GIS knife switch fault diagnosis method and multi-monitoring point sound and vibration acquisition equipment

Abstract

The present invention proposes a fault diagnosis method for a GIS knife switch and a multi-monitoring point sound and vibration collection device, which relate to the technical field of GIS knife switch fault diagnosis. The method collects sound and vibration data of the GIS knife switch under normal working conditions to determine a threshold value, and adopts a multi-monitoring point collection method that combines idle uninterrupted monitoring with formal monitoring. In the idle uninterrupted monitoring, the sound and vibration amplitude frequency thresholds are used to judge whether the idle uninterrupted monitoring data is abnormal. This method only turns on the collection and detection functions of the sound and vibration data, reduces energy consumption, speeds up data processing, and discards sound and vibration data without abnormalities, saving memory space. When an abnormality occurs, the formal monitoring method is entered. This method includes functions such as data collection, data processing and data storage. After the formal sound and vibration information detection is completed, the threshold value judgment is performed on the data monitored by the remaining collection and monitoring points, thereby improving the reliability of the fault diagnosis result.

Description

Technical Field

[0001] The present invention relates to the technical field of GIS knife switch fault diagnosis, and more particularly to a GIS knife switch fault diagnosis method and multi-monitoring point sound and vibration collection equipment. Background Art

[0002] GIS equipment has the advantages of compact structure, high reliability, good seismic resistance, low noise, low maintenance and no influence from external environmental conditions. It has been increasingly widely used in power systems.

[0003] GIS switches are essential equipment in substations for changing system operation and isolating equipment for maintenance. Their operating status directly impacts the safety of switching operations and is crucial for ensuring the safe and stable operation of the power grid. Compared to traditional open-type equipment, they have a lower failure rate. However, once a failure occurs, locating and troubleshooting the fault point is relatively difficult, resulting in slow power restoration and high incident handling costs. Therefore, monitoring the mechanical status of the switch in the combined electrical system (GIS) during operation is crucial. Traditionally, mechanical status assessment relies on on-site observation by experts who listen to the operating sounds and make empirical judgments during operation. The drawback of this approach is that problems can only be detected during fixed maintenance periods, making it impossible to promptly address various switch failures. Furthermore, given the relatively low failure rate of GIS switches, frequent manual maintenance of these switches is not only time-consuming and labor-intensive, but also inevitably leads to misjudgments and missed detections.

[0004] The prior art discloses a sound-based online monitoring system and method for GIS circuit breaker faults. The sound-based online monitoring method for GIS circuit breaker faults includes the following steps: acquiring and storing multiple different types of fault action sound signals of the GIS circuit breaker; collecting the real-time action sound signal of the GIS circuit breaker; calculating the similarity between the real-time action sound signal and multiple different types of fault action sound signals. If the similarity between the real-time action sound signal and one type of fault action sound signal is greater than a first preset threshold, it is determined that the GIS circuit breaker has a fault. This can achieve online monitoring of the mechanical state of the GIS circuit breaker, facilitating the inspection and maintenance of the GIS circuit breaker. However, on the one hand, this solution calculates and compares all the action sound signal data collected in real time in real time, which has the disadvantages of large data storage space, high data storage costs, and high processor operating power consumption. On the other hand, all data in the entire process are judged by a single absolute standard (a first preset threshold), and the fault diagnosis conclusion is directly drawn based on the judgment result, which has a certain degree of misjudgment and low reliability of the fault diagnosis result. Summary of the Invention

[0005] In order to solve the problems in the existing GIS knife switch fault diagnosis method, such as large data storage space occupation, high processing energy consumption and low reliability of fault diagnosis results, the present invention proposes a GIS knife switch fault diagnosis method and multi-monitoring point sound and vibration collection equipment, which adopts a multi-monitoring point collection method that combines uninterrupted monitoring and collection during idle time with formal monitoring and collection, thereby improving the reliability of fault diagnosis results. In the process, unnecessary data is discarded in time, saving storage space occupied by data and reducing data processing energy consumption.

