A GIS circuit breaker closing state monitoring device

Abstract

The present application belongs to the technical field of electrical equipment state monitoring, and particularly relates to a GIS circuit breaker closing state monitoring device, which first synchronously collects vibration signals, closing and opening coil current signals and transmission main shaft angle signals in the closing process, then pre-processes the collected vibration signals, closing and opening coil current signals and transmission main shaft angle signals, extracts multi-dimensional features including vibration signal time-frequency domain features, current waveform key point parameters and angle-time curve motion features, constructs a comprehensive feature vector F, then calculates the health score of the current state and identifies the fault mode based on the constructed comprehensive feature vector F, finally displays the health score Hscore and the identified fault mode, and provides an effective tool for state maintenance and intelligent operation, which helps to reduce the unplanned shutdown risk and improve the power grid operation reliability.

Description

Technical Field

[0001] This invention belongs to the field of electrical equipment condition monitoring technology, specifically relating to a GIS circuit breaker closing status monitoring device. Background Technology

[0002] GIS circuit breakers are core equipment for ensuring the safe operation of the power grid. The reliability of their closing operation is directly related to the continuity of power supply. The closing process involves the coordinated action of multiple components such as electromagnets, springs, and connecting rods. Their mechanical condition deteriorates over time. Traditional GIS circuit breaker condition monitoring mainly relies on periodic maintenance and offline testing, which makes it difficult to reflect the changes in the equipment's condition in actual operation in real time. This results in problems such as maintenance lag and insufficient fault early warning capabilities. Moreover, most existing monitoring methods use single signals such as coil current or vibration for analysis, lacking the synchronous acquisition and fusion of multiple physical quantities. This leads to incomplete fault feature extraction, limited diagnostic accuracy, and the inability to automatically identify fault modes and output fault probability vectors P when the health score Hscore is lower than the preset threshold. This results in poor real-time performance and makes it difficult to meet the needs of rapid response operation and maintenance.

[0003] To address the aforementioned issues, this application presents a GIS circuit breaker closing status monitoring device. Summary of the Invention

[0004] To address the shortcomings of the prior art mentioned in the background section, this application proposes a GIS circuit breaker closing status monitoring device. First, it synchronously collects vibration signals, opening / closing coil current signals, and drive shaft angle signals during the closing process. Then, it preprocesses the collected vibration signals, opening / closing coil current signals, and drive shaft angle signals, extracting multi-dimensional features including time-frequency domain features of the vibration signals, key point parameters of the current waveform, and motion features of the angle-time curve, constructing a comprehensive feature vector F. Next, based on the constructed comprehensive feature vector F, it calculates the current state's health score and identifies fault modes. Finally, it displays the health score Hscore and the identified fault modes, and issues an early warning when the health score Hscore falls below a preset threshold, thereby solving the problems in the background section.

[0005] To achieve the above objectives, this application provides a GIS circuit breaker closing status monitoring device, including a sensor acquisition module, an edge intelligent processing module, an intelligent diagnostic module, and a human-machine interaction and data service module. The sensor acquisition module includes a vibration acceleration sensor, a high-frequency Hall current sensor, and a miniature angular displacement sensor, which are used to synchronously acquire vibration signals, opening and closing coil current signals, and transmission spindle angle signals during the closing process. The edge intelligent processing module has a built-in signal conditioning circuit, synchronous data acquisition card and embedded processor. It is used to preprocess the acquired vibration signal, opening and closing coil current signal and transmission spindle angle signal, and extract multi-dimensional features including vibration signal time-frequency domain features, current waveform key point parameters and angle-time curve motion features, and finally construct a comprehensive feature vector F. The intelligent diagnostic module is used to calculate the health score of the current state and identify fault modes based on the constructed comprehensive feature vector F, including the following steps: S1. Obtain the current comprehensive feature vector Fcurrent, and calculate the Mahalanobis distance DM between the current comprehensive feature vector Fcurrent and the mean vector μ of the pre-stored healthy feature vector set. The formula for calculating the Mahalanobis distance DM is: ; Where S is the covariance matrix, It is the inverse of the covariance matrix; S2. Map the calculated Mahalanobis distance DM to the health score Hscore. The mapping formula for the health score Hscore is: ; Wherein, γ is an adjustable scaling parameter used to control the sensitivity of the score as distance changes; S3. Determine the threshold for the health score Hscore. If the health score Hscore is higher than the preset threshold, the diagnosis ends. If the health score Hscore is lower than the preset threshold, the intelligent diagnosis module further inputs the comprehensive feature vector Fcurrent into the pre-trained fault mode recognition model to perform fault mode recognition and outputs the fault probability vector P.

