructions and feedback signals in the operation log data to obtain the load data of the distribution branch.

[0048] Specifically, the method for generating the status data of the ring main unit's power distribution branches is as follows:

[0049] Based on the temperature-current dual-mode grid, segmented loads are applied to the ring main unit, and corresponding current response curves are collected at different load levels. The time when the ring main unit switching command is issued and the start time of the actual voltage and current curve response are recorded to obtain the interference-free response range during the power supply switching operation of the ring main unit's distribution branch.

[0050] The response delay parameters are structured and stored according to the interference-free interval, and the ring main unit power distribution dataset is generated by combining the current response curve during the power supply load application process. The ring main unit health dataset is used as the basic input parameter of the intelligent ring main unit monitoring and diagnosis model, and the current response curve is reconstructed by polynomial fitting to generate the ring main unit power distribution branch status data.

[0051] Specifically, the temperature-current dual-mode mesh is divided into mesh nodes based on the load stroke of the ring main unit. The temperature data and current data are paired in dual modes, and the corresponding current characteristic parameters and temperature characteristic parameters are recorded on each node. The mesh is interpolated and densified according to the node characteristic parameters to simulate the power distribution load of the ring main unit.

[0052] In this embodiment, as Figure 2 The overall data flow of the remote early warning and maintenance system for the ring main unit shown is as follows:

[0053] Distribution branch load data acquisition → Data preprocessing and fusion module → Intelligent ring network cabinet monitoring and diagnostic model (based on deep learning and time series analysis) → Real-time maintenance command generation bus (based on EtherCAT protocol) → (ΔT feature extraction engine | Raw state data stored in InfluxDB for offline model training) → (Feature vector written to local database | Remote maintenance scheme optimization and dynamic learning) → Remote control command API → Servo drive and execution module.

[0054] The real-time monitoring module enables a ΔT continuous integral monitoring mode, overlaid with a Watermark delay compensation mechanism, allowing the maximum out-of-order time of the ring main unit's sensor data to not exceed 100ms, ensuring accurate alignment between equipment operation status analysis and maintenance command generation. Feedback weight decay under different operating loads employs a piecewise adaptive-load decay function (PALDF), the formula of which is:

[0055] ,

[0056] Where Δt=t conrrol -t feedback , τ crit For load-sensitive inflection point, a i b i This function dynamically assigns parameters from the ring main unit's load attribute library (mapped according to the current-voltage curve and temperature changes). It implements a dynamic scheduling strategy of "rapid attenuation under high load and delayed attenuation under low load," ensuring the system responds promptly to load changes under different operating environments.

[0057] A three-dimensional control matrix (i.e., power distribution branches, load regulation, and insulation performance monitoring) was constructed by dynamically associating and modeling the various power distribution branches of the ring main unit. The key parameters in the matrix are:

[0058] ,

[0059] Among them, W r (t) represents the load response weight of the distribution branch, W f (t) represents the voltage response weight, W c γ(t) represents the temperature response weight, γ(t) represents the coupling compensation factor for each distribution branch, and µ(t) represents the rebound memory factor. The weighting factors are derived by reverse engineering from historical load data.

[0060] ,

[0061] Where λ is the adaptive learning and memory coefficient. When the detected current / voltage fluctuation exceeds the safety threshold, the system triggers the "shaft system independent compensation mechanism," which dynamically optimizes the equipment's operation by intelligently adjusting the load and operating status of each power distribution branch.

[0062] In the remote maintenance module of the ring main unit, when the equipment load monitoring data exceeds the predetermined risk warning threshold, the system pushes remote maintenance instructions to maintenance personnel through the cloud platform, guiding operators to troubleshoot and perform maintenance adjustments. The remote diagnostic and maintenance solutions pushed in real time through the cloud analysis platform can effectively reduce the equipment failure rate and improve the overall operational stability of the system.

[0063] Specifically, the method for constructing the intelligent ring network cabinet monitoring and diagnostic model is as follows:

[0064] Based on the current characteristics extracted during the operation of the ring main unit from the simulated load test, the unstructured power distribution operation status data is transformed into a set of extracted response delay and fault parameters. Combined with the response delay in the ring main unit operation log and the status of the ring main unit power distribution branch, a remote maintenance feature matrix of the ring main unit is generated.