[0006] In order to achieve the above technical effects, the technical solutions of the present invention are as follows:

[0007] A method for diagnosing a fault of a GIS knife switch, the method comprising:

[0008] S1. Measure sound and vibration data multiple times on the GIS knife switch housing. Select the location with the most pronounced sound loudness and vibration amplitude as the acquisition area. Place an external sound card and vibration sensor to collect and analyze sound and vibration data under normal operating conditions of the GIS knife switch to determine the sound loudness frequency threshold and the vibration amplitude frequency threshold.

[0009] S2. Continuous monitoring during off-peak hours: Multiple monitoring points continuously collect sound and vibration data from GIS knife switches, and perform Fourier transform on the collected data to obtain the sound loudness frequency and vibration amplitude frequency;

[0010] S3. Determine whether the sound loudness frequency exceeds the sound loudness frequency threshold or whether the vibration amplitude frequency exceeds the sound loudness frequency threshold. If so, the corresponding acquisition monitoring point triggers an interrupt and records the time point of the interrupt triggering, and executes step S4; otherwise, the GIS knife switch is normal, the collected sound and vibration data are discarded, and return to S2;

[0011] S4. Start formal monitoring: Enter the sound and vibration information processing process of the monitoring point that has triggered the interruption of collection, and obtain the fault type of the GIS knife switch;

[0012] S5. The time point at which the interrupt is triggered and the number of the current acquisition monitoring point that triggers the interruption are recorded as log information, and the fault area is located and the location information is saved;

[0013] S7. Return to step S3 for the data collected from the remaining monitoring points.

[0014] In this technical solution, the sound and vibration data of the GIS knife switch under normal working conditions are first collected to determine the threshold value, and then the idle time uninterrupted monitoring and formal monitoring are combined with multi-monitoring point collection. In the idle time uninterrupted monitoring, the sound loudness frequency threshold and the vibration amplitude frequency threshold are used to judge whether the idle time uninterrupted monitoring data is abnormal. The idle time uninterrupted monitoring process only turns on the collection and detection functions of sound and vibration data, reduces energy consumption, speeds up data processing, and immediately discards sound and vibration data without abnormalities to save memory space; in the event of an abnormality, the formal monitoring mode is entered. The formal monitoring process includes functions such as data collection, data processing and data storage. After the formal sound and vibration information detection is completed, the threshold judgment is performed on the data monitored by the remaining collection and monitoring points, which improves the reliability of the fault diagnosis results.

[0015] Preferably, in step S1, the amplitude and frequency of the sound and vibration data of the GIS knife switch under normal working conditions that have been collected are analyzed respectively, and the first five frequency components and their amplitudes of the sound and vibration data under normal working conditions of the GIS knife switch are recorded respectively, and then the amplitudes of the first five frequency components are averaged and used as the sound loudness frequency threshold and the vibration amplitude frequency threshold respectively.

[0016] Here, the characteristic that the main working frequency of GIS operating equipment remains unchanged is utilized to determine the sound loudness frequency threshold and vibration amplitude frequency threshold.

[0017] Preferably, a plurality of sound and vibration collection monitoring points are provided on the GIS knife switch, and each sound and vibration collection monitoring point is provided with a monitoring device, which includes an external sound card and an acceleration sensor. The sound data is collected by the external sound card, and the vibration data is collected by the acceleration sensor. The external sound card and the acceleration sensor are placed together as a monitoring device for a sound and vibration monitoring point.

[0018] Here, the reliability of fault diagnosis is higher by adopting the method of multiple sound and vibration collection monitoring points.

[0019] Preferably, each external sound card collects the sound of the GIS knife switch during operation at a frequency of H1 and a fixed duration of T seconds to the main controller and saves it as a WAV file. The main controller then converts the WAV file into a PCM file. The PCM file is then transformed directly into a frequency domain graph of the audio segment after fast Fourier transformation. The frequency domain graph is used to analyze the loudness and frequency of the sound. Each acceleration sensor collects the vibration acceleration information of the GIS knife switch during operation at a frequency of H2 and a fixed duration of T seconds to the main controller. In the main controller, the acceleration data is processed into vibration data. Then, after fast Fourier transformation, a frequency domain graph of the vibration data is directly generated. The frequency domain graph is used to analyze the amplitude and frequency of the vibration information.