[0006] In a preferred embodiment based on the above scheme, the vibration acceleration sensor is installed at a key position of the GIS circuit breaker operating mechanism, the high-frequency Hall current sensor is connected in series in the opening and closing coil circuit of the GIS circuit breaker, and the miniature angular displacement sensor is installed at the transmission link of the GIS circuit breaker.

[0007] Based on the above scheme, the preferred embodiment is the comprehensive feature vector F = [Fa, Fi, Fo], where Fa, Fi, and Fo are the vibration feature subset Fa, current feature subset Fi, and motion feature subset Fo extracted from the vibration signal, the opening and closing coil current signal, and the transmission main shaft angle signal, respectively.

[0008] Based on the above scheme, the preferred embodiment is the comprehensive feature vector F = [Fa, Fi, Fo], where Fa, Fi, and Fo are the vibration feature subset Fa, current feature subset Fi, and motion feature subset Fo extracted from the vibration signal, the opening and closing coil current signal, and the transmission main shaft angle signal, respectively.

[0009] In a preferred embodiment based on the above scheme, the vibration feature subset Fa includes the effective vibration value, vibration peak value, and vibration energy in a specific frequency band extracted from the closing event time window; The current feature subset Fi includes the peak value of the extracted closing coil current waveform, the time to reach the peak value, and the approximate slope of the current rise. The motion feature subset Fo includes the total closing time, maximum angular velocity, and overshoot at the end of the motion phase extracted from the angle-time curve.

[0010] Based on the above scheme, the preferred embodiment is the output fault probability vector P = [P1, P2, P3, P4], where P1, P2, P3, and P4 represent the probabilities of the current state being judged as normal, core jamming, spring fatigue, and connecting rod loosening, respectively. .

[0011] Based on the above scheme, the preferred embodiment of the fault mode recognition model is to identify faults by learning the unique patterns of different faults in the feature space. The fault recognition logic is as follows: a. If all features extracted from the vibration signal, the opening and closing coil current signal and the transmission main shaft angle signal fluctuate within the specified safety range and no abnormality occurs, it is judged as normal. The decrease in the health score Hscore is a normal phenomenon caused by the aging of the GIS circuit breaker. b. If the peak value of the closing coil current waveform increases abnormally and the total closing time is significantly prolonged, it is determined to be core jamming. c. If the maximum angular velocity decreases and the vibration energy distribution changes, it is determined to be spring fatigue. d. If the overshoot at the end of the motion is abnormal or the vibration energy in a specific frequency band increases dramatically, it is determined that the connecting rod is loose.

[0012] In a preferred embodiment based on the above scheme, the fault mode recognition model calculates the evidence strength of the current state belonging to multiple predefined fault modes through its internal discriminant function, and finally outputs the fault probability vector P through a normalization function, with the fault category corresponding to the highest probability being used as the primary fault warning.

[0013] Based on the above scheme, the preferred embodiment of the human-computer interaction and data service module includes a local display device, an early warning module, and a remote monitoring backend. The local display device is used to display the health score Hscore and the identified fault modes. The early warning module is used to issue an early warning when the health score Hscore is lower than a preset threshold. The remote monitoring backend is used for historical data query and export.

[0014] In a preferred embodiment based on the above scheme, the edge intelligent processing module is connected to the cloud server to upload the comprehensive feature vector F and the diagnosed fault probability vector P=[P1, P2, P3, P4], and to receive and update the fault mode recognition model parameters.

