An abnormal sound detection device for GIS equipment

Abstract

The utility model relates to GIS equipment detection technical field discloses a GIS equipment's abnormal sound detection device, include: non -contact multichannel microphone array, install in the outside of GIS equipment shell with interval with GIS equipment shell has, non -contact multichannel microphone array configuration is gathered the original sound signal in the process of GIS equipment operation, processor is connected with non -contact multichannel microphone array, and the processor includes signal processing unit, first filter and second filter, and signal processing unit is connected with first filter and second filter respectively. The utility model can improve the detectability and distinguishability of abnormal sound signal, realize the accurate collection of high fidelity, high robustness of abnormal sound signal, guarantee the safe operation of GIS equipment and power grid, and non -contact multichannel microphone array installs in the outside of GIS equipment shell with interval with GIS equipment shell has, can avoid the interference that causes to the normal operation of GIS equipment.

Description

Technical Field

[0001] This utility model relates to the field of GIS equipment testing technology, specifically to a device for detecting abnormal noise in GIS equipment. Background Technology

[0002] Gas-insulated metal-enclosed switchgear (GIS) is a critical component of high-voltage power transmission and transformation systems, and its operating status directly affects the safety and stability of the power grid. During operation, GIS equipment may generate abnormal acoustic phenomena, i.e., unusual noises, due to factors such as loose internal structure, minor contact between components, deformation of supporting members, or breakdown discharge. These signals typically exhibit characteristics such as non-steady-state behavior, low energy, short-duration bursts, frequency sparsity, and spatial directionality, serving as important information carriers reflecting early abnormalities or potential faults in the equipment.

[0003] In contrast, GIS equipment has a closed structure and low mechanical vibration intensity, resulting in extremely low acoustic emission levels under normal operating conditions and a clean acoustic background. Therefore, once an abnormal noise event occurs, its acoustic signal energy is easily submerged by the high-intensity noise in the surrounding environment during propagation. In the typical operating environment of a substation, GIS equipment is often installed below high-voltage overhead busbars, where broadband discharge noise generated by corona discharge is persistent, covering a wide frequency range and with high sound pressure levels. Simultaneously, the low-frequency mechanical and electromagnetic composite noise generated by surrounding large equipment such as transformers and reactors has a high energy density below 1000kHz, highly overlapping with the spectral range of the GIS abnormal noise signal, further exacerbating the masking problem of the target signal.

[0004] Most existing acoustic detection methods rely on single-channel microphones, which have limited spatial coverage, lack directional selectivity, and are difficult to effectively suppress noise interference in non-target areas. Abnormal noise signals from GIS equipment suffer from weak signals, unstable direction, and are easily masked by strong background noise. In complex environments such as strong noise backgrounds, the detectability and distinguishability of abnormal noise signals from GIS equipment are low, making it impossible to accurately collect these signals. Consequently, it is impossible to accurately assess the equipment status of GIS equipment or provide fault warnings, affecting the safe operation of GIS equipment and the power grid. Utility Model Content

[0005] The present invention aims to at least solve the technical problem in the prior art that the detectability and distinguishability of abnormal noise signals of GIS equipment in complex environments such as strong noise backgrounds are low, making it impossible to accurately collect abnormal noise signals of GIS equipment, thereby making it impossible to accurately assess the equipment status of GIS equipment and provide fault warnings for GIS equipment, which affects the safe operation of GIS equipment and power grid.

[0006] To solve the above-mentioned technical problems, this utility model provides an abnormal noise detection device for GIS equipment, comprising:

[0007] A non-contact multi-channel microphone array is installed outside the housing of the GIS equipment and spaced apart from the housing of the GIS equipment. The non-contact multi-channel microphone array is configured to collect raw sound signals during the operation of the GIS equipment.

[0008] A processor, connected to the non-contact multi-channel microphone array, includes a signal processing unit, a first filter, and a second filter. The signal processing unit is connected to the first filter and the second filter, respectively. The signal processing unit is configured to determine the target direction of the abnormal sound signal in the original sound signal through sparse activation modeling. The first filter is configured to perform spatial filtering and energy enhancement on the target sound signal in the target direction to obtain a first enhanced signal. The second filter is configured to perform time-frequency domain filtering and signal enhancement on the first enhanced signal to obtain a second enhanced signal. The signal processing unit is also configured to detect abnormal noise in the GIS device based on the second enhanced signal.