[0020] Preferably, except for the monitoring point that triggers the interrupt, the external sound card and accelerometer at other monitoring points continue to collect sound and vibration data, and save the sound and vibration data to a temporary file via DMA. This part is implemented by DMA, without CPU control, saving space and energy.

[0021] Preferably, the process of obtaining the fault type of the GIS knife switch in step S4 is:

[0022] The sound and vibration data of the GIS knife switch for a fixed time are collected and saved. The sound data is saved after denoising. The vibration data and the denoised sound data are used as input data of the pre-trained machine learning model to output the fault type.

[0023] Preferably, after the sound and vibration information processing process of the collection and monitoring point that has triggered the interruption is completed, the main controller analyzes and judges the temporary files saved at other collection and monitoring points except the collection and monitoring point that triggered the interruption: if the sound loudness frequency exceeds the sound loudness frequency threshold or the vibration amplitude frequency exceeds the sound loudness frequency threshold, it is determined that there is an abnormality, the corresponding collection and monitoring point is interrupted, and the time point of the triggering interrupt is recorded, and step S4 is executed; otherwise, it is determined that there is no abnormality, the temporary file is deleted immediately, and then the main controller enters step S2.

[0024] Preferably, the process of locating the fault area in step S5 is as follows:

[0025] The positions of the two groups of monitoring points that first triggered the interruption in the recorded log information are selected as the arc endpoints of the GIS knife switch shell, and the center of the GIS knife switch cylinder is used as the center of the arc. The sector area is divided accordingly as the fault occurrence area.

[0026] This application also proposes a multi-monitoring point sound and vibration collection device for fault diagnosis of GIS knife switches, comprising: multiple sound and vibration information collection terminals and a main controller, each of which includes an external sound card and an accelerometer positioned alongside the external sound card. Each sound and vibration information collection terminal serves as a collection monitoring point and is positioned on the housing of the GIS knife switch. The multiple sound and vibration information collection terminals are equally spaced around the housing of the GIS knife switch. The main controller is connected to each external sound card and controls the sound data collection process. The external sound card is fixed to the housing of the GIS knife switch. The main controller is connected to each accelerometer via a wire, communicating via I2C. The main controller controls the accelerometer to collect acceleration data and transmits it back to the main controller, which is then processed and converted into vibration data. The accelerometer is positioned on the housing of the GIS knife switch and fixed next to the external sound card, serving as a monitoring point. The device does not require a physical connection to the GIS knife switch and has no impact on the operation of the GIS knife switch.

[0027] Preferably, the sound-to-electricity conversion ratio of each external sound card is the same as the sound collection frequency, and the three-axis acceleration detection accuracy of each acceleration sensor is the same as the acceleration collection frequency.

[0028] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0029] The present invention proposes a fault diagnosis method for a GIS knife switch and a multi-monitoring point sound and vibration collection device. First, the sound and vibration data of the GIS knife switch under normal working conditions are collected to determine a threshold value. Then, a multi-monitoring point collection method that combines idle uninterrupted monitoring with formal monitoring is adopted. In the idle uninterrupted monitoring, the sound loudness frequency threshold and the vibration amplitude frequency threshold are used to judge whether the data of the idle uninterrupted monitoring are abnormal. The idle uninterrupted monitoring process only turns on the collection and detection functions of the sound and vibration data, reduces energy consumption, speeds up data processing, and immediately discards sound and vibration data without abnormalities to save memory space. In the event of an abnormality, the formal monitoring mode is entered. The formal monitoring process includes functions such as data collection, data processing, and data storage. After the formal sound and vibration information detection is completed, the threshold judgment is performed on the data monitored by the remaining collection and monitoring points, thereby improving the reliability of the fault diagnosis result. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 A schematic diagram showing a flow chart of a fault diagnosis method for a GIS knife switch proposed in Example 1 of the present invention;