[0015] Compared with the prior art, the beneficial effects of the present invention are: This invention first comprehensively acquires vibration signals, opening and closing coil current signals, and drive shaft angle signals during the closing process through synchronous acquisition by a vibration acceleration sensor, a high-frequency Hall current sensor, and a miniature angular displacement sensor. Then, an edge intelligent processing module preprocesses these signals, extracting multi-dimensional features including time-frequency domain characteristics of the vibration signals, key parameters of the current waveform, and motion characteristics of the angle-time curve. Finally, a comprehensive feature vector F is constructed. Next, Mahalanobis distance is used to calculate a health score, enabling quantitative assessment and trend monitoring of the closing status. Combined with a pre-trained fault mode recognition model, fault classification and probability output are automatically performed when the health score falls below a threshold, improving the accuracy and interpretability of fault diagnosis. Finally, by combining local display, early warning prompts, and remote data services, this invention meets the needs of on-site personnel to monitor equipment status in real time while also supporting historical data traceability and remote expert diagnosis. It provides an effective tool for condition-based maintenance and intelligent operation and maintenance, helping to reduce the risk of unplanned outages and improve the reliability of power grid operation. Attached Figure Description

[0016] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a system flowchart of a GIS circuit breaker closing status monitoring device according to the present invention; Figure 2 This is a flowchart of the evaluation and diagnosis process of the intelligent diagnosis module in this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] Example: To address the technical problems raised in the background art, this application provides a preferred embodiment: such as Figure 1-2As shown, a GIS circuit breaker closing status monitoring device includes a sensor acquisition module, an edge intelligent processing module, an intelligent diagnostic module, and a human-machine interaction and data service module. The sensor acquisition module includes a vibration acceleration sensor, a high-frequency Hall current sensor, and a miniature angular displacement sensor, which are used to synchronously acquire vibration signals, opening and closing coil current signals, and transmission spindle angle signals during the closing process. The edge intelligent processing module has a built-in signal conditioning circuit, synchronous data acquisition card and embedded processor. It is used to preprocess the acquired vibration signal, opening and closing coil current signal and transmission spindle angle signal, and extract multi-dimensional features including vibration signal time-frequency domain features, current waveform key point parameters and angle-time curve motion features, and finally construct a comprehensive feature vector F. The intelligent diagnostic module is used to calculate the health score of the current state and identify fault modes based on the constructed comprehensive feature vector F, including the following steps: S1. Obtain the current comprehensive feature vector Fcurrent, and calculate the Mahalanobis distance DM between the current comprehensive feature vector Fcurrent and the mean vector μ of the pre-stored healthy feature vector set. The formula for calculating the Mahalanobis distance DM is: ; Where S is the covariance matrix, It is the inverse of the covariance matrix; S2. Map the calculated Mahalanobis distance DM to the health score Hscore. The mapping formula for the health score Hscore is: ; Wherein, γ is an adjustable scaling parameter used to control the sensitivity of the score as distance changes; S3. Determine the threshold for the health score Hscore. If the health score Hscore is higher than the preset threshold, the diagnosis ends. If the health score Hscore is lower than the preset threshold, the intelligent diagnosis module further inputs the comprehensive feature vector Fcurrent into the pre-trained fault mode recognition model to perform fault mode recognition and outputs the fault probability vector P.

