An industrial equipment fault diagnosis method and device based on acoustic imaging and a medium

Abstract

This invention discloses a method, device, and medium for industrial equipment fault diagnosis based on acoustic imaging, belonging to the field of equipment fault diagnosis technology. The method includes: reconstructing the spatial acoustic energy distribution of multi-channel sound signals to form spatial acoustic energy distribution data; performing acoustic imaging reconstruction on the equipment operating area according to the spatial acoustic energy distribution data to form a sound field spatial distribution map; extracting spatial boundary information and spatial overlap information from the sound field spatial distribution map; performing sound field region segmentation on the spatial boundary information and spatial overlap information; and performing multi-device sound field separation on candidate sound field region data and cross-sounding region data to output independent sound field data for each device; and performing continuous time scanning and spatial position change comparison on the independent sound field data for each device to output equipment fault diagnosis information. This invention achieves a clear representation of the device-level sound source structure in complex sound field environments through the collaborative processing of sound field region segmentation and multi-device sound field separation.

Description

Technical Field

[0001] This invention relates to the field of equipment fault diagnosis technology, and in particular to a method, equipment and medium for diagnosing industrial equipment faults based on acoustic imaging. Background Technology

[0002] In recent years, with the continuous improvement of the scale and automation level of industrial equipment operation, equipment condition monitoring and fault diagnosis methods have gradually developed towards intelligence and visualization. Acoustic monitoring, as a non-contact means of equipment condition perception, can reflect the operating status of internal mechanical structures by collecting sound signals generated during equipment operation. It has been widely used in industrial scenarios such as rotating machinery, compressors, and transmission systems. With the development of microphone array technology and digital signal processing technology, researchers have gradually introduced sound source localization and acoustic imaging methods into the field of industrial equipment monitoring. By acquiring multi-channel sound signals and spatially reconstructing the sound energy, acoustic images reflecting the distribution of sound sources are generated, thereby realizing the visual expression of the sound field characteristics of the equipment operating area.

[0003] However, existing methods have shortcomings in sound field differentiation and abnormal sound event identification in multi-device collaborative operation scenarios. Existing methods mostly rely on single sound source localization or simple acoustic feature analysis to determine the operating status of equipment. When multiple industrial devices operate simultaneously, the sound sources generated by each device easily overlap spatially, resulting in a complex sound field structure and difficulty in distinguishing the acoustic contributions of different devices. Furthermore, they do not adequately utilize the spatial structure and temporal evolution of the sound field, making it difficult to accurately identify abnormal sound events and determine their corresponding devices in complex sound field environments. This affects the accuracy and stability of fault diagnosis results, and fails to meet the application requirements of refined condition monitoring and fault diagnosis for industrial equipment. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an industrial equipment fault diagnosis method based on acoustic imaging to address the shortcomings in sound field differentiation and abnormal sound event identification in multi-device collaborative operation scenarios.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for fault diagnosis of industrial equipment based on acoustic imaging, comprising: acquiring multi-channel sound signals of the equipment operating area; performing spatial acoustic energy distribution reconstruction processing on the multi-channel sound signals to form spatial acoustic energy distribution data; performing acoustic imaging reconstruction on the equipment operating area according to the spatial acoustic energy distribution data to generate a continuous acoustic image sequence; identifying sound field feature regions from the continuous acoustic image sequence; performing spatial continuity matching and region attribution labeling on the sound field feature regions to form a sound field spatial distribution map; extracting spatial boundary information and spatial overlap information from the sound field spatial distribution map; performing sound field region splitting on the spatial boundary information and spatial overlap information; outputting candidate sound field region data and cross-sounding region data; performing multi-device sound field separation on the candidate sound field region data and cross-sounding region data; and outputting equipment-independent sound field data; performing continuous time scanning and spatial position change comparison on the equipment-independent sound field data; identifying abnormal sound segments; and determining abnormal sound events of the industrial equipment based on the spatial position distribution and temporal evolution relationship of the abnormal sound segments, and outputting equipment fault diagnosis information.

[0007] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the specific steps for forming spatial acoustic energy distribution data are as follows: Perform spatial response difference analysis between multi-channel audio signals to obtain spatial acoustic energy correlation information; Based on the spatial acoustic energy correlation information, the spatial acoustic energy distribution of the equipment operating area is mapped, and an acoustic energy distribution matrix is ​​output. The acoustic energy distribution matrix is ​​spatially gridded based on the spatial location relationship of the equipment operating area to generate spatial acoustic energy distribution data.

[0008] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the specific steps for generating a continuous acoustic image sequence are as follows: The acoustic image of the equipment operating area is initialized based on the spatial acoustic energy distribution data to generate preliminary acoustic image data; Spatial continuity reconstruction is performed on the preliminary acoustic image data to form enhanced acoustic image data; The enhanced acoustic image data is temporally aligned and spatially registered with the device's operating area to generate a continuous acoustic image sequence.

[0009] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the specific steps for forming the sound field spatial distribution map are as follows: The sound field feature information of each frame of a continuous acoustic image sequence is extracted, and the sound field feature regions with sound energy changes are identified in the sound field feature information. Spatial relationship localization of sound field characteristic regions to determine sound field boundaries and range; The sound field boundary and sound field range are continuously matched according to the temporal order of the continuous acoustic image sequence, and integrated to generate a spatiotemporal acoustic feature dataset. The spatiotemporal acoustic feature dataset is categorized and labeled and spatially clustered to form a spatial distribution map of the sound field.

[0010] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the specific steps for outputting candidate sound field region data and cross-sound emission region data are as follows: The boundary information of the sound field feature regions is extracted from the sound field spatial distribution map, the boundary coordinates and shape of each sound field feature region are determined, and the data are integrated to form sound field boundary data. Based on the sound field feature regions in the sound field spatial distribution map, the sound energy overlap between each sound field feature region is compared using the regional connectivity analysis method, and the overlapping region data is output. Extract the boundaries of overlapping regions from the overlapping region data, and perform boundary comparison and region clipping processing between the boundaries of overlapping regions and the independent sound field regions in the sound field boundary data to generate initial candidate sound field data and initial cross-sound data. The initial candidate sound field data and the initial cross-sound data are unified in spatial coordinates and the regional adjacency relationship is corrected to output the candidate sound field region data and the cross-sound region data.