[0031] Figure 2 A schematic diagram showing the fault location proposed in Example 2 of the present invention;

[0032] Figure 3 A schematic diagram showing the application of the sound and vibration information collection device in the fault diagnosis of the GIS knife switch proposed in Example 3 of the present invention. DETAILED DESCRIPTION

[0033] The accompanying drawings are for illustrative purposes only and are not to be construed as limiting this patent;

[0034] In order to better illustrate this embodiment, some parts of the drawings may be omitted, enlarged, or reduced, and do not represent the actual size;

[0035] It is understandable to those skilled in the art that descriptions of certain well-known contents may be omitted in the drawings.

[0036] The positional relationships described in the drawings are for illustrative purposes only and should not be construed as limiting this patent;

[0037] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0038] Example 1

[0039] This embodiment proposes a fault diagnosis method for a GIS knife switch. Figure 1 The method includes the following steps:

[0040] S1. Measure sound and vibration data multiple times on the GIS knife switch housing, using the locations with the most pronounced sound loudness and vibration amplitude as collection points. Place an external sound card and vibration sensor to collect and analyze sound and vibration data under normal operating conditions of the GIS knife switch to determine the sound loudness frequency threshold and the vibration amplitude frequency threshold.

[0041] S2. Continuous monitoring during off-peak hours: Multiple monitoring points continuously collect sound and vibration data from GIS knife switches, and perform Fourier transform on the collected data to obtain the sound loudness frequency and vibration amplitude frequency;

[0042] In this embodiment, the non-stop monitoring during off-peak hours is 24 hours a day. In this mode, only the collection and detection functions of the sound and vibration data are enabled, which reduces energy consumption and speeds up data processing.

[0043] S3. Determine whether the sound loudness frequency exceeds the sound loudness frequency threshold or whether the vibration amplitude frequency exceeds the sound loudness frequency threshold. If so, trigger an interrupt at the corresponding acquisition monitoring point, record the time of the interruption triggering, and execute step S4. Otherwise, if there is no abnormality in the GIS knife switch, discard the collected sound and vibration data, and return to S2. In this embodiment, when the sound and vibration data do not exceed the threshold, the sound and vibration data are immediately discarded to save memory space.

[0044] S4. Start formal monitoring: Enter the sound and vibration information processing process of the monitoring point that has triggered the interruption of collection, and obtain the fault type of the GIS knife switch;

[0045] S5. The time point at which the interrupt is triggered and the number of the current acquisition monitoring point that triggers the interruption are recorded as log information, and the fault area is located and the location information is saved;

[0046] S7. Return to step S3 for the data collected from the remaining monitoring points.

[0047] In this embodiment, before implementing multi-monitoring point data collection for both off-peak and regular monitoring, sound and vibration data are collected from the GIS knife switch under normal operating conditions to determine the subsequent sound loudness frequency threshold and vibration amplitude frequency threshold. This determination utilizes the fact that the primary operating frequency of GIS operating equipment remains constant. Specifically, the amplitude and frequency of the collected sound and vibration data from the GIS knife switch under normal operating conditions are analyzed. The first five frequency components and their amplitudes are recorded for each of these data. The amplitudes of these first five frequency components are then averaged to serve as the sound loudness frequency threshold and vibration amplitude frequency threshold, respectively.

[0048] In this embodiment, multiple sound and vibration collection monitoring points are used to enhance the reliability of fault diagnosis. The GIS knife switch is equipped with multiple sound and vibration collection monitoring points, each equipped with monitoring equipment. The monitoring equipment includes an external sound card and an accelerometer. The sound data is collected by the external sound card, and the vibration data is collected by the accelerometer. The external sound card and accelerometer are placed together as monitoring equipment for a sound and vibration monitoring point.