[0019] It should be noted that: the comprehensive feature vector F is a general symbol or structural definition. In the patent document, it is used to conceptually define what the feature vector constructed by our system looks like. That is, it is a mathematical vector composed of three feature subsets of vibration, current and angle in a certain order. It describes the output format of the system. The current feature vector Fcurrent is the variable name used in a specific diagnostic operation, specifically referring to the specific feature vector instance extracted and constructed in this closing operation. Compared to simple Euclidean distance or fixed threshold methods, Mahalanobis distance has an overwhelming advantage in equipment health assessment. During the circuit breaker closing process, features are strongly correlated. For example, a longer closing time is usually accompanied by an increase in the peak coil current. If independent thresholds are set for these two features, inaccurate assessment boundaries such as rectangles will be formed. The Mahalanobis distance calculation formula can automatically capture and correct this correlation, solving this problem. In the direction of high feature correlation, the allowed joint fluctuation range is larger, and in the direction of weak correlation, it is more sensitive. It defines an elliptical health boundary that adapts to the natural distribution of data, which is far more reasonable than a rectangular boundary. Moreover, different features such as current amperes, time milliseconds, and vibration energy have huge differences in dimensions and numerical ranges. Mahalanobis distance can automatically complete dimensionless calculation because it contains the variance information of each feature. Features with large variance have large fluctuations and their weights will be relatively reduced, while features with small variance have small fluctuations and are more precise, and their anomalies will be amplified. This is in line with engineering intuition. The Mahalanobis distance calculation formula can successfully transform complex, correlated, multidimensional information into a scientific and interpretable single health indicator. The advantages of the H-score mapping formula are: When fully within the health center, the health score Hscore equals 100. As the value of Hscore increases, the health score decreases towards zero. This formula is derived from... By imposing constraints, when the distance is too large in actual calculations, the score can be stabilized at a lower limit, thus avoiding extreme values.

[0020] The advantages of the above scheme are as follows: First, by synchronously acquiring vibration signals, opening and closing coil current signals, and drive shaft angle signals during the closing process through vibration acceleration sensors, high-frequency Hall current sensors, and miniature angular displacement sensors, the scheme comprehensively obtains vibration signals, opening and closing coil current signals, and drive shaft angle signals during the closing process. Then, through an edge intelligent processing module, the acquired vibration signals, opening and closing coil current signals, and drive shaft angle signals are preprocessed, and multi-dimensional features, including time-frequency domain features of vibration signals, key point parameters of current waveforms, and motion features of angle-time curves, are extracted. Finally, a comprehensive feature vector F is constructed. Next, Mahalanobis distance is used to calculate the health score, realizing quantitative assessment and trend monitoring of the closing status. Combined with a pre-trained fault mode recognition model, when the health score is below a threshold, fault classification and probability output are automatically performed, improving the accuracy and interpretability of fault diagnosis. Finally, by combining local display, early warning prompts, and remote data services, the scheme not only meets the needs of on-site personnel to monitor equipment status in real time, but also supports historical data traceability and remote expert diagnosis, providing an effective tool for condition-based maintenance and intelligent operation and maintenance, helping to reduce the risk of unplanned outages and improve the reliability of power grid operation.

[0021] Furthermore: In an optional embodiment, a vibration acceleration sensor is installed at a key position on the operating mechanism of the GIS circuit breaker, a high-frequency Hall current sensor is connected in series in the opening and closing coil circuit of the GIS circuit breaker, and a miniature angular displacement sensor is installed at the transmission link of the GIS circuit breaker.

[0022] In an optional embodiment, the comprehensive feature vector F = [Fa, Fi, Fo], where Fa, Fi, and Fo are the vibration feature subset Fa, current feature subset Fi, and motion feature subset Fo extracted from the vibration signal, the opening and closing coil current signal, and the transmission main shaft angle signal, respectively.

[0023] In an optional embodiment, the vibration feature subset Fa includes the effective vibration value, vibration peak value, and vibration energy in a specific frequency band extracted from the closing event time window; The current feature subset Fi includes the peak value of the extracted closing coil current waveform, the time to reach the peak value, and the approximate slope of the current rise. The motion feature subset Fo includes the total closing time, maximum angular velocity, and overshoot at the end of the motion, extracted from the angle-time curve.

[0024] Furthermore: In an optional embodiment, the output fault probability vector P = [P1, P2, P3, P4] is used, where P1, P2, P3, and P4 represent the probabilities of the current state being judged as normal, core jamming, spring fatigue, and connecting rod loosening, respectively. .

[0025] It should be noted that the fault mode is not limited to the four modes mentioned above, and can be set according to the specific actual usage.