[0009] In some embodiments, the signal processing unit includes:

[0010] The sparse modeling subunit is configured to perform sparse directional modeling on each frame of acoustic signal to construct a sparse directional response model.

[0011] The activation estimation subunit is configured to perform activation estimation on the sparse direction response model through sparse reconstruction, and extract the main activation direction of the abnormal sound signal as the target direction of the abnormal sound signal in the current frame.

[0012] In some embodiments, the first filter includes a spatial filter configured to perform spatial filtering and energy enhancement on the target acoustic signal in the target direction using a minimum variance distortionless response beamformer.

[0013] The second filter includes a bandpass filter configured to perform bandpass filtering on the first enhanced signal and construct a residual signal spectrum based on the spectral energy of the first enhanced signal.

[0014] In some embodiments, the non-contact multi-channel microphone array is mounted on the outside of the GIS equipment housing via a support device, and the non-contact multi-channel microphone array is a circular array or a linear array.

[0015] In some embodiments, the processor further includes a signal preprocessing unit configured to preprocess the original acoustic signal to obtain multi-channel time series data, wherein the preprocessing includes DC removal, windowing, and frame splitting.

[0016] In some embodiments, the abnormal noise detection device further includes a signal synchronizer connected to the non-contact multi-channel microphone array. The signal synchronizer is configured to convert the analog / digital signals acquired by the non-contact multi-channel microphone array into multi-channel synchronization signals and to time-calibrate each channel of the non-contact multi-channel microphone array.

[0017] In some embodiments, the abnormal noise detection device of the GIS equipment further includes a housing, the non-contact multi-channel microphone array is mounted on the housing, and the processor is mounted inside the housing.

[0018] In some embodiments, the abnormal noise detection device further includes a data interaction unit connected to the processor. The data interaction unit is also connected to a host platform via a communication interface, which includes at least one of an Ethernet interface, a WIFI interface, and a USB interface.

[0019] In some embodiments, the abnormal noise detection device further includes a first display screen, which is mounted on the housing and connected to the processor; and / or

[0020] The host platform is equipped with a second display screen, and the data interaction unit is connected to the second display screen through the signal transceiver interface.

[0021] In some embodiments, the number of array channels of the non-contact multichannel microphone array is greater than or equal to 8 channels.

[0022] The abnormal noise detection device for GIS equipment provided in this embodiment of the utility model includes a non-contact multi-channel microphone array and a processor. The non-contact multi-channel microphone array is installed outside the housing of the GIS equipment and spaced apart from the housing. The non-contact multi-channel microphone array is configured to collect raw sound signals during the operation of the GIS equipment. The processor is connected to the non-contact multi-channel microphone array and includes a signal processing unit, a first filter, and a second filter. The signal processing unit is connected to the first filter and the second filter, respectively. The signal processing unit is configured to determine the target direction of abnormal sound signals in the raw sound signals through sparse activation modeling. The first filter is configured to perform spatial filtering and energy enhancement on the target sound signals in the target direction to obtain the first... A first enhanced signal is obtained by time-frequency domain filtering and signal enhancement of the first enhanced signal by the second filter. The signal processing unit is also configured to detect abnormal noises in the GIS equipment based on the second enhanced signal. Multi-channel sound signals can be collected using a non-contact multi-channel microphone array. Through the cooperation of the signal processing unit, the first filter, and the second filter, sparse activation modeling, directional enhancement, and time-frequency domain adaptive filtering are performed to improve the detectability and distinguishability of abnormal sound signals. This achieves high-fidelity and high-robust accurate acquisition of abnormal sound signals, ensuring the safe operation of the GIS equipment and the power grid. At the same time, the non-contact multi-channel microphone array is installed outside the GIS equipment housing and is spaced apart from the GIS equipment housing to avoid interfering with the normal operation of the GIS equipment. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this utility model or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this utility model. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the abnormal noise detection device for GIS equipment according to an embodiment of the present utility model;

[0025] Figure 2 This is another structural schematic diagram of the abnormal noise detection device for GIS equipment according to an embodiment of the present utility model.