[0049] Each external sound card collects the sound of the GIS knife switch in operation at a frequency of 44100Hz and a fixed duration of 10 seconds and sends it to the main controller, saving it as a WAV file. In actual implementation, it can also be saved as a file in other formats at common frequencies such as 20000Hz and 40000Hz. The main controller then converts the WAV file into a PCM file, and then the PCM file is directly converted into a frequency domain diagram of the audio segment after fast Fourier transform. The frequency domain diagram is used to analyze the loudness and frequency of the sound.

[0050] Each accelerometer collects vibration acceleration information from the GIS knife switch during operation at a fixed frequency of 3200 Hz and a fixed duration of 10 seconds, and transmits it to the main controller. In practice, the external sound card can collect information at common frequencies such as 800 Hz and 1600 Hz. In the main controller, the acceleration data is processed into vibration data and then, after a fast Fourier transform, directly generates a frequency domain plot of this vibration data. The frequency domain plot is used to analyze the amplitude and frequency of the vibration information.

[0051] In this embodiment, except for the monitoring point that triggers the interrupt, the external sound card and accelerometer at other monitoring points continue to collect sound and vibration data, and save the sound and vibration data to a temporary file via DMA. This part is implemented by DMA, which does not require CPU control, does not add excessive CPU burden, and saves space and energy.

[0052] After the sound and vibration data analysis and processing are completed, they are compared with the set threshold. When the amplitude of the non-main frequency component in the sound or vibration data does not exceed the threshold, it is determined that there is no abnormality, the data is directly discarded, and the next data collection is carried out directly.

[0053] When the amplitude of the non-main frequency component in the sound or vibration data exceeds the threshold, it is determined to be abnormal and an interrupt is triggered. The time point of the interrupt triggering is recorded, and then the sound and vibration information processing program is entered to perform formal sound and vibration information detection.

[0054] After the sound and vibration information processing process of the collection monitoring point that has triggered the interruption is completed, the main controller analyzes and judges the temporary files saved by other collection monitoring points except the collection monitoring point that triggered the interruption: if the sound loudness frequency exceeds the sound loudness frequency threshold or the vibration amplitude frequency exceeds the sound loudness frequency threshold, it is determined that there is an abnormality, the corresponding collection monitoring point is interrupted, and the time point of the triggering interrupt is recorded, and step S4 is executed; otherwise, it is determined that there is no abnormality, the temporary file is immediately deleted, and the main controller then enters step S2.

[0055] Example 2

[0056] The process of obtaining the fault type of the GIS knife switch in step S4 is as follows:

[0057] The sound and vibration data of the GIS knife switch for a fixed time are collected and saved. The sound data is saved after denoising. The vibration data and the denoised sound data are used as input data of the pre-trained machine learning model to output the fault type.

[0058] In this embodiment, the pre-trained machine learning model can be an existing relatively mature neural network, such as a convolutional neural network, which is a pre-trained neural network obtained after training with a large number of existing data sets. Currently, support vector machines, BP neural networks, and MobileNet are used in this field.

[0059] The process of locating the fault area in step S5 is as follows:

[0060] The GIS knife switch is a cylinder. The positions of the two sets of collection monitoring points that first triggered the interruption in the recorded log information are selected as the arc endpoints of the GIS knife switch shell. The center of the GIS knife switch cylinder is used as the center of the arc, and a sector-shaped area is divided accordingly as the fault occurrence area.

[0061] like Figure 2 As shown, the acquisition monitoring point 1 (the location of the external sound card 101 and the acceleration sensor 102) and the monitoring point 2 (the location of the external sound card 201 and the acceleration sensor 202) are the two groups of acquisition monitoring points that trigger the interrupt first, so the fault occurrence area can be divided based on this. Figure 2 In the figure, the shaded part is the arc surface, the positions of the two groups of monitoring points are the endpoints of the arc, and the area surrounded by the dotted line is the area where the fault occurs.