[0026] In an optional embodiment, the fault mode recognition model identifies faults by learning the unique patterns of different faults in the feature space. The fault recognition logic is as follows: a. If all features extracted from the vibration signal, the opening and closing coil current signal and the transmission main shaft angle signal fluctuate within the specified safety range and no abnormality occurs, it is judged as normal. The decrease in the health score Hscore is a normal phenomenon caused by the aging of the GIS circuit breaker. b. If the peak value of the closing coil current waveform increases abnormally and the total closing time is significantly prolonged, it is determined to be core jamming. c. If the maximum angular velocity decreases and the vibration energy distribution changes, it is determined to be spring fatigue. d. If the overshoot at the end of the motion is abnormal or the vibration energy in a specific frequency band increases dramatically, it is determined that the connecting rod is loose.

[0027] It should be noted that the fault identification logic is not limited to the above-mentioned situations and can be set according to the specific actual use case.

[0028] In an optional embodiment, the fault mode recognition model calculates the evidence strength of the current state belonging to multiple predefined fault modes through its internal discriminant function, and finally outputs a fault probability vector P through a normalization function, with the fault category corresponding to the highest probability being used as the primary fault warning.

[0029] In an optional embodiment, the human-computer interaction and data service module includes a local display device, an early warning module, and a remote monitoring backend. The local display device is used to display the health score Hscore and the identified fault modes. The early warning module is used to issue an early warning when the health score Hscore is lower than a preset threshold. The remote monitoring backend is used for historical data querying and exporting.

[0030] In an optional embodiment, the edge intelligence processing module is communicatively connected to a cloud server to upload the comprehensive feature vector F and the diagnosed fault probability vector P=[P1, P2, P3, P4], and to receive fault mode recognition model parameters for updates.

[0031] In practice: First, during each closing operation, vibration signals, opening and closing coil current signals, and transmission spindle angle signals are simultaneously triggered and collected; Then, the collected vibration signal, opening and closing coil current signal, and transmission spindle angle signal are filtered, aligned, and feature extracted to construct the comprehensive feature vector F of this operation; Next, obtain the current comprehensive feature vector Fcurrent, and calculate the Mahalanobis distance DM between the current comprehensive feature vector Fcurrent and the mean vector μ of the pre-stored healthy feature vector set; Next, the calculated Mahalanobis distance DM is mapped to the health score Hscore. Next, the health score Hscore is used to determine the threshold. If the health score Hscore is higher than the preset threshold, the diagnosis ends. If the health score Hscore is lower than the preset threshold, the intelligent diagnosis module further inputs the comprehensive feature vector Fcurrent into the pre-trained fault mode recognition model to perform fault mode recognition and outputs the fault probability vector P. Finally, the health score Hscore and the identified fault modes are displayed locally through the human-computer interaction and data service module, and an early warning is issued when the health score Hscore is lower than the preset threshold.

[0032] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A GIS circuit breaker closing status monitoring device, comprising a sensor acquisition module, an edge intelligent processing module, an intelligent diagnostic module, and a human-computer interaction and data service module, characterized in that, The sensing and acquisition module includes a vibration acceleration sensor, a high-frequency Hall current sensor, and a miniature angular displacement sensor, which are used to synchronously acquire vibration signals, opening and closing coil current signals, and transmission shaft angle signals during the closing process. The edge intelligent processing module has a built-in signal conditioning circuit, synchronous data acquisition card and embedded processor. It is used to preprocess the acquired vibration signal, opening and closing coil current signal and transmission spindle angle signal, and extract multi-dimensional features including vibration signal time-frequency domain features, current waveform key point parameters and angle-time curve motion features, and finally construct a comprehensive feature vector F. The intelligent diagnostic module is used to calculate the health score of the current state and identify fault modes based on the constructed comprehensive feature vector F, including the following steps: S1. Obtain the current comprehensive feature vector Fcurrent, and calculate the Mahalanobis distance DM between the current comprehensive feature vector Fcurrent and the mean vector μ of the pre-stored healthy feature vector set. The formula for calculating the Mahalanobis distance DM is: ； Where S is the covariance matrix, It is the inverse of the covariance matrix; S2. Map the calculated Mahalanobis distance DM to the health score Hscore. The mapping formula for the health score Hscore is: ； Wherein, γ is an adjustable scaling parameter used to control the sensitivity of the score as distance changes; S3. Determine the threshold for the health score Hscore. If the health score Hscore is higher than the preset threshold, the diagnosis ends. If the health score Hscore is lower than the preset threshold, the intelligent diagnosis module further inputs the comprehensive feature vector Fcurrent into the pre-trained fault mode recognition model to perform fault mode recognition and outputs the fault probability vector P.

