Substation switch cabinet state sensing and early warning system and method
The substation switchgear status perception and early warning system, which is based on cloud-edge collaboration, utilizes multi-source sensing terminals and edge computing units for real-time analysis. Combined with cloud-based deep learning models, it solves the problems of blind spots and information lag in substation switchgear monitoring, and achieves efficient fault early warning and diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-05
- Publication Date
- 2026-03-13
AI Technical Summary
Existing substation switchgear monitoring systems suffer from large monitoring blind spots, information lag, inability to detect potential faults early, and lack of edge intelligent processing capabilities, making it difficult to meet the needs of "few-personnel" or "unmanned" scenarios.
The substation switchgear status perception and early warning system adopts cloud-edge collaboration. It collects data through multi-source sensing terminals, performs preliminary analysis and generates alarm signals through edge computing units, and combines cloud-based deep learning models for fault diagnosis to achieve edge intelligent analysis and timely early warning.
It enables 24/7 online monitoring and timely early warning of switchgear status, reduces the burden on cloud servers, improves diagnostic accuracy and fault identification speed, and is compatible with different models of switchgear.
Smart Images
Figure CN121663781A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of power technology. Specifically, this invention relates to a substation switchgear status sensing and early warning system and method. Background Technology
[0002] With the continuous improvement of power system capabilities, the real-time perception and early warning capabilities of the internal switchgear operating status of substations, as key nodes in the power grid, are becoming increasingly important. Currently, most switchgear operation relies on periodic inspections and local monitoring, which suffers from problems such as large monitoring blind spots, information lag, and inability to detect potential faults early. In addition, traditional systems mostly adopt a centralized structure and lack edge intelligent processing capabilities, making it difficult to meet the needs of new substations in scenarios requiring "fewer staff" or even "unmanned" operation.
[0003] The existing technology has the following drawbacks: 1. Existing switchgear monitoring systems mostly use single-parameter threshold alarms, such as temperature exceeding limits, which cannot identify more complex faults.
[0004] 2. Centralized cloud processing has high latency and cannot meet the requirements for millisecond-level fault response.
[0005] 3. Existing research relies solely on cloud-based deep learning models, lacking preprocessing capabilities at the edge.
[0006] Therefore, this invention proposes a substation switchgear status sensing and early warning system and method. Summary of the Invention
[0007] This invention aims to overcome the shortcomings of existing technologies and proposes a substation switchgear status perception and early warning system and method to achieve the following objectives: intelligent perception and early warning of substation switchgear status through cloud-edge collaboration.
[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a substation switchgear status sensing and early warning system, the system comprising a cloud server, an edge computing unit, an alarm unit, and a multi-source sensing terminal installed in the substation switchgear, the multi-source sensing terminal comprising a temperature sensor, a partial discharge sensor, a current transformer, and a voltage transformer, all of which are connected to the edge computing unit via a multi-sensor interface board; the edge computing unit is also connected to the cloud server and the alarm unit. Preferably, the edge computing unit is provided with an electromagnetic shielding housing.
[0009] This application also proposes a substation switchgear status sensing and early warning method, using the aforementioned substation switchgear status sensing and early warning system, the method comprising: The multi-source sensing terminal collects multi-source data from the substation switchgear and sends it to the edge computing unit. The multi-source data includes temperature data, electrical data, and partial discharge data. The edge computing unit receives the multi-source data and outputs a switchgear health score based on a pre-established diagnostic model; the edge computing unit makes alarm decisions based on the health score. When an alarm is triggered, the edge computing unit generates an alarm signal based on the health score and sends it to the alarm unit to trigger the alarm; at the same time, it uploads multi-source data, including the alarm time, to the cloud server within a preset continuous time period. Based on multi-source data from preset time intervals before and after the alarm time, the cloud server calls a deep learning model to perform fault diagnosis and adjusts the parameters of the diagnostic model in conjunction with a historical fault case library.
