Method and system for anomaly recognition of a blower based on multi-modal data

Abstract

The application discloses an abnormality identification method and system of a blower fan based on multi-modal data, and relates to the technical field of abnormality identification.A multi-modal framework of the blower fan, noise data and temperature data of the blower fan are used to construct multi-modal data corresponding to the blower fan, thereby improving the accuracy of the multi-modal data corresponding to the blower fan.A plurality of abnormal data are determined according to the detection of the multi-modal data corresponding to the blower fan, and each abnormal data is marked with an abnormal component corresponding thereto.An abnormality identification system is determined according to the plurality of abnormal data, the abnormal component corresponding to each abnormal data and the overall form of the blower fan, thereby improving the accuracy of the abnormality identification system.Meanwhile, the node position of each abnormal influence node, the corresponding node form and the use scene of the blower fan are used to determine the maintenance content to be maintained, and the corresponding abnormal maintenance measures are determined based on the maintenance content, the current working mode of the blower fan and the corresponding maintenance system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of anomaly identification, and particularly to an air blower anomaly identification method and system based on multi-modal data. BACKGROUND

[0002] With the development of technology, air blowers are applied to industrial scenes and serve as industrial fluid mechanical devices. A plurality of working data of the air blower is collected, the working state of the air blower is determined according to the plurality of working data of the air blower, and the working content of the air blower is marked. Meanwhile, corresponding abnormal data is determined based on the detection of the working content, and the consideration of abnormal components corresponding to each abnormal data and the overall form of the air blower is ignored, which affects the accuracy of the anomaly identification system and leads to low accuracy of the abnormal maintenance measures of the air blower. SUMMARY

[0003] The present application aims to overcome the shortcomings of the prior art, and provides an air blower anomaly identification method and system based on multi-modal data.

[0004] The present application provides an air blower anomaly identification method based on multi-modal data, which comprises the following steps:

[0005] When the air blower is in a working state, the real-time image of the air blower is determined based on the visual detection of the air blower, the plurality of component regions are determined based on the recognition of the real-time image of the air blower, and the corresponding vibration data combination is determined according to the vibration detection of each component region.

[0006] The multi-modal framework of the air blower is determined based on the region position of each component region and the corresponding vibration data combination, the multi-modal data corresponding to the air blower is constructed according to the multi-modal framework of the air blower, the noise data and the temperature data of the air blower, the multi-modal framework not only contains the current vibration state, but also implies the three-dimensional geometric structure and fault importance information of the equipment, and the multi-modal data contains the spatial position, the vibration feature, the thermodynamic feature, the acoustic feature and the scene semantics.

[0007] The plurality of abnormal data is determined according to the detection of the multi-modal data corresponding to the air blower, the abnormal components corresponding to each abnormal data are marked, the anomaly identification system is determined according to the plurality of abnormal data, the abnormal components corresponding to each abnormal data and the overall form of the air blower, and the plurality of abnormal regions of different abnormal influence levels are marked, the anomaly identification system is a comprehensive system that fuses the structural dimension and the working condition dimension of the air blower to realize fault grading.

[0008] In each abnormal area, a corresponding abnormal event is determined according to dynamic detection of the abnormal area, a plurality of abnormal influence factors are determined based on each abnormal event, real-time working data of the air blower and the abnormal identification system, and a corresponding abnormal influence route is formed, and the abnormal influence route is used as a panoramic multi-dimensional fault mapping atlas;

[0009] Based on the identification of the abnormal influence route, a plurality of abnormal influence nodes are determined, the maintenance content to be maintained is determined according to the node position of each abnormal influence node, the corresponding node form and the use scene of the air blower, the corresponding abnormal maintenance measure is determined based on the maintenance content to be maintained, the current working mode of the air blower and the corresponding maintenance system, and the abnormal influence node is defined as the most critical physical position of the maintenance operation; the node position of the abnormal influence node covers the corresponding installation coordinates, hierarchical relationship and surrounding envelope boundary; the node form of the abnormal influence node includes the corresponding form topological feature; the maintenance content to be maintained refers to the specific maintenance operation task determined to solve the fault; and the maintenance system includes the mapping of the fault mode and the repair scheme.