[0026] Figure label:

[0027] 1-Non-contact multi-channel microphone array, 2-Processor, 21-Signal processing unit, 211-Sparse modeling sub-unit, 212-Activation estimation sub-unit, 22-First filter, 23-Second filter, 3-Signal synchronizer, 4-Data interaction unit; 20-Upper platform. Detailed Implementation

[0028] Various embodiments and features of this utility model are described herein with reference to the accompanying drawings.

[0029] It should be understood that various modifications can be made to the embodiments described herein. Therefore, the above description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of this invention will be apparent to those skilled in the art.

[0030] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present invention and, together with the general description of the present invention given above and the detailed description of the embodiments given below, serve to explain the principles of the present invention.

[0031] These and other features of the present invention will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0032] It should also be understood that although the present invention has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of the present invention, which have the features described in the claims and are therefore all within the scope of protection defined herein.

[0033] The above and other aspects, features and advantages of the present invention will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0034] Specific embodiments of the present invention will now be described with reference to the accompanying drawings; however, it should be understood that the claimed embodiments are merely examples of the present invention, which may be implemented in various ways. Well-known and / or repeated functions and structures have not been described in detail to avoid unnecessary or redundant details that could obscure the present invention. Therefore, the specific structural and functional details claimed herein are not intended to be limiting, but merely to serve as the basis and representative basis for the claims to teach those skilled in the art to use the present invention in a variety of substantially any suitable detailed structures.

[0035] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in other embodiments,” all of which may refer to one or more of the same or different embodiments according to the present invention.

[0036] Figure 1 A schematic diagram of the abnormal noise detection device for GIS equipment according to an embodiment of this utility model is shown. Figure 1 As shown, this utility model embodiment provides an abnormal noise detection device for GIS equipment, comprising:

[0037] A non-contact multi-channel microphone array 1 is installed outside the GIS equipment housing and is spaced apart from the GIS equipment housing. The non-contact multi-channel microphone array is configured to collect raw sound signals during the operation of the GIS equipment.

[0038] Processor 2 is connected to the non-contact multi-channel microphone array. The processor includes a signal processing unit 21, a first filter 22, and a second filter 23. The signal processing unit 21 is connected to the first filter 22 and the second filter 23 respectively. The signal processing unit 21 is configured to determine the target direction of the abnormal sound signal in the original sound signal through sparse activation modeling. The first filter 22 is configured to perform spatial filtering and energy enhancement on the target sound signal in the target direction to obtain a first enhanced signal. The second filter 23 performs time-frequency domain filtering and signal enhancement on the first enhanced signal to obtain a second enhanced signal. The signal processing unit is also configured to detect abnormal noise in the GIS device based on the second enhanced signal.

[0039] The non-contact multi-channel microphone array (hereinafter referred to as microphone array) consists of several high-sensitivity miniature microphones forming a fixed array structure. It is installed on the outside of the GIS equipment housing in a non-contact manner, and the microphone array is spatially isolated from the GIS equipment (non-contact) to avoid interfering with the normal operation of the GIS equipment.

[0040] In the non-contact multi-channel microphone array 1, the position and array shape of the microphones can be adjusted according to the structural characteristics of the GIS equipment to enable the microphone array to have good spatial direction perception capabilities. For example, the non-contact multi-channel microphone array can be configured as a circular array, linear array, or local multi-unit layout according to the on-site GIS equipment layout conditions to adapt to different equipment housing shapes and installation space requirements. While meeting the array's spatial resolution requirements, the number of channels in the microphone array is greater than or equal to 8 channels, such as 8 channels, 16 channels, or 32 channels, to accommodate different sensitivity and sampling bandwidth requirements.

[0041] The microphone is preferably a MEMS (Micro-Electro-Mechanical Systems Sound Sensor) to improve the signal-to-noise ratio and signal acquisition consistency. The signal-to-noise ratio (SNR) is a core indicator of signal quality, directly reflecting the relationship between the strength of effective information and interference noise in the signal. A higher SNR value indicates a clearer signal and less noise interference.

[0042] In some embodiments, the non-contact multichannel microphone array is mounted on the outside of the GIS equipment housing via a support device.

[0043] The support device can be a bracket installed outside the GIS equipment housing. The non-contact multi-channel microphone array 1 is installed on the bracket and is at a certain distance from the GIS equipment housing to realize non-contact and reliable acquisition of sound signals.