[0062] Example 3

[0063] This embodiment proposes a sound and vibration information collection device for fault diagnosis of GIS knife switch, such as Figure 3 As shown, it includes: 3 sound and vibration information collection terminals and a main controller. Each sound and vibration information collection terminal includes an external sound card and an acceleration sensor placed together with the external sound card. Each sound and vibration information collection terminal is a collection monitoring point and is set on the outer shell of the GIS knife switch. Several sound and vibration information collection terminals are arranged around the outer shell of the GIS knife switch and are evenly spaced.

[0064] In this embodiment, see Figure 3, the external sound card includes a first external sound card 1, a second external sound card 3 and a third external sound card 5, the acceleration sensor includes a first acceleration sensor 2, a second acceleration sensor 4 and a third acceleration sensor 6, the main controller is connected to each external sound card through a USB port, and the main controller controls it to collect sound data. In the specific implementation, the external sound card is fixed to the housing of the GIS knife switch through thickened double-sided tape; the main controller is connected to each acceleration sensor through a wire, and uses I2C for communication, controls the acceleration sensor to collect acceleration data and transmit it back to the main controller, and converts it into vibration data after processing. The acceleration sensor is magnetically set on the housing of the GIS knife switch in turn through the magnetic stickers attached to the back, and is fixed next to the external sound card, and serves as a monitoring point. The sound and vibration information collection device proposed in this embodiment does not need to be physically connected to the GIS knife switch, and has no effect on the operation of the GIS knife switch body.

[0065] The sound-to-electricity conversion ratio of each external sound card is the same as the sound acquisition frequency, and the three-axis acceleration detection accuracy of each accelerometer is the same as the acceleration acquisition frequency. Figure 3 As shown in the figure, an acceleration sensor and an external sound card form a group of monitoring points. The installation positions of the monitoring points of each group are the same on the GIS knife switch housing. Figure 3 In the example, the installation interval between every two groups of monitoring points on the GIS knife switch housing is Ld.

[0066] Obviously, the above embodiments of the present invention are merely examples for the purpose of clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Those skilled in the art will appreciate that other variations or modifications can be made based on the above description. It is not necessary and impossible to enumerate all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the claims of the present invention.

Claims

1. A fault diagnosis method for a GIS knife switch, characterized in that: The method comprises: S1. Measure sound and vibration data multiple times on the GIS knife switch housing, using the locations with the most pronounced sound loudness and vibration amplitude as collection points. An external sound card and vibration sensor are placed to collect and analyze sound and vibration data under normal operating conditions of the GIS knife switch to determine the frequency thresholds for sound loudness and vibration amplitude. The vibration sensor uses an accelerometer. S2. Continuous monitoring during off-peak hours: Multiple monitoring points continuously collect sound and vibration data from GIS knife switches, and perform Fourier transform on the collected data to obtain the sound loudness frequency and vibration amplitude frequency; S3. Determine whether the sound loudness frequency exceeds the sound loudness frequency threshold, or whether the vibration amplitude frequency exceeds the vibration amplitude frequency threshold. If so, the corresponding acquisition monitoring point triggers an interrupt, and records the time point of the trigger interrupt, and executes step S4; otherwise, the GIS knife switch is normal, the collected sound and vibration data are discarded, and return to S2; S4. Start formal monitoring: Enter the sound and vibration information processing process of the monitoring point that has triggered the interruption of collection, and obtain the fault type of the GIS knife switch; The process of obtaining the fault type of the GIS knife switch in step S4 is as follows: Collect and save the sound and vibration data of the GIS knife switch for a fixed period of time. After de-noising, save the sound data. Use the vibration data and the de-noised sound data as input data for the pre-trained machine learning model to output the fault type. S5. The time point at which the interrupt is triggered and the number of the current acquisition monitoring point that triggers the interruption are recorded as log information, and the fault area is located and the location information is saved; S6. Return the remaining data collected from the monitoring points to step S3; Except for the monitoring point that triggers the interrupt, the external sound card and acceleration sensor of other monitoring points continue to collect sound and vibration data, and save the sound and vibration data as a temporary file through DMA. After the sound and vibration information processing process of the collection monitoring point that has triggered the interruption is completed, the main controller analyzes and judges the temporary files saved by other collection monitoring points except the collection monitoring point that triggered the interruption: if the sound loudness frequency exceeds the sound loudness frequency threshold or the vibration amplitude frequency exceeds the vibration amplitude frequency threshold, it is determined that there is an abnormality, the corresponding collection monitoring point is interrupted, and the time point of the triggering interrupt is recorded, and step S4 is executed; otherwise, it is determined that there is no abnormality, the temporary file is immediately deleted, and then the main controller enters step S2.