[0011] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the output device provides independent sound field data, and the specific steps are as follows: Extract the spatial inclusion relationship and boundary intersection relationship between the candidate sound field region data and the cross-sound emission region data, and decompose the overlapping sound energy distribution in the cross-sound emission region data according to the spatial inclusion relationship and boundary intersection relationship to generate sound energy allocation data; The sound energy distribution data is mapped to the candidate sound field region data to form independent sound field segment data; Independent sound field segment data is integrated with candidate sound field region data to form device-independent sound field data.

[0012] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the specific steps for identifying abnormal sound segments are as follows: The device's independent sound field data is segmented into time frame sequences to generate sound field frame sequence data; The sound field change trajectory data is obtained by comparing adjacent time frames in the sound field frame sequence data frame by frame. Identify the enhanced fluctuation segment and the continuous offset segment in the sound field change trajectory data, and extract segments from the enhanced fluctuation segment and the continuous offset segment to output abnormal sound segments.

[0013] As a preferred embodiment of the industrial equipment fault diagnosis method based on acoustic imaging described in this invention, the specific steps for outputting equipment fault diagnosis information are as follows: Based on the spatial distribution of abnormal sound segments, the abnormal sound segments are matched to the corresponding spatial locations of industrial equipment to generate abnormal sound event location data. Extract the acoustic intensity variation characteristics, frequency distribution characteristics, and time duration characteristics of abnormal sound segments from the abnormal sound event location data, and perform feature matching between the acoustic intensity variation characteristics, frequency distribution characteristics, and time duration characteristics and preset fault acoustic characteristics to output equipment fault diagnosis information.

[0014] In a second aspect, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein the computer program, when executed by the processor, implements any step of the industrial equipment fault diagnosis method based on acoustic imaging as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the industrial equipment fault diagnosis method based on acoustic imaging as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: Through the collaborative processing of sound field region segmentation and multi-device sound field separation, a clear expression of the device-level sound source structure in complex sound field environments is achieved. By extracting spatial boundary information and spatial overlap information from the sound field spatial distribution map and performing sound field region segmentation, the sound field structure within the device operating area can be orderly divided according to spatial boundary relationships, forming candidate sound field region data and cross-sound emission region data. This provides a stable spatial representation basis for the attribution of sound sources in complex sound fields. Multi-device sound field separation is performed on the candidate sound field region data and cross-sound emission region data, allowing the overlapping sound energy distribution in the cross-sound emission regions to be decomposed according to spatial inclusion and boundary intersection relationships, forming independent sound field data for each device. This provides clear data support for the identification of abnormal sound segments and the determination of abnormal sound events, establishing a stable correspondence between the operating status of industrial equipment and changes in the sound field. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of an industrial equipment fault diagnosis method based on acoustic imaging.

[0019] Figure 2 This is a flowchart for outputting the spatial distribution map of the sound field.

[0020] Figure 3 A flowchart for outputting independent sound field data for the device.

[0021] Figure 4 This is a flowchart for outputting device fault diagnosis information. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for diagnosing industrial equipment faults based on acoustic imaging, comprising the following steps: S1. Acquire multi-channel sound signals from the operating area of ​​the equipment, and perform spatial sound energy distribution reconstruction processing on the multi-channel sound signals to form spatial sound energy distribution data.

[0026] Acquire multi-channel sound signals from the equipment operating area, perform spatial response difference analysis between the multi-channel sound signals, and obtain spatial acoustic energy correlation information.

[0027] Specifically, within the equipment's operating area, microphones synchronously collect the sounds generated during equipment operation. Each microphone corresponds to an independent sound channel, and each microphone receives samples to acquire multi-channel sound signals. The multi-channel sound signals contain information on the sound amplitude changes (sound energy values), time series information, and microphone position correspondence information for each sound channel. After obtaining the multi-channel sound signals, the time series of each sound channel are uniformly aligned. Then, according to the microphone position correspondence information, the sound amplitude changes of each sound channel at the same time position are compared channel by channel to extract the differences in sound amplitude, sound energy changes, and sound arrival time between different sound channels. The differences in sound amplitude, sound energy changes, and sound arrival time are then organized according to the microphone position distribution relationship to form spatial sound energy correlation information. Spatial sound energy correlation information refers to the data set formed by time alignment and inter-channel difference analysis of the multi-channel sound signals, which reflects the correspondence between the differences in sound amplitude, sound energy changes, and sound arrival time between different sound channels and their spatial positions.

[0028] Furthermore, when acquiring multi-channel sound signals from the equipment operating area, multiple sound acquisition points are deployed within the operating area. These points are distributed according to their spatial location within the operating area, maintaining a distance between them to form a spatial distribution structure. Each sound acquisition point corresponds to an independent sound channel. Each sound acquisition point synchronously acquires the sound signals generated during equipment operation at a uniform sampling frequency, and the time series of each sound channel is aligned based on a unified time reference to ensure time consistency between different sound channels. A spatial coordinate relationship consistent with the operating area is established based on the spatial location of each sound acquisition point, and the spatial location corresponding to each sound acquisition point is calibrated to ensure a correspondence between the sound signals of each sound channel and the spatial location within the operating area.

[0029] After establishing the spatial coordinate relationship of the equipment operating area, the installation positions of each industrial device in the equipment operating area are registered to obtain the position coordinate range of each industrial device in the equipment operating area. That is, based on the actual layout position of each industrial device in the equipment operating area, the spatial boundary range, center position coordinates or coverage area coordinates of each industrial device are determined, and the position coordinate information of each industrial device is uniformly registered in the spatial coordinate relationship to form industrial device position correspondence data, so that the actual spatial position of the industrial device is established with the spatial position in the equipment operating area.

[0030] Background noise suppression processing is performed on the acquired multi-channel sound signals. Stable noise components in each sound channel are identified and weakened to reduce environmental noise interference. Simultaneously, reflection paths during sound propagation are suppressed to reduce the impact of spatial reflection on sound energy distribution. After time alignment and spatial location calibration, the sound intensity of each sound channel at the same time location is compared to extract spatial response difference information between different sound channels. Based on the spatial location coordinate relationship, the spatial response difference information is mapped to the corresponding spatial location in the device operating area. During the mapping process, the sound intensity of each sound channel is weighted according to the spatial distance between each sound acquisition point and the corresponding spatial location (the weight is obtained based on the spatial distance between each sound acquisition point and the corresponding spatial location, the closer the distance, the greater the weight, and the sound intensity weight of each sound channel is obtained by normalizing the reciprocal of each distance). This allows the sound energy value of each spatial location to comprehensively reflect the contribution of multiple sound channels, thereby forming spatial sound energy distribution data.