[0044] Processor 2 is used to process the raw acoustic signals acquired by the non-contact multi-channel microphone array 1. Signal processing unit 21, first filter 22, and second filter 23 can be integrated and mounted on a circuit board, and electrically connected to each other. First filter 22 and second filter 23 can utilize signal processing unit 21 as signal transmission nodes; that is, signal processing unit 21 is connected to first filter 22 and second filter 23 respectively. The signal processed by first filter 22 is then transmitted to second filter 23 through signal processing unit 21. Second filter 23 processes the signal to obtain a second enhanced signal and then sends it back to signal processing unit 21. Using signal processing unit 21 as a signal transmission node ensures centralized signal processing and storage. First filter 22 and second filter 23 can also be interconnected, sending the first enhanced signal obtained by first filter 22 to second filter 23 for processing without needing to send it back to signal processing unit 21, thus improving processing efficiency.

[0045] The signal processing unit 21 extracts the activation direction of the abnormal sound signal (abnormal noise signal) from the raw sound signal collected by the microphone array through directional sparsity modeling and activation estimation, thereby determining the possible spatial direction of the abnormal sound signal. The signal processing unit 21 can model the sparsity and spatial orientation characteristics of the abnormal noise signal, thereby accurately determining the possible spatial direction.

[0046] The activation direction obtained by the signal processing unit 21 from the original sound signal is the instantaneous activation direction, which can dynamically determine the activation direction of the abnormal sound signal.

[0047] After the signal processing unit 21 obtains the activation direction of the abnormal acoustic signal, the first filter 22 performs spatial filtering and energy enhancement on the target acoustic signal in the target direction (activation direction), thereby suppressing background interference from non-target directions. The first filter 22 performs filtering and signal enhancement in the spatial domain.

[0048] After the first filter 22 performs spatial filtering and energy enhancement on the target sound signal, it sends the processed first enhanced signal to the second filter 23. The second filter 23 performs time-frequency domain filtering and signal enhancement on the first enhanced signal to obtain the second enhanced signal. The second enhanced signal can effectively highlight the significance of the abnormal sound signal in the time-frequency domain, which is convenient for subsequent detection.

[0049] After the second enhanced signal is obtained through the processing of the first filter 22 and the second filter 23, the signal processing unit 21 receives the second enhanced signal and performs abnormal noise detection and identification on it.

[0050] The signal processing unit 21 can extract key parameters that reflect the suddenness and non-steady-state characteristics of abnormal sound signals from the second enhanced signal, construct a feature vector as the basis for judging abnormal noise events; then, it can identify abnormal noise events in the operation of the GIS equipment based on the feature vector.

[0051] After constructing the feature vector, the signal processing unit 21 performs abnormal noise event identification based on the feature vector. The specific abnormal noise event identification process includes: if the feature value of one or more components in the feature vector exceeds a corresponding preset threshold, it can be determined as an abnormal noise event; for example, when the instantaneous energy is high and has high kurtosis, while the feature values ​​corresponding to other feature components are normal, it is determined that an abnormal noise event may exist. Alternatively, if the feature vector as a whole deviates from the standard feature vector of the normal state by a distance exceeding a preset threshold (the feature vector as a whole significantly deviates from the standard feature vector of the normal state), it can be determined as an abnormal noise event.

[0052] In the above-mentioned abnormal noise event identification process, the feature vector is directly compared with the standard vector to identify the abnormal noise event, achieving rapid identification of the abnormal noise event. In other embodiments, the signal processing unit 21 can have a built-in preset abnormal noise event identification model, which is used to identify abnormal noise events. The abnormal noise event identification model is constructed and trained based on the feature vector. That is, in this embodiment, the signal processing unit 21 can use a classifier to train and discriminate the feature vector, improving the accuracy and efficiency of abnormal noise event identification.