[0065] Based on the remote maintenance feature matrix of the ring main unit, the response delay and fault parameter set are received, the ring main unit maintenance monitoring features are extracted and processed through feature vectors, and power distribution maintenance weights are allocated according to the ring main unit maintenance monitoring features.

[0066] Based on the power distribution maintenance weight analysis of the ring main unit operation execution data and feedback results, and under the constraint of the temperature-current dual-modal grid, the parameter weights of the intelligent ring main unit monitoring and diagnosis model for monitoring the status data of the ring main unit power distribution branches are dynamically corrected by gradient descent, thereby constructing the intelligent ring main unit monitoring and diagnosis model.

[0067] Specifically, the simulated load test simulates the micro-load execution process of the ring main unit by synchronously sampling the switching response of the interference-free ring main unit, and verifies the current curve scheme based on the test results.

[0068] Specifically, the verification method for the insulation degradation cycle of the ring main unit is as follows: obtain the failure rate data of the insulation parts during the operation of the ring main unit, conduct accelerated aging tests on the insulation components of the ring main unit through electromagnetic interference simulation, determine the degradation rate index based on the changes in characteristic parameters before and after accelerated aging, and perform curve fitting between the degradation rate index and the operation data of the remote maintenance scheme of the ring main unit to calculate the insulation degradation cycle of the ring main unit.

[0069] Specifically, the ring main unit status early warning information is constructed by a recurrent neural network to form a time-series feature matrix, and the health score of each distribution branch is calculated based on the time-series feature matrix. When the health score is lower than the risk warning value, the ring main unit status early warning information corresponding to the risk level is output through the intelligent ring main unit monitoring and diagnosis model.

[0070] In this embodiment, the intelligent operation scenario of the 35kV distribution ring network cabinet in the industrial park is achieved by building a unified operation and maintenance platform through artificial intelligence-optimized operating system, realizing a closed-loop operation and maintenance process of equipment status acquisition, intelligent diagnosis, simulation verification and adaptive maintenance strategy generation.

[0071] 1. System operating environment

[0072] The system is deployed on an AI-Enhanced OS, which incorporates: a multi-threaded neural network inference engine; a high-dimensional matrix operation accelerator; a multi-modal feature shared memory module; and a low-latency heterogeneous scheduling kernel. It also loads AI middleware as the cross-module data exchange backbone, enabling: dynamic orchestration of model services; unified scheduling of data and event streams; collaborative computing among multiple models; and adaptive allocation of heterogeneous computing nodes, providing a highly reliable AI execution environment for ring network cabinet monitoring, prediction, and early warning.

[0073] 2. Multi-source data acquisition and intelligent preprocessing

[0074] The ring main unit is equipped with various sensors, including those for current, voltage, temperature, partial discharge, and sound; a visual acquisition unit is also deployed. The data acquisition module loads computer vision and audio analytics software to perform the following tasks: visual temperature rise hotspot identification; partial discharge spectral feature extraction; mechanical motion noise signal analysis; and depth image recognition of the switching mechanism status.

[0075] The software components used for data collection call functions from the AI ​​function library:

[0076] High-frequency filtering function (HF Filter);

[0077] Multi-scale temporal convolution operator (MS-TCN);

[0078] Visual inspection operator (CV-Op);

[0079] Transformer noise depth separation operator (Audio-DSS).

[0080] Real-time enhancement, noise reduction, and structuring of electrical and physical signals.

[0081] 3. Intelligent diagnostic model and health assessment

[0082] Multimodal features are fed into an intelligent diagnostic model that runs on AI middleware. This model consists of the following components:

[0083] Voltage-current curve auto-encoder analyzer;

[0084] Dual-modal temperature-current behavior graph neural network (TC-GNN).

[0085] The Transformer module performs log sequence analysis.

[0086] The AI ​​function library provides the Dynamic Attention Operator (DA-Op) and the Topology Preserving Mapping Operator (TPM-Op).