[0010] The embodiment of the application provides an abnormal identification system of an air blower based on multi-modal data.

[0011] Compared with the prior art, the beneficial effects of the application are:

[0012] (1) The multi-modal framework of the air blower is determined based on the region position of each component region and the corresponding vibration data, the multi-modal data corresponding to the air blower is constructed according to the multi-modal framework of the air blower, noise data and temperature data of the air blower, the multi-modal framework of the air blower is introduced, the multi-modal framework of the air blower, the noise data and the temperature data of the air blower are considered, and the accuracy of the multi-modal data corresponding to the air blower is improved.

[0013] (2) A plurality of abnormal data are determined according to the detection of the multi-modal data corresponding to the air blower, and each abnormal component corresponding to each abnormal data is marked, the abnormal identification system is determined according to the plurality of abnormal data, each abnormal component corresponding to each abnormal data and the overall form of the air blower, and a plurality of abnormal areas of different abnormal influence levels are marked, each abnormal data is further controlled, and the corresponding abnormal component is marked, the overall consideration of the plurality of abnormal data, each abnormal component corresponding to each abnormal data and the overall form of the air blower is realized, and the accuracy of the abnormal identification system is improved.

[0014] (3) According to the dynamic detection of the abnormal area, the corresponding abnormal event is determined, a plurality of abnormal influence factors are determined based on each abnormal event, real-time working data of the air blower and an abnormal identification system, and a corresponding abnormal influence route is formed; a plurality of abnormal influence nodes are determined based on the identification of the abnormal influence route, the to-be-maintained content is determined according to the node position of each abnormal influence node, the corresponding node form and the use scene of the air blower, the corresponding abnormal maintenance measure is determined based on the to-be-maintained content, the current working mode of the air blower and the corresponding maintenance system, and the overall consideration of the to-be-maintained content, the current working mode of the air blower and the corresponding maintenance system is realized, and the accuracy of the abnormal maintenance measure of the air blower is improved. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is a flowchart of an abnormal identification method of an air blower based on multi-modal data in an embodiment of the present application;

[0016] Figure 2 is a flowchart of step S11 in the abnormal identification method of the air blower based on multi-modal data in an embodiment of the present application;

[0017] Figure 3 is a flowchart of step S12 in the abnormal identification method of the air blower based on multi-modal data in an embodiment of the present application;

[0018] Figure 4 is a flowchart of step S13 in the abnormal identification method of the air blower based on multi-modal data in an embodiment of the present application;

[0019] Figure 5 is a flowchart of step S14 in the abnormal identification method of the air blower based on multi-modal data in an embodiment of the present application;

[0020] Figure 6 is a flowchart of step S15 in the abnormal identification method of the air blower based on multi-modal data in an embodiment of the present application;

[0021] Figure 7 is a structural composition diagram of an abnormal identification system of the air blower based on multi-modal data in an embodiment of the present application. DETAILED DESCRIPTION

[0022] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0023] Please refer to Figures 1 to 7 An abnormal identification method of an air blower based on multi-modal data is applied to an abnormal identification scene; the abnormal identification method of the air blower based on multi-modal data comprises:

[0024] Step S11: determining a real-time image of the air blower based on the visual detection of the air blower when the air blower is in operation, determining a plurality of component regions based on the identification of the real-time image of the air blower, and determining a corresponding vibration data combination according to the vibration detection of each component region;

[0025] Step S12: determining a multi-modal framework of the air blower based on the region position of each component region and the corresponding vibration data combination, and constructing multi-modal data corresponding to the air blower according to the multi-modal framework of the air blower, noise data and temperature data of the air blower;

[0026] Step S13: determining a plurality of abnormal data according to the detection of the multi-modal data corresponding to the air blower, marking an abnormal component corresponding to each abnormal data, determining an abnormal recognition system according to the plurality of abnormal data, the abnormal component corresponding to each abnormal data and the overall form of the air blower, and marking a plurality of abnormal regions of different abnormal influence levels;

[0027] Step S14: in each abnormal region, determining a corresponding abnormal event according to the dynamic detection of the abnormal region, determining a plurality of abnormal influence factors based on each abnormal event, real-time working data of the air blower and the abnormal recognition system, and forming a corresponding abnormal influence route;

[0028] Step S15: determining a plurality of abnormal influence nodes based on the identification of the abnormal influence route, determining the maintenance content according to the node position of each abnormal influence node, the corresponding node form and the use scene of the air blower, determining the corresponding abnormal maintenance measures based on the maintenance content, the current working mode of the air blower and the corresponding maintenance system.