[0031] The formula for calculating acoustic energy value is as follows: ; in, The x-coordinate of the sound acquisition point. The vertical coordinate represents the sound acquisition point. Indicates the sound collection points in the operating area of ​​the device. The sound energy value, Indicates the first The sequence number of each audio channel. Indicates the number of audio channels. Indicates the first The sound intensity of each sound channel at the current moment. Indicates the first Each sound channel corresponds to a sound acquisition point. Sound intensity weighting.

[0032] Based on the spatial acoustic energy correlation information, the spatial acoustic energy distribution of the equipment operating area is mapped, and the acoustic energy distribution matrix is ​​output.

[0033] Specifically, when performing spatial acoustic energy distribution mapping processing on the equipment operating area according to spatial acoustic energy correlation information, the system reads the recorded sound amplitude differences, acoustic energy change differences, sound arrival time differences, and microphone position information from the spatial acoustic energy correlation information. Based on the microphone position information, spatial position coordinates are established in the equipment operating area, and the sound amplitude differences, acoustic energy change differences, and sound arrival time differences are mapped to the spatial position coordinates in the equipment operating area. Within the spatial position coordinates, the acoustic energy change state at each spatial position is organized position by position, and the acoustic energy change relationship between adjacent spatial positions is continuously organized. That is, the acoustic energy change state at different spatial positions is arranged according to the spatial position order of the equipment operating area. The arranged acoustic energy values ​​of each spatial position are then matrix-organized in a row and column manner, so that the acoustic energy distribution state of different spatial positions in the equipment operating area forms a structured arrangement result, and the acoustic energy distribution matrix is ​​output. The acoustic energy distribution matrix is ​​a matrix formed by spatial acoustic energy reconstruction of multi-channel sound signals, representing the acoustic energy values ​​at each spatial position in the equipment operating area, and is used to reflect the distribution of sound field intensity.

[0034] The acoustic energy distribution matrix is ​​spatially gridded based on the spatial location relationship of the equipment operating area to generate spatial acoustic energy distribution data.

[0035] Specifically, when spatially gridding the acoustic energy distribution matrix based on the spatial location relationship of the equipment operating area, the spatial coordinate information of each row and column in the acoustic energy distribution matrix is ​​read, and the covered area is divided into spatial grids according to the spatial location relationship of the equipment operating area, that is, the equipment operating area is divided into continuous spatial grid positions. The acoustic energy value of each row and column in the acoustic energy distribution matrix is ​​mapped to the divided spatial grid position, with each spatial grid position corresponding to one acoustic energy value; the spatial grid positions are arranged according to the spatial location order of the equipment operating area to ensure a continuous spatial correspondence between the spatial grid positions; the spatial grid positions and their corresponding acoustic energy values ​​are organized to form a regular expression of the acoustic energy distribution state of different spatial grid positions within the equipment operating area, thereby generating spatial acoustic energy distribution data.

[0036] S2. Perform acoustic imaging reconstruction on the equipment operating area according to the spatial acoustic energy distribution data to generate a continuous acoustic image sequence. Identify the sound field feature regions from the continuous acoustic image sequence, match the spatial continuity relationship of the sound field feature regions and assign them to regions to form a sound field spatial distribution map.

[0037] The acoustic imaging reconstruction process is based on the spatial mapping relationship of spatial acoustic energy distribution data in the operating area of ​​the equipment, which is used to reflect the energy distribution state of the sound source in space.

[0038] The acoustic image of the equipment operating area is initialized based on the spatial acoustic energy distribution data to generate preliminary acoustic image data.

[0039] Specifically, the acoustic energy values ​​and spatial coordinates of each spatial grid location recorded in the spatial acoustic energy distribution data are read. An image position frame consistent with the spatial coordinates is established based on the spatial position range of the equipment operating area. The corresponding image positions are set in the image position frame according to the spatial position order of the equipment operating area. The acoustic energy values ​​corresponding to each spatial grid location are read one by one in the image position frame and mapped to the corresponding image positions, so that the numerical state in the image position and the spatial grid position in the equipment operating area are in a one-to-one correspondence. After completing the mapping of all spatial grid positions to image positions, the acoustic energy values ​​of each image position in the image position frame are arranged in the spatial position order of the equipment operating area, so that the acoustic energy distribution state of each spatial grid location in the equipment operating area can form a complete expression structure in the image position frame, thereby generating preliminary acoustic image data.

[0040] Spatial continuity reconstruction is performed on the preliminary acoustic image data to form enhanced acoustic image data.

[0041] Specifically, the acoustic energy values ​​and coordinate information of each image location in the preliminary acoustic image data are read. Following the distribution order of the image location coordinates within the device's operating area, each image location is arranged point by point to determine the spatial adjacency relationship between adjacent images in the horizontal and vertical directions. Based on this spatial adjacency relationship, the acoustic energy value differences, acoustic energy change directions, and acoustic energy change amplitudes of adjacent image locations are compared one by one to determine whether the acoustic energy changes between adjacent image locations have continuous characteristics. For adjacent image locations where the acoustic energy value differences are within a preset range and the change directions are consistent, corresponding spatial continuity constraints are established to create a continuous connection between adjacent image locations. For adjacent image locations where the acoustic energy value differences exceed a preset range, the acoustic energy change boundaries are identified by combining the spatial distance between adjacent image locations and the acoustic energy distribution of surrounding adjacent image locations. This identifies the boundary between areas of continuous acoustic energy change and areas of abrupt acoustic energy change, preventing incorrect connections between different sound field regions. After establishing the spatial continuity constraint relationship between each image position, the acoustic energy value corresponding to each image position in the continuously changing area is adjusted for consistency according to the established spatial continuity constraint relationship between each image position. This ensures a smooth transition in the acoustic energy change state between adjacent image positions in the continuously changing area and preserves the original change boundary in the acoustic energy abrupt change area. After the consistency adjustment, each image position is reorganized according to its spatial coordinates so that the acoustic energy value corresponding to each image position in the equipment operating area forms a continuously distributed image representation structure in the horizontal and vertical directions, thus obtaining enhanced acoustic image data.