[0087] By automatically identifying abnormal loads, abnormal response delays, and harmonic distortion trends, the system infers the insulation degradation rate and potential fault locations, and outputs the equipment health status and risk level.

[0088] 4. Digital Twin and Intelligent Simulation Verification

[0089] A digital twin of the ring main unit is constructed by calling an AI simulation function library, including: an electromagnetic field simulation module; an insulation aging simulation module; and a load change response prediction module.

[0090] Real-time simulations are performed on an AI-optimized operating system to cross-validate the prediction results: the difference between actual and simulated current waveforms is compared synchronously, the fault evolution path of high-risk nodes is inferred through simulation, and the parameter weights of the diagnostic model are automatically corrected.

[0091] Specifically, the process of pushing remote maintenance instructions for the ring main unit is as follows: the maintenance instructions generated in the cloud are sent to the maintenance terminal platform based on the multi-stream communication transmission protocol, and the on-site execution data of the ring main unit is cached and verified according to the edge computing node. After receiving the maintenance instructions, the maintenance terminal platform automatically generates an execution confirmation signal and verifies the maintenance instructions in combination with the feedback information of the execution confirmation signal.

[0092] Specifically, the voltage-current curve is based on the power distribution branch of the ring main unit. By extracting the peak value, phase difference, and response delay parameters of the curve during load changes and switching command triggering, the abnormal fluctuation characteristics of the ring main unit under different load conditions are identified.

[0093] Specifically, the simulation process of the digital simulation control method is as follows: the control commands in the remote maintenance scheme are mapped to the virtual space to perform load switching and insulation status simulation modeling, and the simulation modeling results are combined with the differences between the simulation modeling results and the abnormal fluctuation characteristics of the voltage-current response curve measured by the actual object to generate the ring main unit simulation maintenance feedback results.

[0094] Specifically, a remote early warning and maintenance system for ring main unit status integrating AI diagnostics is characterized by comprising:

[0095] Distribution data acquisition module: collects distribution branch load data. The distribution branch load data is based on the pulse current method. When the ring main unit is partially discharged, the instantaneous pulse current signal is analyzed by the intelligent load analysis algorithm. The instantaneous pulse current signal is used to construct a temperature-current dual-modal grid to simulate the load test of the ring main unit and generate the distribution branch status data of the ring main unit.

[0096] Intelligent detection and diagnosis module: Constructs an intelligent ring main unit monitoring and diagnosis model, inputs the status data of the power distribution branch of the ring main unit as input parameters into the intelligent ring main unit monitoring and diagnosis model, predicts and analyzes the voltage-current curves under normal and abnormal loads through deep neural networks, sets ring main unit maintenance information based on the changes in the voltage-current curves, simulates the operating status of the ring main unit based on intelligent drive, and generates a remote maintenance plan for the ring main unit.

[0097] Digital simulation control module: Simulates the execution of the remote maintenance scheme of the ring main unit through digital simulation control method, analyzes the operating status data of the ring main unit, extracts response delay, harmonic and failure rate data from the ring main unit operation log through convolutional neural network for the simulated load test, and verifies the insulation degradation cycle of the ring main unit based on electromagnetic countermeasure interference network to obtain ring main unit status early warning information;

[0098] Maintenance push early warning module: Based on the remote maintenance scheme of the ring main unit, multi-level risk early warning values ​​are set. When the current power distribution branch data of the ring main unit exceeds the risk early warning value, the maintenance cycle and inspection strategy are dynamically optimized, and remote maintenance instructions and ring main unit status early warning information are pushed to the maintenance personnel through the cloud intelligent reasoning monitoring platform.