[0029] Reference Figure 2 In step S11, the specific steps are as follows:

[0030] S111: marking the position of the air blower, collecting a plurality of working data of the air blower, determining the working state of the air blower according to the plurality of working data of the air blower and the corresponding overall form, controlling the working state of the air blower, triggering the positioning shooting of a plurality of cameras around the position of the air blower to visually detect the air blower, and collecting a plurality of sub-images of the air blower in different dimensions, and constructing a real-time image of the air blower according to the synthesis of the plurality of sub-images;

[0031] S112: dynamically identifying the real-time image of the air blower, determining a plurality of component markers in the identification process, determining the corresponding component region according to the tracing of each component marker, triggering the vibration detection of each component region, determining a plurality of vibration data according to the vibration detection of each component region, and determining the corresponding vibration data combination based on the plurality of vibration data, the vibration detection position and the corresponding component region.

[0032] In the embodiments of the present application, the precise coordinates of the air blower in the physical space are determined, and the system collects multi-dimensional working data (such as current, voltage, speed, flow, etc.) of the air blower in real time through an industrial Internet of Things interface; at the same time, the system carries out preliminary digital modeling based on the spatial geometric form of the air blower, and maps these real-time working data to the device model, so as to comprehensively judge whether the air blower is currently in a steady state, a transient state (such as a start-stop process) or an overload state.

[0033] Based on the determined spatial position of the air blower, the system triggers the visual acquisition terminal deployed in the periphery. Here, "triggering" is not simply shooting, but active visual guidance based on spatial geometric relationship. The system calculates the optimal camera combination that can cover the key appearance features of the air blower according to the volume, height of the air blower and the field of view (FOV) of the surrounding cameras. These cameras perform synchronous or asynchronous positioning shooting of the air blower from different dimensions (such as front, side, top view, etc.). The acquisition process involves optical techniques such as optical flow control and focal length locking to obtain clear and unobstructed raw sub-image data.

[0034] The acquired multi-dimensional sub-images are transmitted to the processing unit. Since these sub-images come from different angles and time points, the system uses feature matching algorithms in computer vision (such as SIFT, SURF or feature extraction based on deep learning) to extract key feature points in each sub-image, and through the calculation of homography matrix or three-dimensional reconstruction, the sub-images are geometrically corrected and spliced. A high-resolution, seamless spliced real-time panoramic image containing complete appearance information of the air blower is constructed, which will serve as the basis for subsequent component region identification and vibration data spatial mapping.

[0035] Specifically, suppose there is an air blower deployed in an industrial workshop. The device runs with relatively large vibration, and there is an obstruction in the surrounding environment. The system locks the air blower at position 3 in workshop B area through the factory digital twin platform. At the same time, the SCADA system collects real-time data of the air blower: the driving end current is 45A, the outlet pressure is 0.25MPa, and the speed is 2950r / min. The system determines that the air blower is currently in a "high load stable running" state according to these parameters and the overall running mode of the air blower. In order to ensure that the visual detection can capture the slight deformation or features under high load, the system adjusts the priority of visual acquisition to the highest and locks the current timestamp as T0.

[0036] Based on the coordinates of the blower at position No. 3 in area B, the system calculates the spatial coverage of surrounding cameras C1 (located in front of the left side of the device), C2 (located on the right side of the device), and C3 (located at the tail of the motor); the system determines that a single camera cannot simultaneously capture the details of the impeller end and the bearing seat, and therefore triggers the automatic positioning and shooting functions of the three cameras simultaneously; camera C1 adjusts the focal length to focus on the air inlet box of the blower, camera C2 focuses on the exhaust side bearing seat, and camera C3 focuses on the coupling and motor part; the system obtains three sub-images of different dimensions: Img_C1 (intake view), Img_C2 (exhaust bearing view), and Img_C3 (drive side view).