[0042] Furthermore, the preset variation range is determined based on the acoustic energy fluctuation characteristics of the equipment under normal operating conditions. By statistically analyzing the acoustic energy differences between adjacent spatial locations in historical normal operating data, the upper limit of the variation range is obtained. The preset variation range is then relaxed based on the upper limit of the variation range. The specific value is generally set to 1.2 to 1.5 times the maximum value of normal acoustic energy difference, so as to avoid cross-sound field misconnection while ensuring continuous sound field connection.

[0043] The enhanced acoustic image data is temporally aligned and spatially registered with the device's operating area to generate a continuous acoustic image sequence.

[0044] Specifically, based on the image position coordinates and acoustic energy values ​​recorded in the enhanced acoustic image data, a one-to-one correspondence is made with the spatial position coordinates of the device operating area to match the image positions in the enhanced acoustic image data with the spatial positions of the device operating area. Based on the time series information of the device operating area, the enhanced acoustic image data is arranged in chronological order to ensure that each frame of the enhanced acoustic image data is aligned in the actual acquisition order in the time dimension. Images at different time positions in the enhanced acoustic image data are organized and arranged in the order of time change, so that the correspondence between the acoustic energy value at each time point in the device operating area and the device operating area in the spatial position forms a continuous image sequence, i.e., a continuous acoustic image sequence.

[0045] The sound field feature information of each frame of a continuous acoustic image sequence is extracted, and the sound field feature regions with sound energy changes are identified.

[0046] Specifically, the sound field feature information of each frame in the continuous acoustic image sequence is read, including the spatial coordinates and corresponding sound energy value of each frame; the sound energy value of each frame is analyzed for differences to compare the sound energy changes between different spatial locations; based on the magnitude of the sound energy value change, the region with sound energy change is identified. That is, the identification of the sound field feature region is based on the difference in sound energy value at adjacent spatial locations. Through a preset sound energy change threshold, the sound energy change region where the difference in sound energy value exceeds the sound energy change threshold is extracted, the sound energy change region is spatially located, the boundary of the sound field feature region is determined, and the sound field feature region with sound energy change is formed.

[0047] Furthermore, the preset acoustic energy change threshold is set according to the operating characteristics of the equipment and environmental conditions, and is usually determined through statistical analysis of historical acoustic energy data. The preset acoustic energy change threshold is set based on the maximum value of acoustic energy fluctuation during normal operation of the equipment and the abnormal acoustic energy change value that may be generated when the equipment malfunctions. The specific value is generally selected above the normal acoustic energy fluctuation range, usually set to 1.5 to 2 times the normal acoustic energy fluctuation range, to ensure that areas of abnormal acoustic energy change with a large amplitude can be identified, while avoiding misjudgment. The acoustic energy change threshold can be adjusted according to the actual equipment type, operating status and experimental data to ensure accurate identification of acoustic energy change areas.

[0048] Spatial relationship localization of sound field characteristic regions to determine sound field boundaries and range; Specifically, the differences in acoustic energy between adjacent spatial locations within the acoustic field feature region are compared, the amplitude of acoustic energy value changes is statistically analyzed, and regions with continuous changes in acoustic energy value are identified. Based on the differences in acoustic energy values ​​between adjacent spatial locations, the coordinates of each spatial location are compared one by one to determine that different spatial locations belong to the same acoustic field feature region. Based on the relationship between spatial location coordinates and acoustic energy values, the acoustic field boundary is expanded, and the acoustic field range is clarified.

[0049] The sound field boundary and sound field range are continuously matched according to the temporal order of the continuous acoustic image sequence to generate a spatiotemporal acoustic feature dataset.

[0050] Specifically, the sound field boundary and range of each frame in the continuous acoustic image sequence are obtained, and the spatial coordinates and acoustic energy values ​​of each frame are extracted. Each frame in the continuous acoustic image sequence is arranged in chronological order to ensure that the position of each frame is consistent with the temporal sequence of the preceding and following frames. Within the time dimension, position alignment is performed based on the spatial coordinates and acoustic energy values ​​of the preceding and following frames to ensure that the boundary and range of each frame continue along the time axis. The sound field boundaries of adjacent image frames are matched to ensure that the sound field range can continuously expand within the time sequence, avoiding broken or discontinuous boundaries. The sound field feature information of each frame is integrated through a spatiotemporal matching process to generate a spatiotemporal acoustic feature dataset, ensuring that the sound field features at different time points are spatially consistent, forming a complete spatiotemporal acoustic feature dataset.

[0051] The spatiotemporal acoustic feature dataset is categorized and labeled and spatially clustered to form a spatial distribution map of the sound field.

[0052] Specifically, based on the changes in acoustic energy values ​​in the spatiotemporal acoustic feature dataset, different sound field feature regions are categorized and labeled according to the range, fluctuation amplitude, and trend of acoustic energy values. During spatial clustering, the spatial relationships of sound field feature regions in each frame of the image are analyzed to identify spatial similarities between adjacent regions. Spatial clustering is then performed based on spatial distance and differences in acoustic energy values, grouping spatially identical sound field feature regions into the same group. Through category classification and spatial clustering, sound field feature regions from different time frames are integrated to generate a sound field spatial distribution map. This map, generated through acoustic imaging, displays the spatial distribution of sound field feature regions within the equipment's operating area, aiding in the identification of sound source locations and the analysis of sound field changes. The sound field spatial distribution map refers to the data result representing the spatial distribution status and changing relationships of sound field feature regions within the equipment's operating area, formed after extracting sound field feature regions from a continuous acoustic image sequence and undergoing temporal continuity matching and spatial clustering.

[0053] S3. Extract spatial boundary information and spatial overlap information from the sound field spatial distribution map, perform sound field region splitting on the spatial boundary information and spatial overlap information, output candidate sound field region data and cross-sounding region data, and perform multi-device sound field separation on the candidate sound field region data and cross-sounding region data to output device-independent sound field data.