[0010] Preferably, the diagnostic model is represented as follows: HS = (W1× S1) + (W2× S2) + (W3× S3); Wherein, HS represents the switchgear health score, S1, S2, and S3 respectively represent the temperature, electrical, and partial discharge sub-scores obtained based on temperature data, electrical data, and partial discharge data, with a score range of 0-100 points; W1, W2, and W3 are the preset weights of the temperature, electrical, and partial discharge sub-scores and satisfy W1+W2+W3=1.
[0011] Preferably, the temperature sub-score S1 is represented as follows: S1=100-[C1×|Tmax-Tmaxn|+C2×|ΔT_rate-ΔT_raten|+C3×|ΔT_phase-ΔT_phasen|]; Wherein, Tmax represents the real-time maximum temperature of the switchgear, and Tmaxn represents the preset maximum temperature threshold; ΔT_rate represents the real-time maximum temperature rise rate of the switchgear, and ΔT_raten represents the preset maximum temperature rise rate threshold; ΔT_phase represents the real-time maximum temperature difference between the three phases in the cabinet, and ΔT_phasen represents the preset maximum temperature difference threshold; C1, C2, and C3 respectively represent the preset penalty coefficients for the maximum temperature of the switchgear, the maximum temperature rise rate of the switchgear, and the maximum temperature difference between the three phases in the cabinet.
[0012] Preferably, the electrical sub-score S2 is represented as follows: S2 = 100 - [K1×|I_eff - I_effn|+K2×|THD - THDn|+K3×|H - Hn|]; Where I_eff represents the real-time effective value of current, and I_effn represents the preset threshold value of effective current; THD represents the real-time total harmonic distortion rate, and THDn represents the preset threshold value of total harmonic distortion rate; H represents the real-time odd harmonic content, and Hn represents the preset threshold value of odd harmonic content; K1, K2, and K3 respectively represent the preset penalty coefficients for effective current value, total harmonic distortion rate, and odd harmonic content.
[0013] Preferably, the partial discharge sub-score S3 is represented as follows: S3=100-[P1×|N-Nn|+P2×|Q-Qn|+P3×|B-Bn|]; Where N represents the number of discharge pulses per power frequency cycle in real time, and Nn represents the preset threshold for the number of discharge pulses per power frequency cycle; Q represents the real-time average discharge quantity, and Qn represents the preset threshold for the average discharge quantity; B represents the skewness / kurtosis of the discharge phase distribution, and Bn represents the preset threshold for skewness / kurtosis; P1, P2, and P3 respectively represent the preset penalty coefficients for the number of discharge pulses per power frequency cycle, the average discharge quantity, and the skewness / kurtosis of the discharge phase distribution.
[0014] Preferably, the cloud server adjusts the parameters of the diagnostic model by combining historical fault case databases, including: Based on the historical fault case database, if it is detected that the frequency of the same fault occurring in a certain model of switchgear exceeds the preset frequency within a preset historical period, the cloud server will adjust the diagnostic model parameters of all switchgear of the corresponding model according to the corresponding fault type to improve the sensitivity of the fault alarm. The parameter adjustment of the diagnostic model includes: adjusting the thresholds or penalty coefficients of various parameters in the temperature, electrical, and partial discharge sub-scoring formulas, and adjusting the weighting coefficients in the health scoring formula.
[0015] Preferably, based on the health score, the alarm decision made by the edge computing unit adopts a hierarchical alarm strategy, including: No alarm will be triggered when the first scoring threshold is less than or equal to the health score and the score is less than 100. When the second scoring threshold is less than the health score and less than the first scoring threshold, a primary alarm signal is generated. When 0 < health score ≤ second score threshold, an advanced alarm signal is generated.
[0016] The technical effects of this invention are as follows: This invention achieves 24 / 7 online monitoring of switchgear operation status through multi-source sensing terminals. Simultaneously, the system possesses edge intelligent analysis capabilities, enabling early warning at the edge based on data collected by the multi-source sensing terminals. Fault identification is then achieved by combining this with a cloud server, reducing the burden on the cloud server compared to traditional solutions that rely solely on it. The system structure is simple, installation is flexible, and it is compatible with different switchgear models.