[0037] After receiving Img_C1, Img_C2, and Img_C3, the system uses a feature extraction algorithm to identify common features in each view (such as the edges of the device base, paint texture, connecting bolts, etc.), and through image registration algorithm, aligns the left side of Img_C1 with the right side of Img_C3 in the overlapping area, and simultaneously integrates the perspective projection corrected Img_C2 into the overall structure; the system generates a real-time panoramic composite image of the blower at T0; in this image, the left side clearly shows the intake state, the middle shows the coupling connection, and the right side presents the details of the exhaust bearing seat, and all spatial coordinates are unified.

[0038] Further, the real-time image of the blower is dynamically identified, and multiple component markers are determined during the identification process, the corresponding component areas are determined according to the tracing of each component marker, to trigger vibration detection of each component area, and at the same time, multiple vibration data are determined according to the vibration detection of each component area, based on the multiple vibration data, the corresponding vibration detection positions, and the corresponding component areas, the corresponding vibration data combinations are determined, and the vibration data combinations are introduced.

[0039] At this time, the real-time panoramic image constructed in step S111 is analyzed frame by frame using a deep learning target detection algorithm (such as YOLO, MaskR-CNN, or a variant network based on instance segmentation); since the device is in working condition, the image contains blurring or slight positional offset caused by movement, therefore the algorithm needs to have dynamic robustness; the system identifies the key independent components of the blower (such as impeller, bearing seat, motor shell, base, etc.) in the image, and uses bounding boxes or semantic masks to accurately mark these components; each marker is assigned a unique ID, forming a component marker set.

[0040] According to the generated component label, the system maps back and forth between the image coordinate system and the physical world coordinate system, which converts the two-dimensional pixel area in the image into the component area coordinates in the three-dimensional physical space by using the preset camera calibration parameters (internal and external parameters); the system then queries the sensor list associated with the physical area, triggering the corresponding vibration data acquisition terminal. Here, "trigger" means sending an acquisition instruction to a specific vibration sensor (such as a piezoelectric accelerometer) to require it to perform high sampling rate synchronous sampling based on the current component's working condition characteristics (such as rotational speed frequency multiplication).

[0041] The system receives raw vibration signals (usually time-domain waveforms) from different sensors; in order to correspond these discrete data points with the visual component area, the system pre-processes the data (such as denoising, filtering), and binds the metadata of the data packet (such as sensor ID, acquisition timestamp) with the physical location information; the system encapsulates the vibration characteristics (such as effective value RMS, peak factor, kurtosis, etc.) belonging to the same component area with the spatial coordinates of the component. This encapsulation is not just a list of values, but forms a "vibration data combination" with spatial attributes.

[0042] Specifically, the system has obtained the real-time panoramic composite image of the blower; the system runs a target detection algorithm on the real-time panoramic image of the blower; the algorithm quickly identifies three key independent objects in the image: motor drive end, blower impeller shell, and gearbox bearing seat; the system generates three bounding boxes on the image and assigns label IDs: Part-01 (motor drive end), Part-02 (blower impeller shell), and Part-03 (gearbox bearing seat); even in the case of slight motion blur at the edges of the image due to high-speed operation of the blower, the algorithm still locks the accurate contours of these components through edge feature learning.

[0043] According to the pixel coordinates of Part-03 (gearbox bearing seat) in the image, the system calculates its position in the physical space as the east side of the device, 1.2 meters high, combined with the camera calibration matrix; the system traces the database and finds that the physical location is pre-installed with a three-axis vibration accelerometer with sensor number VS-003; the system sends a trigger instruction to VS-003, requiring it to capture data in the high frequency band (such as 0-10 kHz) to capture high-frequency fault signals generated by gear meshing.