[0099] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for remote early warning and maintenance of the state of a ring main unit integrated with AI diagnosis, characterized in that, The method comprises the following steps: S1: Collecting power distribution branch load data, which is based on analyzing transient pulse current signals in ring main unit partial discharge by pulse current method, and simulating load test of ring main unit according to the transient pulse current signals to generate ring main unit power distribution branch state data; S2: Constructing an intelligent ring main unit monitoring and diagnosis model, inputting the ring main unit power distribution branch state data as input parameters into the intelligent ring main unit monitoring and diagnosis model, analyzing voltage-current curves under normal and abnormal loads, setting ring main unit maintenance information in combination with the voltage-current curve changes, evaluating ring main unit load operation state, and generating a ring main unit remote maintenance scheme; S3: Simulating execution of the ring main unit remote maintenance scheme by digital simulation control method, analyzing operation state data of the ring main unit, identifying response delay, harmonic and failure rate data in the ring main unit operation log for the simulated load test by feature vector extraction method, and verifying ring main unit insulation degradation period based on electromagnetic interference simulation method to obtain ring main unit state early warning information; S4: Setting multi-level risk early warning values based on the ring main unit remote maintenance scheme, dynamically optimizing maintenance period and inspection strategy when detecting that the current ring main unit power distribution branch data exceeds the risk early warning value, and pushing ring main unit remote maintenance instructions and ring main unit state early warning information to maintenance personnel through a cloud monitoring platform; The temperature-current dual-mode grid is based on grid node division according to ring main unit load travel, performs dual-mode pairing of the temperature data and the current data, records corresponding current characteristic parameters and temperature characteristic parameters on each node, interpolates and encrypts the grid according to node characteristic parameters, and simulates ring main unit power distribution load; The construction method of the intelligent ring main unit monitoring and diagnosis model is as follows: According to the simulated load test, the current characteristics in the ring main unit operation process are extracted, unstructured power distribution operation state data is converted into a set of extracted response delay and fault parameters, the response delay in the ring main unit operation log is associated with the ring main unit power distribution branch state, and a ring main unit remote maintenance feature matrix is generated; Based on the ring main unit remote maintenance feature matrix, the set of response delay and fault parameters is received, ring main unit maintenance monitoring features are processed through feature vector extraction, and power distribution maintenance weights are distributed according to the ring main unit maintenance monitoring features; According to the power distribution maintenance weights, ring main unit operation execution data and feedback results are analyzed, parameter weights of the intelligent ring main unit monitoring and diagnosis model for monitoring ring main unit power distribution branch state data are dynamically corrected through gradient descent under the constraint of the temperature-current dual-mode grid, and the intelligent ring main unit monitoring and diagnosis model is constructed.

2. The method of claim 1, wherein, The collection method of the power distribution branch load data is as follows: analyzing operation parameters in the ring main unit operation process, the operation parameters including current data, temperature data and operation log data, performing time sequence modal fusion on the operation parameters, and obtaining power distribution branch load data in combination with switching instructions and feedback signals in the operation log data.

3. The method of claim 1, wherein, The method for generating ring main unit power distribution branch state data is as follows: The ring main unit is segmented and loaded based on the temperature-current dual-mode grid, current response curves are collected at different load levels, the start time of the actual voltage and current curve response when the ring main unit switching instruction is issued is recorded, and an undisturbed response interval in the power supply switching operation process of the ring main unit distribution branch is obtained; According to the undisturbed response interval, the response delay parameter is stored in a structured manner, and a ring main unit power distribution dataset is generated in combination with the current response curve in the power supply load application process. The ring main unit health degree dataset is used as the basic input parameter of the intelligent ring main unit monitoring and diagnosis model, and the current response curve is reconstructed by polynomial fitting to generate ring main unit distribution branch state data.

4. The method of claim 1, wherein, The simulation load test simulates the ring main unit micro-load execution process by synchronously sampling the undisturbed ring main unit switching response, and verifies the current curve scheme according to the test results.

5. The method of claim 1, wherein, The verification method of the ring main unit insulation degradation period is: obtaining the failure rate data of the insulation part in the operation process of the ring main unit, implementing accelerated aging test on the insulation part of the ring main unit by electromagnetic interference simulation method, and determining the degradation rate index based on the change of characteristic parameters before and after acceleration aging. Curve fitting is performed on the degradation rate index and the operation data of the ring main unit remote maintenance scheme to calculate the ring main unit insulation degradation period.

6. The method of claim 2, wherein, The ring main unit state early warning information is constructed by a recurrent neural network to generate a time series feature matrix, and the health degree score of each distribution branch is calculated based on the time series feature matrix. When the health degree score is lower than the risk warning value, the intelligent ring main unit monitoring and diagnosis model outputs the ring main unit state early warning information of the corresponding risk level.