[0054] The boundary information of the sound field feature regions is extracted from the spatial distribution map of the sound field, the boundary coordinates and shape of each sound field feature region are determined, and the data are integrated to form the sound field boundary data.

[0055] Specifically, the sound field feature region information of each frame in the sound field spatial distribution map is read, the sound energy value of each sound field feature region is compared, and the regions where the sound energy value changes are identified; the boundary position of the sound field feature region is determined by comparing the sound energy difference between adjacent spatial positions, and the boundary range is clarified by combining the amplitude of sound energy change; the accurate determination of the boundary is ensured by locating the sound energy change of adjacent spatial positions, and the coordinates and shape of the boundary of each sound field feature region are drawn. The boundary information of all sound field feature regions is integrated to form sound field boundary data.

[0056] Sound field region splitting and multi-device sound field separation are based on the continuous characteristics of sound energy in spatial distribution and the sound source contribution relationship formed by the differences in multi-channel response.

[0057] Based on the sound field feature regions in the sound field spatial distribution map, the acoustic energy overlap between each sound field feature region is compared using the regional connectivity analysis method, and the overlapping region data is output.

[0058] Specifically, the system reads the sound field feature region information of each frame in the sound field spatial distribution map, obtains the spatial coordinates and corresponding acoustic energy value of each sound field feature region, compares the spatial coordinates of adjacent sound field feature regions, identifies the intersecting parts between sound field feature regions through the region connectivity analysis method, and analyzes the overlap of acoustic energy values ​​in the intersection region. The region connectivity analysis method identifies the intersection of regions by checking the spatial coordinate relationship between adjacent sound field feature regions, that is, by comparing the spatial coordinates of two sound field feature regions to determine whether there is an overlapping region. Based on the coordinate overlap, the system analyzes the changes in acoustic energy values ​​within the overlapping region. If there is a difference in acoustic energy values ​​within the overlapping region, it indicates acoustic energy overlap. The boundary of the overlapping region is located, and the overlapping region data is output.

[0059] The overlapping region boundaries are extracted from the overlapping region data, and the overlapping region boundaries are compared with the independent sound field regions in the sound field boundary data and the region is clipped to generate initial candidate sound field data and initial cross-sound data.

[0060] Specifically, the overlapping region boundaries are extracted from the overlapping region data. By comparing the boundary coordinates of the overlapping region boundaries with the boundary coordinates of the independent sound field regions in the sound field boundary data, the intersection of the boundaries is identified. Region clipping is performed on the intersection to remove regions that do not belong to the overlapping region, ensuring that the acoustic energy value of the overlapping region is consistent with the boundary data. The clipped overlapping region data is integrated with the independent sound field regions in the sound field boundary data to generate initial candidate sound field data and initial cross-sound data, ensuring that the boundary of the overlapping region does not conflict with other regions.

[0061] The initial candidate sound field data and the initial cross-sound data are unified in spatial coordinates and the regional adjacency relationship is corrected to output the candidate sound field region data and the cross-sound region data.

[0062] Specifically, the coordinates of each spatial location in the initial candidate sound field region data and the initial cross-sound region data are compared to check for deviations or inconsistencies. For spatial locations with deviations, the coordinates are corrected based on the spatial relationship of the equipment's operating area to ensure consistency between the two sets of data. After unifying the spatial coordinates, the adjacency relationships between different regions are analyzed to identify overlapping parts and gaps between adjacent regions. Based on the adjacency relationship correction, the boundaries of overlapping regions are adjusted to remove sound field regions that do not belong to the same sound field, ensuring that the boundaries of each sound field region are clear and accurate. After the correction is completed, the corrected candidate sound field region data and cross-sound region data are integrated to ensure that the data output after the adjacency relationship correction is complete and consistent candidate sound field region data and cross-sound region data. Cross-sound region data refers to the region data with independent sound field affiliation tendency extracted based on sound field boundaries and regional relationships before splitting the overlapping regions in the sound field spatial distribution map. Cross-sound region data refers to the region data containing overlapping sound energy distribution information formed by the overlap of multiple sound field feature regions at the same spatial location.

[0063] The spatial inclusion relationship and boundary intersection relationship between the candidate sound field region data and the cross-sound emission region data are extracted. The overlapping sound energy distribution in the cross-sound emission region data is decomposed according to the spatial inclusion relationship and boundary intersection relationship to generate sound energy allocation data.

[0064] Specifically, the spatial inclusion and boundary intersection relationships are extracted from the candidate sound field region data and the cross-sound emission region data. The spatial coordinates and boundary data of each region are read, and the coordinates between different regions are compared to identify the inclusion relationship between the candidate sound field region and the cross-sound emission region. The boundaries of the candidate sound field region and the cross-sound emission region are compared to determine the position of the overlapping part and obtain the spatial coordinates and sound energy value of the overlapping region. According to the spatial inclusion relationship, the sound energy contained in the cross-sound emission region is allocated to the cross-sound emission region. According to the boundary intersection relationship, the overlapping sound energy is decomposed between the candidate sound field region and the cross-sound emission region, and the sound energy value of the overlapping region is allocated according to the coordinates and sound energy value to generate sound energy allocation data. The sound energy allocation data refers to the data result used for sound energy attribution mapping after the overlapping sound energy in the cross-sound emission region data is decomposed according to the contribution ratio of the candidate sound field region.

[0065] Furthermore, in the process of decomposing the overlapping acoustic energy distribution in the data of the cross-sound-emitting area, for the case where multiple sound sources are superimposed at the same spatial location, the contribution ratio of the acoustic energy value at each spatial location within the cross-sound-emitting area is statistically processed. Specifically, the acoustic energy value corresponding to each spatial location within the cross-sound-emitting area and the acoustic energy distribution information of the adjacent candidate sound field areas are read, and the acoustic energy change trend and acoustic energy intensity level of each candidate sound field area within the neighborhood of that spatial location are obtained respectively. Based on the acoustic energy continuity, acoustic energy change direction, and acoustic energy intensity ratio of each candidate sound field area within its neighborhood, the contribution ratio of each candidate sound field area to the acoustic energy value of the current spatial location is statistically calculated. The acoustic energy value of that spatial location within the cross-sound-emitting area is decomposed according to the contribution ratio, and the decomposed acoustic energy is allocated to the corresponding candidate sound field areas, thereby obtaining the decomposed acoustic energy value corresponding to each candidate sound field area. By repeatedly performing the acoustic energy contribution ratio statistical and decomposition processing on all spatial locations within the cross-sound-emitting area, the overlapping acoustic energy is split among multiple candidate sound field areas according to the acoustic energy distribution characteristics, forming acoustic energy allocation data.