[0044] The VS-003 sensor returns raw time-domain waveform data; the system performs fast Fourier transform (FFT) analysis on the data to extract characteristic values (for example, horizontal direction vibration speed is 4.5 mm / s, vertical direction is 3.2 mm / s, and an obvious sideband appears at the gear meshing frequency); the system strongly binds these vibration characteristics with the physical location information (gearbox on the east side of the blower) and image ID of Part-03; the final generated “vibration data combination” not only contains the above-mentioned numerical values, but also contains spatial semantics: “the bearing seat area of the blower gearbox shows high-frequency vibration characteristics”.

[0045] Reference Figure 3 In step S12, the specific steps are:

[0046] S121: In the plurality of component regions, the corresponding region positions are determined based on the detection of each component region, and the multi-modal framework of the blower is determined according to the region position of each component region, the corresponding region priority, and the corresponding vibration data combination;

[0047] S122: A plurality of environment data are determined based on the detection of the location where the blower is located; noise data and temperature data of the blower are determined according to the screening of the plurality of environment data, and the use scene of the blower is marked; the preliminary data are determined according to the multi-modal framework of the blower and the use scene of the blower, and the multi-modal data corresponding to the blower are constructed based on the preliminary data, the noise data of the blower, and the temperature data of the blower.

[0048] In the embodiment of the present application, the system uses image processing technology to spatially analyze each component mark identified in step S112, and through pre-calibrated camera internal parameters (focal length, principal point) and external parameters (rotation matrix, translation vector), combined with binocular vision or structured light depth estimation technology, converts the pixel coordinates of the components in the two-dimensional image into three-dimensional space coordinates in the physical world. This process establishes the absolute position and relative position relationship of each component region relative to the reference point (such as the center of the base) of the blower, forming a spatial point set describing the geometric structure of the device.

[0049] On the basis of determining the spatial position, the system calculates the priority according to the physical properties, functional importance, and fault sensitivity of the components. This process usually uses the analytic hierarchy process (AHP) or rule-based expert system logic; the system gives different components different weight levels, for example: high-speed rotating components and bearing seats that bear core loads are usually given the highest priority; auxiliary connecting parts or housings are given lower priority; in addition, the system also dynamically adjusts the priority in combination with real-time working conditions (such as sudden speed changes), to ensure that key components occupy a dominant position in the allocation of computing resources or abnormality determination.

[0050] The system constructs a weighted topological network structure with the above determined "regional position" as the node, "regional priority" as the attribute weight of the node, and "vibration data combination" as the load of the node; in this structure, each component region is no longer an isolated data point, but is related to each other through the geometric adjacency relationship of the mechanical transmission chain; the system strongly couples the time-domain vibration characteristics with the spatial coordinates to generate a multi-modal framework; the multi-modal framework not only contains the current vibration state, but also implicitly contains the three-dimensional geometric structure and fault importance information of the equipment.

[0051] Specifically, the system has identified three key components of the blower through visual recognition. The system analyzes the real-time image of the blower and identifies the "gearbox high-speed shaft bearing seat" region; using the pre-calibrated visual matrix, the system calculates the coordinates of the center point of this component in the physical space as (X: 1.25m, Y: 0.80m, Z: 1.50m), with an offset of (+0.4m, 0m, +1.2m) relative to the reference point of the blower base. Similarly, the system determines the spatial coordinates of the "motor driving end" as (X: 0.5m, Y: 0.80m, Z: 0.5m). These accurate three-dimensional coordinates are locked as spatial anchor points for subsequent data mapping.

[0052] The system evaluates according to the equipment design parameters and fault model library; it determines that the "gearbox high-speed shaft bearing seat" is responsible for the core link of power transmission, has the highest linear speed, and has the highest risk of failure, so its priority is set to LevelP0 (key core level); while the "motor driving end" is important, but the load is relatively stable under this working condition, so its priority is set to LevelP1 (important level); at the same time, considering that it is currently in a high-speed working condition, the system temporarily increases the monitoring weight coefficient of rotating components by 20%.