[0053] The abnormal noise detection device for GIS equipment provided in this embodiment of the utility model comprises a non-contact multi-channel microphone array 1 and a processor 2. The non-contact multi-channel microphone array 1 is installed outside the housing of the GIS equipment and is spaced apart from the housing. The non-contact multi-channel microphone array 1 is configured to collect raw sound signals during the operation of the GIS equipment. The processor 2 is connected to the non-contact multi-channel microphone array 1 and includes a signal processing unit 21, a first filter 22, and a second filter 23. The signal processing unit 21 is connected to the first filter 22 and the second filter 23 respectively. The signal processing unit 21 is configured to determine the target direction of the abnormal sound signal in the raw sound signal through sparse activation modeling. The first filter 22 is configured to perform spatial filtering and energy filtering on the target sound signal in the target direction. The signal processing unit is further configured to perform time-frequency domain filtering and signal enhancement on the first enhanced signal to obtain a second enhanced signal. The signal processing unit is also configured to detect abnormal noises in the GIS equipment based on the second enhanced signal. A non-contact multi-channel microphone array can be used to collect multi-channel sound signals. Through the cooperation of the signal processing unit 21, the first filter 22, and the second filter 23, sparse activation modeling, directional enhancement, and time-frequency domain adaptive filtering are performed to improve the detectability and distinguishability of abnormal sound signals. This achieves high-fidelity and robust accurate acquisition of abnormal sound signals, ensuring the safe operation of the GIS equipment and the power grid. Simultaneously, the non-contact multi-channel microphone array 1 is installed outside the GIS equipment housing and spaced apart from it, avoiding interference with the normal operation of the GIS equipment.

[0054] Specifically, in this embodiment, through the cooperation of the signal processing unit 21 with the first filter 22 and the second filter 23, spatial focusing and time-frequency domain enhancement of abnormal sound signals are achieved under non-contact acquisition conditions, which can significantly improve the detectability of abnormal sound signals in complex environments. Furthermore, the second enhanced signal can accurately extract and distinguish abnormal noise events from GIS equipment, exhibiting good burst detection capability and noise robustness, improving the distinguishability of abnormal sound signals, and achieving high-fidelity, high-robust, and accurate acquisition of abnormal sound signals. In addition, the above-mentioned abnormal noise detection device can improve the signal-to-noise ratio of abnormal sound signal acquisition, thereby improving the detectability and distinguishability of abnormal sound signals, and achieving high signal-to-noise ratio extraction of abnormal sound signals under multi-interference conditions.

[0055] Processor 2 can be a graphics processing unit (GPU), a field-programmable gate array (FPGA), or an application-specific integrated circuit (ASIC), integrating the aforementioned signal processing unit 21 and filters. Processor 2 can also be a novel embedded processor integrating edge AI functionality, where the signal processing unit 1 includes components such as a neural processing unit (NPU) and / or a digital signal processor (DSP), integrated on a graphics processing unit (GPU), to address the issues of high power consumption of GPUs and FPGAs and poor flexibility of ASICs, while also being lower in cost.

[0056] In some embodiments, the processor 2 further includes a signal preprocessing unit configured to preprocess the original acoustic signal to obtain multi-channel time series data, wherein the preprocessing includes DC removal, windowing, and frame splitting.

[0057] The signal preprocessing unit can be a digital signal processor (DSP) electrically connected to the signal processing unit 21. It performs standard preprocessing on the raw acoustic signals acquired by the non-contact multi-channel microphone array 1, including DC removal, windowing, and framing, thereby constructing multi-channel time-series data. For example, when the signal preprocessing unit performs framing, it sets the frame length to L and the frame shift to H, dividing the signal of each channel into M frames for subsequent processing by the signal processing unit 21.

[0058] In some embodiments, such as Figure 1 As shown, the abnormal noise detection device also includes a signal synchronizer 3, which is connected to the non-contact multi-channel microphone array 1. The signal synchronizer 3 is configured to convert the analog / digital signals collected by the non-contact multi-channel microphone array 1 into multi-channel synchronization signals, and to perform time calibration on each channel of the non-contact multi-channel microphone array 1.

[0059] Signal synchronizer 3 performs synchronous acquisition and clock calibration of multi-channel acoustic signals, that is, it performs unified sampling (sampling at a unified sampling rate) and time synchronization on all acoustic signal channels in the array. The clock control unit in signal synchronizer 3 can align the time delays between channels to ensure the consistency and synchronization of subsequent input data. Through signal synchronizer 3, the analog / digital signals acquired by the multi-channel microphone array can be converted into multi-channel synchronization signals.