[0066] When assigning and decomposing overlapping acoustic energy distributions in cross-sound-emitting area data, continuous temporal comparisons are performed on similar sound field areas that continuously overlap within the same time frame. Consistency judgment is made by combining the acoustic energy change trend, spatial continuity direction, and regional stability in adjacent time frames. For boundary changes caused by equipment position offset, partial obstruction, or simultaneous operation of multiple similar devices, the acoustic energy assignment results are corrected based on the continuity relationship of the sound field area within continuous time to reduce the situation of incorrect decomposition and incorrect assignment of overlapping sound fields.

[0067] The sound energy distribution data is mapped onto the candidate sound field region data to form independent sound field segment data.

[0068] Specifically, based on the decomposed acoustic energy values ​​corresponding to each spatial location in the acoustic energy allocation data and the spatial coordinate relationship of the candidate sound field regions, the decomposed acoustic energy at each spatial location is assigned and mapped according to its corresponding candidate sound field region. For each spatial location, the acoustic energy after contribution ratio decomposition is allocated to the corresponding spatial location in the relevant candidate sound field region, so that each candidate sound field region obtains the corresponding acoustic energy component. After completing the acoustic energy assignment mapping of all spatial locations, the acoustic energy values ​​in each candidate sound field region are integrated to form the corresponding independent sound field segment data. Independent sound field segment data refers to the data that characterizes the local acoustic energy distribution state of a single candidate sound field after the acoustic energy allocation data is mapped to the corresponding candidate sound field region.

[0069] Independent sound field segment data is integrated with candidate sound field region data to form device-independent sound field data.

[0070] Specifically, the system reads the spatial coordinates and acoustic energy values ​​from the independent sound field segment data and the candidate sound field region data, and compares the spatial coordinate relationships between the independent sound field segment data and the candidate sound field region data. For regions with overlapping or adjacent spatial coordinates, the acoustic energy values ​​of the independent sound field segment data and the candidate sound field region data are merged to ensure that the acoustic energy value of each spatial location correctly corresponds to the corresponding region. By checking the spatial location and acoustic energy values ​​within the region, overlapping areas or gaps are corrected to ensure that the acoustic energy value within each region is unique and correct. During the integration process, the connectivity and consistency between regions are ensured, and the independent sound field data of the device is output.

[0071] By accurately extracting sound field feature regions, identifying and processing overlapping regions, and generating independent sound field data for each device, multiple sound field region data are effectively split and integrated in a refined manner. This ensures that the sound energy distribution of different devices and sound sources can be clearly distinguished, accurately identifying and separating the sound field characteristics of different devices, reducing sound field overlap and interference, improving the accuracy and reliability of fault diagnosis, and ensuring the accuracy of independent sound field data for each device. This provides more effective support for acoustic fault diagnosis of industrial equipment.

[0072] S4. Perform continuous time scanning and spatial position change comparison on the independent sound field data of the equipment to identify abnormal sound segments. Based on the spatial position distribution and temporal evolution of the abnormal sound segments, determine the abnormal sound events of the industrial equipment and output equipment fault diagnosis information.

[0073] The device's independent sound field data is segmented into time frame sequences to generate sound field frame sequence data.

[0074] Specifically, the continuous acoustic energy value sequence in the independent acoustic field data of the device is read, and the acoustic energy data is divided into multiple time frames with equal time intervals according to the time sampling frequency of the microphone on the industrial equipment. Each frame of data represents the acoustic field characteristics and acoustic energy distribution of the industrial equipment within a time period. Each time frame includes the spatial distribution of the acoustic field region, specifically including the acoustic energy values ​​of each acoustic field region at different time points during the operation of the equipment. In the process of dividing the time frame sequence, it is ensured that the acoustic field data of each time period is independent and complete, and that the data of each time frame can accurately reflect the equipment status within the corresponding time period, thereby generating acoustic field frame sequence data.

[0075] The sound field change trajectory data is obtained by comparing adjacent time frames in the sound field frame sequence data frame by frame.

[0076] Specifically, the system reads adjacent time frames from the sound field frame sequence data. Each frame contains information on the spatial distribution and acoustic energy values ​​of the sound field region at different time points. It compares the changes in acoustic energy values ​​between adjacent time frames and statistically analyzes the amplitude and positional changes of acoustic energy values ​​between different time frames. Based on the differences in acoustic energy values ​​and spatial positional changes in each frame, it identifies the trend of sound field changes in the time dimension. The continuity and amplitude changes of the sound field are organized into sound field change trajectory data to ensure that the sound field changes between each time frame can accurately reflect the operating status of the equipment. The sound field change trajectory data is used as the basis for subsequent anomaly detection and diagnosis.

[0077] Identify the enhanced fluctuation segment and the continuous offset segment in the sound field change trajectory data, and extract segments from the enhanced fluctuation segment and the continuous offset segment to output abnormal sound segments.

[0078] Specifically, by comparing the differences in acoustic energy values ​​between adjacent time frames in the acoustic field change trajectory data, the fluctuation enhancement segment and the continuous offset segment are identified. The fluctuation enhancement segment refers to the change in acoustic energy value in the time series, while the continuous offset segment refers to the portion of the acoustic energy value that continuously deviates from the normal range (the normal range refers to the normal amplitude of acoustic energy value fluctuation during the normal operation of industrial equipment; the value is usually set according to the acoustic energy fluctuation range of industrial equipment under different working conditions, and statistical analysis is performed based on the historical operating data of the equipment to obtain the average acoustic energy value and the standard deviation of acoustic energy fluctuation under normal working conditions; usually, the normal range can be set to within 1.5 to 2 times the standard deviation of the acoustic energy value under normal working conditions; this setting is to ensure that abnormal acoustic energy changes that significantly deviate from the normal state are identified and to avoid misjudgment). Based on the amplitude change of the acoustic field, the fluctuation enhancement segment and the continuous offset segment are extracted to determine the time interval of the abnormal sound segment; the abnormal sound segment is output as the recognition result to ensure that the abnormal sound segment can accurately reflect the abnormal state of the equipment.