[0053] The system begins to construct the framework; taking the coordinates (1.25m, 0.80m, 1.50m) as the node center, the vibration data combination (including time-domain waveform, frequency-domain spectrum, and kurtosis index) of the "gearbox high-speed shaft bearing seat" is loaded into the node, and the LevelP0 weight label is added; at the same time, the topological connection relationship between this node and the "motor driving end" node is established, indicating the coupling between the two in mechanical transmission; the system generates the multi-modal framework of the blower: a weighted point cloud network distributed in three-dimensional space with priority attributes and encapsulating real-time vibration data. This framework clearly tells the system that at the highest priority position (1.25m, 0.80m, 1.50m), it is currently bearing a specific vibration energy characteristic.

[0054] Further, the multiple environment data are determined based on detection of the position of the air blower; the noise data and the temperature data of the air blower are determined according to screening of the multiple environment data, and the use scene of the air blower is marked, the primary data are determined according to the multi-modal framework of the air blower and the use scene of the air blower, and the multi-modal data corresponding to the air blower are constructed based on the primary data, the noise data and the temperature data of the air blower, which is compatible with the overall consideration of the multi-modal framework of the air blower and the use scene of the air blower, guarantees the accuracy of the primary data, and meanwhile, the multi-modal framework of the air blower is introduced, which is compatible with the consideration of the multi-modal framework of the air blower, the noise data and the temperature data of the air blower, and improves the accuracy of the multi-modal data corresponding to the air blower.

[0055] At this time, the system continuously collects original environment signals through an industrial sensor array (such as a sound pressure sensor array, a non-contact infrared thermal imager, and an environment temperature and humidity transmitter) deployed around the air blower; in order to eliminate the interference of background noise, the system uses signal processing technology for screening: for sound data, beamforming or spectrum analysis technology is used to filter out background noise in directions other than the device (such as other operation sounds in the workshop), and to accurately extract specific frequency band noise data from the air blower body; for temperature data, the device thermal field distribution model is combined to segment the region of interest (ROI) from the global thermal map of the thermal imager, eliminate the influence of environmental radiation and sunlight radiation, and lock the temperature data reflecting the true thermal state of the component.

[0056] The system extracts pure physical quantities while combining the DCS (Distributed Control System) operation parameters of the air blower for scene reasoning; the system labels the current running state as a specific use scene (such as: “heavy load full load running”, “empty load start”, “surge boundary running”, etc.) according to the current load rate, medium flow, valve opening degree, and control instructions (such as frequency converter frequency) through clustering algorithm or rule matching; the system uses the scene label to correct the multi-modal framework constructed in step S121; for example, in the “heavy load” scene, the system dynamically adjusts the baseline threshold of the vibration data, and eliminates the normal physical quantity changes caused by the increase of load, to generate “primary data” filtered by scene semantics.

[0057] The system maps the filtered noise data and temperature data to the primary data skeleton after scene correction, which involves strict data alignment: in the time dimension, the instantaneous noise peak and the temperature change value are synchronized to the same sampling time slice of the vibration data based on the unified timestamp (such as the PTP protocol); in the spatial dimension, the noise signal is attributed to the specific component node in the multi-modal framework using sound source positioning technology, and the temperature heat map data is pixel-level registered with the geometric area of the component; the system finally constructs multi-modal data containing spatial position, vibration characteristics, thermodynamic characteristics, acoustic characteristics and scene semantics, providing all-round input for subsequent deep anomaly recognition.

[0058] Specifically, the blower is located in a noisy industrial workshop, and the system collects mixed sound wave signals in the workshop through a microphone array; using an adaptive filtering algorithm, the system filters out the frequency of 2 kHz of the cutting machine noise in the background, and accurately extracts the blade pass frequency (BPF) noise data specific to the blower, with a sound pressure level of 88 dB; at the same time, the infrared thermal imager scans the surface of the equipment, and the system excludes the thermal radiation interference of the surrounding high-temperature pipeline through a heat map segmentation algorithm, and locks the real temperature data of the surface of the "gearbox bearing seat" of the blower, which is displayed as 72℃.