[0060] Optionally, the signal synchronizer 3 has a sampling rate of ≥48kHz and 24-bit sampling accuracy, and is capable of multi-channel high-precision analog-to-digital conversion.

[0061] In specific implementations, the signal synchronizer 3 may also include a buffer unit to realize sliding frame processing and array delay compensation, thereby improving the accuracy of response to transient abnormal noise events.

[0062] In some embodiments, such as Figure 2 As shown, the signal processing unit 21 includes:

[0063] The sparse modeling subunit 211 is configured to perform sparse directional modeling on each frame of acoustic signal to construct a sparse directional response model.

[0064] The activation estimation subunit 212 is configured to perform activation estimation on the sparse direction response model through sparse reconstruction, and extract the main activation direction of the abnormal sound signal as the target direction of the abnormal sound signal in the current frame.

[0065] The sparse modeling subunit 211 performs sparse directional modeling and a sparse directional response model on each frame of signal. The activation estimation subunit 212 performs activation estimation by solving the sparse directional response model through sparse reconstruction. Based on the solution results, it extracts the maximum activation direction as the main activation direction of the abnormal sound signal in the current frame, thereby determining the target direction of the abnormal sound signal in the current frame. The solvers used by the sparse modeling subunit 211 and the activation estimation subunit 212 can be flexibly selected according to the resources of the power system, such as the number of GIS devices and the power grid operating time period, which can improve the modeling accuracy of the sparse modeling subunit 211 and the processing efficiency of the activation estimation subunit 212.

[0066] The signal processing unit 21 further includes a direction focusing subunit. After the activation estimation subunit 212 determines the target direction of the abnormal sound signal in the current frame, the direction focusing subunit can combine the target direction of the abnormal sound signal in the current frame with the main activation direction of the abnormal sound signal in the preceding frame to form a preceding direction sequence. The abnormal sound signal in the current frame is then dynamically oriented around the preceding direction sequence to determine the possible activation regions of the abnormal sound signal. In other words, the signal processing unit 21 can combine the abnormal sound signal in the preceding frame with the abnormal sound signal obtained through sparse activation modeling to perform dynamic direction focusing, further improving the detectability of the abnormal sound signal in complex environments.

[0067] In some embodiments, the first filter 22 includes a spatial filter configured to perform spatial filtering and energy enhancement on the target acoustic signal in the target direction using a minimum variance distortionless response beamformer.

[0068] The second filter 23 includes a bandpass filter configured to perform bandpass filtering on the first enhanced signal and construct a residual signal spectrum based on the spectral energy of the first enhanced signal.

[0069] The first filter 22 can use a minimum variance distortion-free response beamformer to perform spatial filtering and energy enhancement on the target acoustic signal in the target direction, effectively suppressing background interference from non-target directions and improving the detectability of abnormal acoustic signals in complex environments.

[0070] The first filter 22 obtains the first enhanced signal through spatial dynamic focusing and minimum variance distortion-free response beamforming. In specific implementation, under resource-constrained conditions, it can also be switched to a fixed-weight delay summation beamforming mode as an approximate enhancement strategy to ensure the real-time acquisition of the sound signal on site and avoid missed acquisition.

[0071] The second filter 23 performs a Fast Fourier Transform on the first enhanced signal, converting the time-domain signal into a frequency-domain signal and extracting the spectral energy of the frequency-domain signal. Then, a preset frequency range (e.g., 100Hz to 10000Hz) is selected to perform bandpass filtering on the frequency-domain signal to achieve frequency band focusing. Furthermore, based on the difference between the spectral energy and the historical average spectral energy, a residual signal spectrum is constructed to enhance the residual signal. The residual signal in the constructed residual signal spectrum can effectively highlight the prominence of sudden abnormal noises in the frequency domain.

[0072] In some embodiments, the signal processing unit 21 further includes an abnormal noise identification subunit. This subunit extracts key parameters from the second enhanced signal that reflect the suddenness and non-steady-state characteristics of the abnormal sound signal, constructs a multi-dimensional feature vector, and identifies abnormal noise events during the operation of the GIS equipment based on the feature vector. The feature vector used for abnormal noise detection and identification includes at least one of instantaneous energy, spectral centroid, energy mutation rate, and residual signal peak frequency.