[0079] Furthermore, when identifying abnormal sound segments, abnormal fluctuations that are too short in duration, have a discrete spatial distribution, or do not have a continuous evolution relationship are excluded. For transient sound energy changes generated during equipment start-up and shutdown, normal sound energy fluctuations caused by load changes, and non-fault abnormal sounds generated by people talking or tools hitting, they are distinguished by differences in duration, spatial stability, and frequency distribution to avoid misjudging non-fault sounds as equipment abnormal events.

[0080] Based on the spatial distribution of abnormal sound segments, the abnormal sound segments are matched to the corresponding spatial locations of industrial equipment to generate abnormal sound event location data.

[0081] Specifically, the system reads the spatial coordinates of the abnormal sound fragments and the corresponding data of the industrial equipment locations. It then compares the spatial coordinates of the abnormal sound fragments with the spatial boundary ranges corresponding to each industrial equipment to determine if the spatial location of the abnormal sound fragment falls within the spatial boundary range of a particular industrial equipment. When the spatial coordinates of the abnormal sound fragment fall within the spatial boundary range of the corresponding industrial equipment, the abnormal sound fragment is matched to the corresponding industrial equipment spatial location. When the spatial coordinates of the abnormal sound fragment are located at the intersection of multiple industrial equipment spatial boundary ranges, the corresponding industrial equipment spatial location is determined by combining the main sound energy distribution location of the abnormal sound fragment with the spatial distance relationship between adjacent industrial equipment, and the abnormal sound event location data is output. The abnormal sound event location data refers to the data result formed after matching the spatial coordinates of the abnormal sound fragments with the corresponding industrial equipment location data, used to characterize the location of the industrial equipment corresponding to the abnormal sound.

[0082] Extract the acoustic intensity variation characteristics, frequency distribution characteristics, and time duration characteristics of abnormal sound segments from the abnormal sound event location data, and perform feature matching between the acoustic intensity variation characteristics, frequency distribution characteristics, and time duration characteristics and preset fault acoustic characteristics to output equipment fault diagnosis information.

[0083] Specifically, based on the time-series data of each abnormal sound segment in the abnormal sound event location data, the changes in sound energy intensity over time are compared, and the amplitude and trend of sound energy intensity changes are statistically analyzed. The frequency distribution of the sound field is analyzed using Fourier transform to extract the frequency domain features of the abnormal sound segments, such as the main peak position and spectral width. The duration of the abnormal sound segments is statistically analyzed to determine their temporal length and stability of change, outputting sound energy intensity change characteristics, frequency distribution characteristics, and time duration characteristics. These characteristics are then matched with preset fault acoustic features. Specifically, by comparing the sound energy intensity change characteristics, frequency distribution characteristics, and time duration characteristics of the abnormal sound segments with various feature patterns in the preset fault acoustic features, it is determined whether the abnormal sound segments match known fault acoustic patterns, ensuring the accuracy of the matching. Through feature matching, equipment fault diagnosis information is output to confirm whether the equipment has a fault and to provide the fault type or location.

[0084] Furthermore, the preset acoustic characteristics of the fault are set based on the equipment's historical operating data and fault modes. These preset acoustic characteristics include the amplitude of sound energy intensity variation, frequency characteristics (such as the position of the main peak and the spectral width), and sound energy duration characteristics (such as the duration and stability of sound energy variation). The sound energy intensity variation characteristics, frequency distribution characteristics, and duration characteristics are determined by collecting and analyzing the sound field data of the equipment under normal and fault conditions, combined with the equipment's working principle, to determine the correlation between different acoustic characteristics and fault types. For example, for motor faults, the acoustic characteristics of the fault may include abnormal increases in sound energy intensity in a specific frequency band, changes in specific frequency components in the spectrum (such as low-frequency peaks), and sudden fluctuations in sound energy. The sound energy intensity variation characteristics, frequency distribution characteristics, and duration characteristics are obtained by comparing and analyzing the acoustic signals of the equipment under normal operation and when a fault occurs.

[0085] In establishing the pre-defined acoustic features of faults, multiple sets of historical acoustic data were collected under both normal operating conditions and known fault conditions of industrial equipment. The historical acoustic data included the acoustic energy value sequence, frequency distribution information, and duration information within the corresponding time period. The collected historical acoustic data were classified and labeled according to the equipment operating status, with normal operating status data serving as the baseline data and fault status data labeled according to different fault types. Based on the labeled historical acoustic data, acoustic energy intensity change features, frequency distribution features, and time duration features were extracted for each type of data. The feature data under the same category were statistically analyzed to obtain the distribution range and variation interval of each type of feature. The corresponding feature threshold interval was determined according to the distribution range of each type of feature, and the feature threshold interval was used as the discrimination criterion to construct a set of fault acoustic features.

[0086] The characteristic threshold range is determined based on historical acoustic data statistics under normal and fault conditions. The characteristic threshold range for sound energy intensity change is set to the range that exceeds the normal operating average by 1.5 to 2 times and falls within the concentrated distribution range of fault data. The characteristic threshold range for frequency distribution is set to the concentrated distribution range of the main peak frequency and spectral width under fault conditions. The characteristic threshold range for time duration is set to the range that exceeds the upper limit of normal duration and falls within the distribution range of fault duration.

[0087] During feature matching, the acoustic intensity variation features, frequency distribution features, and time duration features of the abnormal sound segment to be detected are compared with the feature threshold intervals in each fault acoustic feature set. When the feature to be detected falls into the threshold interval of the corresponding fault feature, it is determined to be a match for the corresponding fault type. For different equipment types, corresponding fault acoustic feature sets are established, and the fault acoustic feature sets are updated and adjusted based on newly added historical data during equipment operation to improve the accuracy of fault identification. The process of constructing fault acoustic features corresponds to the process of classifying and labeling historical acoustic data, extracting features, and determining statistical intervals.

[0088] This embodiment also provides a computer device applicable to the case of an industrial equipment fault diagnosis method based on acoustic imaging, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the industrial equipment fault diagnosis method based on acoustic imaging as proposed in the above embodiment.