[0017] The edge computing unit of this invention assesses the status of the switch cabinet by calculating a health score through a diagnostic model. The diagnostic model combines multi-source data and is also optimized based on feedback from the cloud server, which improves the accuracy of diagnosis and thus enables timely and effective early warning. Attached Figure Description
[0018] Figure 1 This is a structural block diagram of a substation switchgear status sensing and early warning system provided in an embodiment of the present invention. Detailed Implementation
[0019] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. This is to help those skilled in the art to have a more complete, accurate, and in-depth understanding of the inventive concept and technical solutions of the present invention, and to facilitate its implementation. It should be noted that the terms "first," "second," etc., used in this application are only for the convenience of describing the technical solutions and to distinguish components; the corresponding component configurations may be the same or different, and are not intended to limit the scope of this application. To make the technical solutions of the present invention clearer, the present invention will be explained and illustrated through the following embodiments.
[0020] This embodiment provides a substation switchgear status sensing and early warning system, such as Figure 1 As shown, the system includes a cloud server, an edge computing unit, an alarm unit, and a multi-source sensing terminal installed in the substation switchgear. The multi-source sensing terminal includes temperature sensors, partial discharge sensors, current transformers, voltage transformers, etc., all of which are connected to the edge computing unit through a multi-sensor interface board. The edge computing unit is also connected to the cloud server and the alarm unit.
[0021] A temperature sensor is installed inside the switchgear to collect temperature data. In this embodiment, an infrared camera with a calcium fluoride lens is used to achieve temperature measurement without blind spots inside the cabinet. Current transformers and voltage transformers are used to collect real-time electrical data of the switchgear. A partial discharge sensor is used to collect partial discharge characteristic data of the switchgear; typically, ultra-high frequency sensors or similar sensors are used to capture signals generated by partial discharge. In specific implementations, the selection of each sensor in the multi-source sensing terminal can be flexibly configured according to the actual situation.
[0022] Multi-source data collected by temperature sensors, partial discharge sensors, current transformers, and voltage transformers are aggregated to the edge computing unit via a multi-sensor interface board. The edge computing unit then performs preliminary early warning analysis based on the multi-source data and sends the data to a cloud server for further fault identification. The alarm unit includes devices such as a display screen, buzzer, and alarm lights, all connected to the edge computing unit and used to trigger alarms based on alarm signals from the edge computing unit.
[0023] Edge computing units establish connections with cloud servers via wireless communication technologies (such as LoRa wireless communication) to achieve two-way data communication. Edge computing units are typically deployed inside substations. To avoid electromagnetic interference from electrical equipment within the substation, edge computing units are equipped with electromagnetic shielding housings to ensure the stability and reliability of the equipment.
[0024] Correspondingly, this embodiment provides a substation switchgear status sensing and early warning method, using the aforementioned substation switchgear status sensing and early warning system, the method includes: The multi-source sensing terminal collects multi-source data from the substation switchgear and sends it to the edge computing unit. The multi-source data includes temperature data, electrical data, and partial discharge data. The edge computing unit receives the multi-source data and outputs a switchgear health score based on a pre-established diagnostic model; the edge computing unit makes alarm decisions based on the health score. When an alarm is triggered, the edge computing unit generates an alarm signal based on the health score and sends it to the alarm unit to trigger the alarm; at the same time, it uploads multi-source data, including the alarm time, to the cloud server within a preset continuous time period. Based on multi-source data from preset time intervals before and after the alarm time, the cloud server calls a deep learning model to perform fault diagnosis and adjusts the parameters of the diagnostic model in conjunction with a historical fault case library.
[0025] Specifically, the multi-source sensing terminal collects multi-source data from the substation switchgear and sends it to the edge computing unit. This multi-source data includes temperature data, electrical data, and partial discharge data. This data is first preprocessed and feature extracted by the edge computing unit to obtain temperature features, electrical features, and partial discharge features. The preprocessing includes data cleaning, noise reduction, and filtering to improve data reliability. Then, feature extraction is performed to obtain temperature features, electrical features, and partial discharge features. Specifically: Temperature characteristics include maximum temperature value, maximum temperature rise rate (°C / s), and maximum temperature difference between the three phases inside the cabinet; Electrical characteristics: RMS current, total harmonic distortion (THD), odd harmonic content (including 3rd, 5th, 7th, etc.). Partial discharge characteristics: number of discharge pulses per power frequency cycle, average discharge quantity, and skewness / kurtosis of discharge phase distribution (PRPD).