[0073] Instantaneous energy represents the total energy of the signal within the current frame, reflecting the intensity of the acoustic signal; the spectral centroid measures the "center of gravity" of the energy distribution in the frequency domain, reflecting the spectral shift trend; the energy mutation rate reflects the amplitude of energy changes between adjacent frames and is used to identify signal burst boundaries; the residual signal peak frequency extracts the main frequency position from the residual signal spectrum and is used to distinguish the frequency structure characteristics of different types of abnormal noise sources. Using these multidimensional feature vectors for abnormal noise identification can extract multidimensional acoustic feature parameters in the time and frequency domains. Combined with residual signal enhancement strategies, it enables accurate identification and output of weak abnormal noise signals, effectively improving the ability to distinguish different types of abnormal noise events, accurately identifying abnormal noise events in GIS equipment, and exhibiting good abnormal noise burst identification capability and noise robustness.

[0074] In some embodiments, the abnormal noise detection device of the GIS equipment further includes a housing (not shown in the figure), the non-contact multi-channel microphone array 1 is mounted on the housing, and the processor 2 is mounted inside the housing.

[0075] The abnormal noise detection device for GIS equipment can be an independent abnormal noise signal acquisition and detection terminal, including a housing, a non-contact multi-channel microphone array 1 installed on the housing, and a processor 2 installed inside the housing. It is easy to carry and ensures high efficiency in signal acquisition and processing by combining the non-contact multi-channel microphone array 1 and the processor 2.

[0076] In other embodiments, the non-contact multi-channel microphone array 1 and the processor 2 can be arranged separately. For example, the non-contact multi-channel microphone array 1 is installed outside the GIS equipment housing, and the processor 2 is connected to the power grid master station monitoring system or edge processing platform and communicates with the non-contact multi-channel microphone array 1.

[0077] In this embodiment, the abnormal noise detection device for GIS equipment has a simple structure and flexible deployment, and can significantly improve the accuracy and stability of abnormal noise collection of GIS equipment while ensuring signal quality.

[0078] In some embodiments, such as Figure 1 As shown, the abnormal noise detection device also includes a data interaction unit 4, which is connected to the processor 2. The data interaction unit 4 is also connected to the host platform 20 through a communication interface, which includes at least one of an Ethernet interface, a WIFI interface, and a USB interface.

[0079] The upper-level platform 20 is equipped with a user interface and can be deployed on a laptop, industrial PC, or remote server. It allows users to configure acquisition parameters, view real-time beam output, and play back historical signals and spectra. It can also perform manual auxiliary judgment of abnormal noises or complete fault identification of GIS equipment based on the abnormal noise event identification results of the abnormal noise detection device.

[0080] The results of abnormal noise event identification can include the time when the abnormal noise occurred, the direction of the abnormal noise, the original sound signal data after filtering and enhancement, and the feature values ​​corresponding to the original sound signal data.

[0081] In one specific embodiment, the data interaction unit 4 is housed within a casing and electrically connected to the processor 2. The casing is provided with a signal transceiver interface connected to the data interaction unit 4, and the signal transceiver interface is also connected to a host platform. The signal transceiver interface is, for example, a wireless communication interface.

[0082] The data interaction unit also includes a storage unit (memory), which has local caching and offline writing functions. It can realize real-time transmission (writing or reading) and storage archiving of multi-level data such as sound signal acquisition data, signal filtering and enhancement results, feature vectors, and event recognition results. For example, data can be stored in the form of time segments or event triggers.

[0083] In some embodiments, the abnormal noise detection device further includes a first display screen, which is mounted on the housing and connected to the processor 2; and / or

[0084] The host platform is equipped with a second display screen, and the data interaction unit is connected to the second display screen through the signal transceiver interface.

[0085] The abnormal noise event identification results can be output to the first display screen of the abnormal noise detection device or to the second display screen of the upper platform, making it convenient for users to view and handle the abnormal noise. For example, the second enhanced signal after being filtered and enhanced by the second filter 23 can be used as the abnormal noise event identification result, allowing users to view the prominent features of the abnormal noise and intuitively understand the intensity and location of the abnormal noise. In specific implementations, the original, unenhanced sound signal corresponding to the abnormal noise (e.g., the original sound signal acquired by the non-contact multi-channel microphone array 1) can also be output and displayed.