2. The fault diagnosis method for GIS knife switch according to claim 1 is characterized in that: In step S1, the amplitude and frequency of the sound and vibration data of the GIS knife switch under normal working conditions are analyzed respectively, and the first five frequency components and their amplitudes of the sound and vibration data under normal working conditions of the GIS knife switch are recorded respectively. Then, the amplitudes of the first five frequency components are averaged and used as the sound loudness frequency threshold and the vibration amplitude frequency threshold, respectively.

3. The fault diagnosis method for GIS knife switch according to claim 2 is characterized in that: There are multiple sound and vibration collection monitoring points on the GIS knife switch. Each sound and vibration collection monitoring point is equipped with monitoring equipment. The monitoring equipment includes an external sound card and an acceleration sensor. The sound data is collected by the external sound card, and the vibration data is collected by the acceleration sensor. The external sound card and the acceleration sensor are placed together as monitoring equipment for a sound and vibration monitoring point.

4. The fault diagnosis method for GIS knife switch according to claim 3 is characterized in that: Each external sound card collects the sound of the GIS knife switch during operation at a frequency of H1 and a fixed duration of T seconds to the main controller and saves it as a WAV file. The main controller then converts the WAV file into a PCM file. The PCM file is then transformed directly into a frequency domain graph of this audio segment through fast Fourier transform. The frequency domain graph is used to analyze the loudness and frequency of the sound. Each acceleration sensor collects the vibration acceleration information of the GIS knife switch during operation at a frequency of H2 and a fixed duration of T seconds to the main controller. In the main controller, the acceleration data is processed into vibration data, and then transformed directly into a frequency domain graph of this vibration data through fast Fourier transform. The frequency domain graph is used to analyze the amplitude and frequency of the vibration information.

5. The fault diagnosis method for GIS knife switch according to claim 2 is characterized in that: The process of locating the fault area in step S5 is as follows: The positions of the two groups of monitoring points that first triggered the interruption in the recorded log information are selected as the arc endpoints of the GIS knife switch shell, and the center of the GIS knife switch cylinder is used as the center of the arc. The sector area is divided accordingly as the fault occurrence area.

6. A multi-monitoring point sound and vibration acquisition device for fault diagnosis of GIS knife switch, characterized in that: include: Several sound and vibration information collection terminals and a main controller, the main controller is used to implement the fault diagnosis method of the GIS knife switch described in claim 1; each sound and vibration information collection terminal includes an external sound card and an acceleration sensor placed together with the external sound card, each sound and vibration information collection terminal is a collection monitoring point, and is arranged on the housing of the GIS knife switch, and several sound and vibration information collection terminals are arranged around the housing of the GIS knife switch and are evenly spaced; the main controller is connected to each external sound card, and is controlled by the main controller to collect sound data, and the external sound card is fixed to the housing of the GIS knife switch; the main controller is connected to each acceleration sensor through a wire, uses I2C for communication, controls the acceleration sensor to collect acceleration data and transmits it back to the main controller, and converts it into vibration data after processing; the acceleration sensor is arranged on the housing of the GIS knife switch and fixed next to the external sound card, and serves as a monitoring point.

7. The multi-monitoring point sound and vibration acquisition device for fault diagnosis of GIS knife switch according to claim 6 is characterized in that: The sound-to-electricity conversion ratio of each external sound card is the same as the sound acquisition frequency, and the three-axis acceleration detection accuracy of each acceleration sensor is the same as the acceleration acquisition frequency.

Images