[0026] Subsequently, temperature characteristics, electrical characteristics, and partial discharge characteristics are input into a preset diagnostic model, which then outputs a health score for the switchgear. The diagnostic model in this embodiment is expressed by the following formula: HS = (W1× S1) + (W2× S2) + (W3× S3); Among them, HS represents the switchgear health score, S1, S2, and S3 respectively represent the temperature, electrical, and partial discharge sub-scores obtained based on temperature data, electrical data, and partial discharge data, with a score range of 0-100 points; W1, W2, and W3 are the preset weights of the temperature, electrical, and partial discharge sub-scores, and satisfy W1+W2+W3=1. In specific implementation, the initial value of this weight can be set according to expert experience and flexibly adjusted according to the actual working conditions in subsequent work.
[0027] The temperature sub-score S1 is expressed as follows: S1=100-[C1×|Tmax-Tmaxn|+C2×|ΔT_rate-ΔT_raten|+C3×|ΔT_phase-ΔT_phasen|]; Wherein, Tmax represents the real-time maximum temperature of the switchgear, and Tmaxn represents the preset maximum temperature threshold; ΔT_rate represents the real-time maximum temperature rise rate of the switchgear, and ΔT_raten represents the preset maximum temperature rise rate threshold; ΔT_phase represents the real-time maximum temperature difference between the three phases in the cabinet, and ΔT_phasen represents the preset maximum temperature difference threshold; C1, C2, and C3 respectively represent the preset penalty coefficients for the maximum temperature of the switchgear, the maximum temperature rise rate of the switchgear, and the maximum temperature difference between the three phases in the cabinet.
[0028] The electrical electronics score S2 is expressed as follows: S2 = 100 - [K1×|I_eff - I_effn|+K2×|THD - THDn|+K3×|H - Hn|]; Where I_eff represents the real-time effective value of current, and I_effn represents the preset threshold value of effective current; THD represents the real-time total harmonic distortion rate, and THDn represents the preset threshold value of total harmonic distortion rate; H represents the real-time odd harmonic content, and Hn represents the preset threshold value of odd harmonic content; K1, K2, and K3 respectively represent the preset penalty coefficients for effective current value, total harmonic distortion rate, and odd harmonic content.
[0029] The score for the S3 move is expressed as follows: S3=100-[P1×|N-Nn|+P2×|Q-Qn|+P3×|B-Bn|]; Where N represents the number of discharge pulses per power frequency cycle in real time, and Nn represents the preset threshold for the number of discharge pulses per power frequency cycle; Q represents the real-time average discharge quantity, and Qn represents the preset threshold for the average discharge quantity; B represents the skewness / kurtosis of the discharge phase distribution, and Bn represents the preset threshold for skewness / kurtosis; P1, P2, and P3 respectively represent the preset penalty coefficients for the number of discharge pulses per power frequency cycle, the average discharge quantity, and the skewness / kurtosis of the discharge phase distribution.
[0030] The diagnostic model in this embodiment integrates multi-dimensional features, enabling a more comprehensive and sensitive reflection of the switchgear's health status. After obtaining the switchgear's health score, the edge computing unit in this embodiment makes alarm decisions accordingly and sets up a tiered alarm strategy, including: No alarm will be triggered when the first scoring threshold is less than or equal to the health score and the score is less than 100. When the second scoring threshold is less than the health score and less than the first scoring threshold, a primary alarm signal is generated. When 0 < health score ≤ second score threshold, an advanced alarm signal is generated.
[0031] For example, the first rating threshold is 80 and the second rating threshold is 60.
[0032] According to the tiered alarm strategy, the edge computing unit generates different alarm signals for the alarm unit. The alarm unit then uses different alarm methods depending on the alarm signal to alert the user to the severity of the alarm and improve the user experience. For example, for a basic alarm signal, the alarm unit alerts the user to monitoring via the display screen, and the alarm light illuminates yellow; for a high-level alarm signal, the alarm unit alerts the user to a fault via the display screen, and the alarm light illuminates red and the buzzer sounds.