[0086] The abnormal noise event identification result may also include the cause of the abnormal noise, etc. The signal processing unit 21 can determine the cause of the abnormal noise based on the time of occurrence of the abnormal noise, the direction of the abnormal noise, the original sound signal data after filtering and enhancement, and the feature value corresponding to the original sound signal data, and output and display it through the first display screen and / or the second display screen.

[0087] The above description is merely a preferred embodiment of this utility model and an explanation of the techniques used. Those skilled in the art should understand that the scope of disclosure involved in this utility model is not limited to the technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in this utility model.

[0088] Furthermore, although the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. Multitasking and parallel processing may be advantageous in certain environments. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this invention. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0089] Although the subject matter has been described using language specific to structural features and / or logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A noise detection device for GIS equipment, characterized in that, include: A non-contact multi-channel microphone array is installed outside the housing of the GIS equipment and is spaced apart from the housing of the GIS equipment. The non-contact multi-channel microphone array is configured to collect raw sound signals during the operation of the GIS equipment. A processor, connected to the non-contact multi-channel microphone array, includes a signal processing unit, a first filter, and a second filter. The signal processing unit is connected to the first filter and the second filter, respectively. The signal processing unit is configured to determine the target direction of the abnormal sound signal in the original sound signal through sparse activation modeling. The first filter is configured to perform spatial filtering and energy enhancement on the target sound signal in the target direction to obtain a first enhanced signal. The second filter is configured to perform time-frequency domain filtering and signal enhancement on the first enhanced signal to obtain a second enhanced signal. The signal processing unit is also configured to detect abnormal noise in the GIS device based on the second enhanced signal.

2. The abnormal noise detection device according to claim 1, characterized in that, The signal processing unit includes: The sparse modeling subunit is configured to perform sparse directional modeling on each frame of acoustic signal to construct a sparse directional response model. The activation estimation subunit is configured to perform activation estimation on the sparse direction response model through sparse reconstruction, and extract the main activation direction of the abnormal sound signal as the target direction of the abnormal sound signal in the current frame.

3. The abnormal noise detection device according to claim 1, characterized in that, The first filter includes a spatial filter configured to perform spatial filtering and energy enhancement on the target acoustic signal in the target direction using a minimum variance distortion-free response beamformer; The second filter includes a bandpass filter configured to perform bandpass filtering on the first enhanced signal and construct a residual signal spectrum based on the spectral energy of the first enhanced signal.

4. The abnormal noise detection device according to claim 1, characterized in that, The non-contact multi-channel microphone array is mounted on the outside of the GIS equipment housing via a support device. The non-contact multi-channel microphone array can be a circular array or a linear array.

5. The abnormal noise detection device according to claim 1, characterized in that, The processor further includes a signal preprocessing unit configured to preprocess the original acoustic signal to obtain multi-channel time series data, wherein the preprocessing includes DC removal, windowing, and frame segmentation.

6. The abnormal noise detection device according to claim 1, characterized in that, The abnormal noise detection device further includes a signal synchronizer connected to the non-contact multi-channel microphone array. The signal synchronizer is configured to convert the analog / digital signals acquired by the non-contact multi-channel microphone array into multi-channel synchronization signals, and to perform time calibration on each channel of the non-contact multi-channel microphone array; and / or The sampling rate of the signal synchronizer is ≥48kHz, and the signal synchronizer has 24-bit sampling accuracy.

7. The abnormal noise detection device according to claim 1, characterized in that, The abnormal noise detection device of the GIS equipment also includes a housing, the non-contact multi-channel microphone array is mounted on the housing, and the processor is installed inside the housing.

8. The abnormal noise detection device according to claim 1, characterized in that, The abnormal noise detection device also includes a data interaction unit, which is connected to the processor. The data interaction unit is also connected to the host platform through a communication interface, which includes at least one of an Ethernet interface, a WIFI interface, and a USB interface.

9. The abnormal noise detection device according to claim 8, characterized in that, The abnormal noise detection device further includes a first display screen, which is mounted on the housing and connected to the processor; and / or The host platform is equipped with a second display screen, and the data interaction unit is connected to the second display screen through the signal transceiver interface.

10. The abnormal noise detection device according to claim 1, characterized in that, The number of channels in the non-contact multichannel microphone array is greater than or equal to 8.

Images