[0089] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0090] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the industrial equipment fault diagnosis method based on acoustic imaging as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0091] In summary, this invention achieves a clear representation of the device-level sound source structure in complex sound field environments through the collaborative processing of sound field region segmentation and multi-device sound field separation. By extracting spatial boundary and spatial overlap information from the sound field spatial distribution map and performing sound field region segmentation, the sound field structure within the device operating area can be orderly divided according to spatial boundary relationships, forming candidate sound field region data and cross-sound emission region data. This provides a stable spatial representation basis for the attribution of sound sources in complex sound fields. Multi-device sound field separation is performed on the candidate sound field region data and cross-sound emission region data, allowing the overlapping sound energy distribution in the cross-sound emission regions to be decomposed according to spatial inclusion and boundary intersection relationships, forming independent sound field data for each device. This provides clear data support for the identification of abnormal sound segments and the determination of abnormal sound events, establishing a stable correspondence between the operating status of industrial equipment and changes in the sound field.

[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for fault diagnosis of industrial equipment based on acoustic imaging, characterized in that: include, Acquire multi-channel sound signals from the equipment operating area, and perform spatial sound energy distribution reconstruction processing on the multi-channel sound signals to form spatial sound energy distribution data; Acoustic imaging reconstruction is performed on the equipment operating area based on spatial acoustic energy distribution data to generate a continuous acoustic image sequence. Sound field feature regions are identified from the continuous acoustic image sequence, and spatial continuity relationship matching and region attribution are performed on the sound field feature regions to form a sound field spatial distribution map. Spatial boundary information and spatial overlap information are extracted from the spatial distribution map of the sound field. The spatial boundary information and spatial overlap information are split into sound field regions, and candidate sound field region data and cross-sound emission region data are output. Multi-device sound field separation is performed on the candidate sound field region data and cross-sound emission region data, and device-independent sound field data are output. The system performs continuous time scanning and spatial position change comparison of independent sound field data of equipment to identify abnormal sound segments. Based on the spatial position distribution and temporal evolution of abnormal sound segments, it determines abnormal sound events in industrial equipment and outputs equipment fault diagnosis information.

2. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The specific steps for forming spatial acoustic energy distribution data are as follows: Perform spatial response difference analysis between multi-channel audio signals to obtain spatial acoustic energy correlation information; Based on the spatial acoustic energy correlation information, the spatial acoustic energy distribution of the equipment operating area is mapped, and an acoustic energy distribution matrix is ​​output. The acoustic energy distribution matrix is ​​spatially gridded based on the spatial location relationship of the equipment operating area to generate spatial acoustic energy distribution data.

3. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The specific steps for generating a continuous acoustic image sequence are as follows: The acoustic image of the equipment operating area is initialized based on the spatial acoustic energy distribution data to generate preliminary acoustic image data; Spatial continuity reconstruction is performed on the preliminary acoustic image data to form enhanced acoustic image data; The enhanced acoustic image data is temporally aligned and spatially registered with the device's operating area to generate a continuous acoustic image sequence.

4. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The specific steps for forming the spatial distribution map of the sound field are as follows: The sound field feature information of each frame of a continuous acoustic image sequence is extracted, and the sound field feature regions with sound energy changes are identified in the sound field feature information. Spatial relationship localization of sound field characteristic regions to determine sound field boundaries and range; The sound field boundary and sound field range are continuously matched according to the temporal order of the continuous acoustic image sequence to generate a spatiotemporal acoustic feature dataset. The spatiotemporal acoustic feature dataset is categorized and labeled and spatially clustered to form a spatial distribution map of the sound field.

5. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The specific steps for outputting candidate sound field region data and cross-sound emission region data are as follows: The boundary information of the sound field feature regions is extracted from the sound field spatial distribution map, the boundary coordinates and shape of each sound field feature region are determined, and the data are integrated to form sound field boundary data. Based on the sound field feature regions in the sound field spatial distribution map, the sound energy overlap between each sound field feature region is compared using the regional connectivity analysis method, and the overlapping region data is output. Extract the boundaries of overlapping regions from the overlapping region data, and perform boundary comparison and region clipping processing between the boundaries of overlapping regions and the independent sound field regions in the sound field boundary data to generate initial candidate sound field data and initial cross-sound data. The initial candidate sound field data and the initial cross-sound data are unified in spatial coordinates and the regional adjacency relationship is corrected to output the candidate sound field region data and the cross-sound region data.

6. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The output device provides independent sound field data, and the specific steps are as follows: Extract the spatial inclusion relationship and boundary intersection relationship between the candidate sound field region data and the cross-sound emission region data, and decompose the overlapping sound energy distribution in the cross-sound emission region data according to the spatial inclusion relationship and boundary intersection relationship to generate sound energy allocation data; The sound energy distribution data is mapped to the candidate sound field region data to form independent sound field segment data; Independent sound field segment data is integrated with candidate sound field region data to form device-independent sound field data.

7. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The specific steps for identifying abnormal sound segments are as follows: The device's independent sound field data is segmented into time frame sequences to generate sound field frame sequence data; The sound field change trajectory data is obtained by comparing adjacent time frames in the sound field frame sequence data frame by frame. Identify the enhanced fluctuation segment and the continuous offset segment in the sound field change trajectory data, and extract segments from the enhanced fluctuation segment and the continuous offset segment to output abnormal sound segments.

8. The industrial equipment fault diagnosis method based on acoustic imaging as described in claim 1, characterized in that: The specific steps for obtaining the output device fault diagnosis information are as follows: Based on the spatial distribution of abnormal sound segments, the abnormal sound segments are matched to the corresponding spatial locations of industrial equipment to generate abnormal sound event location data. Extract the acoustic intensity variation characteristics, frequency distribution characteristics, and time duration characteristics of abnormal sound segments from the abnormal sound event location data, and perform feature matching between the acoustic intensity variation characteristics, frequency distribution characteristics, and time duration characteristics and preset fault acoustic characteristics to output equipment fault diagnosis information.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the industrial equipment fault diagnosis method based on acoustic imaging as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the industrial equipment fault diagnosis method based on acoustic imaging as described in any one of claims 1 to 8.

Images