[0033] Furthermore, in this embodiment, once an alarm signal is generated, the corresponding edge computing unit uploads multi-source data, including multi-source data from a preset continuous period of time (e.g., 5 minutes before and 5 minutes after the alarm time), to the cloud server. This data often contains fault characteristics, which helps improve the speed and accuracy of fault identification.
[0034] In this embodiment, the cloud server is pre-trained with a deep learning model for fault identification using a historical fault case database. This model typically employs an LSTM (Long Short-Term Memory) network. For time-varying temperature, partial discharge, and electrical data, the LSTM model can effectively capture long-term dependencies in the time series, improving the accuracy of fault identification. In practical applications, the fault identification results can be obtained by inputting real-time uploaded multi-source data into the deep learning model.
[0035] Based on the fault identification results, the cloud server can push them to the user terminal via wireless communication technology, facilitating subsequent maintenance and other operations by the user. Simultaneously, the cloud server also adjusts the parameters of the diagnostic model based on the fault identification results and a historical fault case database, including: Based on the historical fault case database, if it is detected that the frequency of the same fault occurring in a certain model of switchgear exceeds the preset frequency within a preset historical period, the cloud server will adjust the diagnostic model parameters of all switchgear of the corresponding model according to the corresponding fault type to improve the sensitivity of fault alarms, thereby enabling timely detection and early warning before the fault occurs. The parameter adjustments for the diagnostic model include: adjustments to various parameter thresholds or penalty coefficients in the temperature, electrical, and partial discharge sub-scoring formulas, and adjustments to the weighting coefficients in the health scoring formula.
[0036] For example, if a certain type of switchgear experiences "poor contact of phase A" (which causes the switchgear temperature to rise) more than 3 times in the past month, the maximum temperature difference threshold in the diagnostic model of all switchgear of the corresponding model will be adjusted from the original 15℃ to 12℃, thereby improving the sensitivity of the diagnostic model.
[0037] This invention enables 24 / 7 online monitoring of switchgear operating status (temperature rise, abnormal current, etc.). Simultaneously, the system possesses edge intelligence analysis capabilities, enabling early warning at the edge, followed by fault identification via a cloud server. This reduces the burden on the cloud server compared to traditional solutions that rely solely on it. The system structure is simple, installation is flexible, and it is compatible with different switchgear models.
[0038] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution; or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.
Claims
1. A substation switchgear status sensing and early warning system, characterized in that: The system includes a cloud server, an edge computing unit, an alarm unit, and a multi-source sensing terminal installed in the substation switchgear. The multi-source sensing terminal includes a temperature sensor, a partial discharge sensor, a current transformer, and a voltage transformer, all of which are connected to the edge computing unit through a multi-sensor interface board. The edge computing unit is also connected to the cloud server and the alarm unit.
2. The substation switchgear status sensing and early warning system according to claim 1, characterized in that: The edge computing unit is equipped with an electromagnetic shielding housing.
3. A method for sensing and early warning of substation switchgear status, using a substation switchgear status sensing and early warning system according to any one of claims 1-2, characterized in that: The method includes: The multi-source sensing terminal collects multi-source data from the substation switchgear and sends it to the edge computing unit. The multi-source data includes temperature data, electrical data, and partial discharge data. The edge computing unit receives the multi-source data and outputs a switchgear health score based on a pre-established diagnostic model; the edge computing unit makes alarm decisions based on the health score. When an alarm is triggered, the edge computing unit generates an alarm signal based on the health score and sends it to the alarm unit to trigger the alarm; at the same time, it uploads multi-source data, including the alarm time, to the cloud server within a preset continuous time period. Based on multi-source data from preset time intervals before and after the alarm time, the cloud server calls a deep learning model to perform fault diagnosis and adjusts the parameters of the diagnostic model in conjunction with a historical fault case library.