2. The GIS circuit breaker closing status monitoring device according to claim 1, characterized in that: The vibration acceleration sensor is installed at a key position in the operating mechanism of the GIS circuit breaker, the high-frequency Hall current sensor is connected in series in the opening and closing coil circuit of the GIS circuit breaker, and the miniature angular displacement sensor is installed at the transmission link of the GIS circuit breaker.

3. The GIS circuit breaker closing status monitoring device according to claim 1, characterized in that: The comprehensive feature vector F = [Fa, Fi, Fo], where Fa, Fi, and Fo are the vibration feature subset Fa, current feature subset Fi, and motion feature subset Fo extracted from the vibration signal, the opening and closing coil current signal, and the transmission main shaft angle signal, respectively.

4. The GIS circuit breaker closing status monitoring device according to claim 3, characterized in that: The vibration feature subset Fa includes the effective vibration value, vibration peak value, and vibration energy in a specific frequency band extracted from the closing event time window; The current feature subset Fi includes the peak value of the extracted closing coil current waveform, the time to reach the peak value, and the approximate slope of the current rise. The motion feature subset Fo includes the total closing time, maximum angular velocity, and overshoot at the end of the motion phase extracted from the angle-time curve.

5. The GIS circuit breaker closing status monitoring device according to claim 4, characterized in that: The output fault probability vector P = [P1, P2, P3, P4], where P1, P2, P3, and P4 represent the probabilities of the current state being judged as normal, core jamming, spring fatigue, and connecting rod loosening, respectively. .

6. The GIS circuit breaker closing status monitoring device according to claim 5, characterized in that: The fault mode recognition model identifies faults by learning the unique patterns of different faults in the feature space. The fault recognition logic is as follows: a. If all features extracted from the vibration signal, the opening and closing coil current signal and the transmission main shaft angle signal fluctuate within the specified safety range and no abnormality occurs, it is judged as normal. The decrease in the health score Hscore is a normal phenomenon caused by the aging of the GIS circuit breaker. b. If the peak value of the closing coil current waveform increases abnormally and the total closing time is significantly prolonged, it is determined to be core jamming. c. If the maximum angular velocity decreases and the vibration energy distribution changes, it is determined to be spring fatigue. d. If the overshoot at the end of the motion is abnormal or the vibration energy in a specific frequency band increases dramatically, it is determined that the connecting rod is loose.

7. The GIS circuit breaker closing status monitoring device according to claim 6, characterized in that: The fault mode recognition model calculates the evidence strength of the current state belonging to multiple predefined fault modes through its internal discriminant function, and finally outputs the fault probability vector P through a normalization function, with the fault category corresponding to the highest probability being used as the primary fault warning.

8. The GIS circuit breaker closing status monitoring device according to claim 1, characterized in that: The human-computer interaction and data service module includes a local display device, an early warning module, and a remote monitoring backend. The local display device is used to display the health score Hscore and the identified fault modes. The early warning module is used to issue an early warning when the health score Hscore is lower than a preset threshold. The remote monitoring backend is used for historical data query and export.

9. A GIS circuit breaker closing status monitoring device according to claim 1, characterized in that: The edge intelligent processing module is connected to the cloud server to upload the comprehensive feature vector F and the diagnosed fault probability vector P=[P1, P2, P3, P4], and to receive and update the fault mode recognition model parameters.

Images