7. The method of claim 1, wherein, The pushing process of the ring main unit remote maintenance instruction is: based on the multi-flow communication transmission protocol, the maintenance instruction generated in the cloud is issued to the maintenance terminal platform, and the edge computing node is used to cache and verify the on-site execution data of the ring main unit. The maintenance terminal platform automatically generates an execution confirmation signal after receiving the maintenance instruction, and verifies the maintenance instruction combined with the return information of the execution confirmation signal.

8. The method of claim 1, wherein, The voltage-current curve is based on the ring main unit distribution branch, and the peak value, phase difference and response delay parameter of the curve are extracted in the process of load change and switching instruction triggering to identify the abnormal fluctuation characteristics of the ring main unit under different load conditions.

9. The method of claim 7, wherein, The process simulated by the digital simulation control method is: mapping the control instruction in the remote maintenance scheme to the virtual space for load switching and insulation state simulation modeling, and generating ring main unit simulation maintenance feedback results combined with the difference between the simulation modeling results and the abnormal fluctuation characteristics of the voltage-current response curve measured by the physical object.

10. An integrated AI diagnosis ring main unit state remote early warning maintenance system, characterized in that, It comprises: A power distribution data acquisition module: collecting power distribution branch load data, the power distribution branch load data is based on the analysis of transient pulse current signals during ring main unit partial discharge based on the pulse current method, and the ring main unit distribution branch state data is generated by simulating the load test of the ring main unit based on the temperature-current dual-mode grid constructed according to the transient pulse current signals; The intelligent detection and diagnosis module: an intelligent ring main unit monitoring and diagnosis model is constructed, the ring main unit power distribution branch state data is input into the intelligent ring main unit monitoring and diagnosis model as an input parameter, the voltage-current curve under normal and abnormal loads is analyzed, the ring main unit maintenance information is set in combination with the voltage-current curve change, the ring main unit load operation state is evaluated, and a ring main unit remote maintenance scheme is generated; The digital simulation control module: the ring main unit remote maintenance scheme is simulated and executed through a digital simulation control method, the operation state data of the ring main unit is analyzed, the response delay, harmonic and failure rate data in the ring main unit operation log are identified through a feature vector extraction method for the simulated load test, and the ring main unit insulation degradation period is verified based on an electromagnetic interference simulation method, and ring main unit state early warning information is obtained; The maintenance pushing early warning module: based on the ring main unit remote maintenance scheme, a multi-level risk early warning value is set, when it is detected that the current ring main unit power distribution branch data exceeds the risk early warning value, the maintenance cycle and inspection strategy are dynamically optimized, and the ring main unit remote maintenance instruction and ring main unit state early warning information are pushed to the maintenance personnel through a cloud monitoring platform; The temperature-current dual-mode grid is based on ring main unit load travel to divide grid nodes, the temperature data and current data are dual-mode paired, and the corresponding current characteristic parameters and temperature characteristic parameters are recorded on each node, the grid is interpolated according to the node characteristic parameters, and the ring main unit power distribution load is simulated; The construction method of the intelligent ring main unit monitoring and diagnosis model is: According to the simulated load test, the current characteristics in the ring main unit operation process are extracted, the unstructured power distribution operation state data is converted into a response delay and fault parameter set, the response delay in the ring main unit operation log is associated with the ring main unit power distribution branch state, and a ring main unit remote maintenance feature matrix is generated; Based on the ring main unit remote maintenance feature matrix, the response delay and fault parameter set are received, the ring main unit maintenance monitoring features are processed through feature vector extraction, and the power distribution maintenance weight is distributed according to the ring main unit maintenance monitoring features; According to the power distribution maintenance weight, the ring main unit operation execution data and feedback results are analyzed, and the parameter weight of the intelligent ring main unit monitoring and diagnosis model for monitoring the ring main unit power distribution branch state data is dynamically corrected through gradient descent under the constraint of the temperature-current dual-mode grid, and the intelligent ring main unit monitoring and diagnosis model is constructed.

Images