4. The method for substation switchgear status sensing and early warning according to claim 3, characterized in that: The diagnostic model is represented as follows: HS = (W1× S1) + (W2× S2) + (W3× S3); Wherein, HS represents the switchgear health score, S1, S2, and S3 respectively represent the temperature, electrical, and partial discharge sub-scores obtained based on temperature data, electrical data, and partial discharge data, with a score range of 0-100 points; W1, W2, and W3 are the preset weights of the temperature, electrical, and partial discharge sub-scores and satisfy W1+W2+W3=1.
5. The method for substation switchgear status sensing and early warning according to claim 4, characterized in that: The temperature sub-score S1 is represented as follows: S1=100-[C1×|Tmax-Tmaxn|+C2×|ΔT_rate-ΔT_raten|+C3×|ΔT_phase-ΔT_phasen|]; Wherein, Tmax represents the real-time maximum temperature of the switchgear, and Tmaxn represents the preset maximum temperature threshold; ΔT_rate represents the real-time maximum temperature rise rate of the switchgear, and ΔT_raten represents the preset maximum temperature rise rate threshold; ΔT_phase represents the real-time maximum temperature difference between the three phases in the cabinet, and ΔT_phasen represents the preset maximum temperature difference threshold; C1, C2, and C3 respectively represent the preset penalty coefficients for the maximum temperature of the switchgear, the maximum temperature rise rate of the switchgear, and the maximum temperature difference between the three phases in the cabinet.
6. The method for substation switchgear status sensing and early warning according to claim 4, characterized in that: The electrical electronics score S2 is expressed as follows: S2 = 100 - [K1×|I_eff - I_effn|+K2×|THD - THDn|+K3×|H - Hn|]; Where I_eff represents the real-time effective value of current, and I_effn represents the preset threshold value of effective current; THD represents the real-time total harmonic distortion rate, and THDn represents the preset threshold value of total harmonic distortion rate; H represents the real-time odd harmonic content, and Hn represents the preset threshold value of odd harmonic content; K1, K2, and K3 respectively represent the preset penalty coefficients for effective current value, total harmonic distortion rate, and odd harmonic content.
7. The method for substation switchgear status sensing and early warning according to claim 4, characterized in that: The partial discharge sub-score S3 is expressed as follows: S3=100-[P1×|N-Nn|+P2×|Q-Qn|+P3×|B-Bn|]; Where N represents the number of discharge pulses per power frequency cycle in real time, and Nn represents the preset threshold for the number of discharge pulses per power frequency cycle; Q represents the real-time average discharge quantity, and Qn represents the preset threshold for the average discharge quantity; B represents the skewness / kurtosis of the discharge phase distribution, and Bn represents the preset threshold for skewness / kurtosis; P1, P2, and P3 respectively represent the preset penalty coefficients for the number of discharge pulses per power frequency cycle, the average discharge quantity, and the skewness / kurtosis of the discharge phase distribution.
8. A method for substation switchgear status sensing and early warning according to any one of claims 3-7, characterized in that: The cloud server adjusts the parameters of the diagnostic model by combining historical fault case databases, including: Based on the historical fault case database, if it is detected that the frequency of the same fault in a certain model of switchgear exceeds the preset frequency within a preset historical period, the cloud server will adjust the diagnostic model parameters of all switchgear of the corresponding model according to the corresponding fault type to improve the sensitivity of fault alarm.
9. The method for substation switchgear status sensing and early warning according to claim 8, characterized in that: The parameter adjustments for the diagnostic model include: adjusting the thresholds or penalty coefficients of various parameters in the temperature, electrical, and partial discharge sub-scoring formulas, and adjusting the weighting coefficients in the health scoring formula.
10. A method for substation switchgear status sensing and early warning according to any one of claims 3-7, characterized in that: Based on the health score, the edge computing unit makes alarm decisions using a tiered alarm strategy, including: No alarm will be triggered when the first scoring threshold is less than or equal to the health score and the score is less than 100. When the second scoring threshold is less than the health score and less than the first scoring threshold, a primary alarm signal is generated. When 0 < health score ≤ second score threshold, an advanced alarm signal is generated.