[0059] The system reads the real-time running parameters of the blower: the outlet valve opening is 100%, and the motor current reaches 95% of the rated value; based on this, the system marks the current running state as the "full load steady state running" scene; the system calls the standard model under this scene, and determines that the specific high-frequency component in the vibration data belongs to the normal physical response under full load; therefore, the system corrects the multi-modal framework in step S121, eliminates the structural vibration false alarm factor caused by load increase, and generates "primary data" reflecting the true health status of the equipment under full load.

[0060] The system injects the above noise data (88 dB) and temperature data (72℃) into the primary data; in space, the temperature value of 72℃ is bound to the "gearbox bearing seat" node with coordinates (X: 1.25m, Y: 0.80m, Z: 1.50m) in the multi-modal framework; in time, the noise value of 88 dB is timestamped and aligned with the current vibration acceleration RMS value (4.5mm / s) of the node. The system constructs the multi-modal data object of the blower: {Node: Gearbox Bearing Seat | Position: (1.25, 0.80, 1.50) | Priority: P0 | Vibration: 4.5mm / s | Temperature: 72℃ | Noise: 88dB | Scenario: Full Load Steady State}.

[0061] Reference Figure 4In step S13, the specific steps are:

[0062] S131: dynamically detecting the multi-modal data corresponding to the air blower, determining a plurality of data anomaly markers in the detection process, determining corresponding abnormal data according to the traceability of each data anomaly marker, and collecting a plurality of abnormal data;

[0063] S132: determining the corresponding abnormal components based on the detection of each abnormal data, determining the first abnormal identification content according to the overall form of the air blower and the abnormal components corresponding to each abnormal data, determining the second abnormal identification content according to the current working mode of the air blower and each abnormal data, and determining the abnormal identification system according to the first abnormal identification content and the second abnormal identification content;

[0064] S133: in the abnormal identification system, determining a plurality of abnormal areas based on the identification of the abnormal identification system, and determining the corresponding abnormal influence level according to the data combination of each abnormal area, the corresponding area position and the corresponding component function.

[0065] In the embodiment of the present application, the system uses a time series sliding window algorithm to scan the multi-modal data stream frame by frame; in this link, the system does not monitor a single dimension of value, but performs dynamic detection based on a multi-parameter fusion strategy; the system aligns the vibration frequency spectrum features, temperature change trends and noise sound pressure levels in the same time window at the feature level; the machine learning model (such as an automatic encoder Autoencoder or a long short-term memory network LSTM) is used to calculate the “reconstruction error” or “residual” of the current input data and the standard normal model under the working condition; when the residual exceeds the dynamically set confidence threshold, or it is detected that the parameters violate the physical coupling law (for example, the vibration energy surges but the acoustic energy does not increase accordingly), the system determines that an abnormal state is detected, and generates a “data anomaly marker”.

[0066] After generating the data anomaly marker, the system uses the accurate timestamp and spatial index information contained in the marker to perform reverse retrieval in the high-speed data cache or the historical database; the traceability process aims to reconstruct the complete context of the abnormality: the system not only extracts the peak data at the time of the abnormality, but also extracts the “time sequence slice” (such as data from T-5 seconds to T+5 seconds) before and after the abnormality, in order to capture the transient characteristics of the fault evolution; at the same time, according to the spatial index, the system locates which node (such as which component area) in the multi-modal framework the abnormality originates from, so as to bind the marker to a specific physical entity.

[0067] According to the traceability result, the system peels off the relevant data segment from the multi-modal data stream and performs encapsulation collection. This process not only intercepts numerical values, but also extracts high-level feature vectors (such as kurtosis, factor, spectral barycenter, etc.) of the data segment; the system packages these original waveforms, feature parameters, and corresponding metadata (device ID, component ID, working condition label) into structured "abnormal data" objects. This step ensures that subsequent fault diagnosis analysis is based on complete, high-quality data blocks with context information, rather than isolated single-point alarm values.

[0068] Specifically, the blower is in the "full load steady state operation" mode, and the system is monitoring its real-time multi-modal data stream, focusing on detecting the condition of the gearbox area; the system scans the multi-modal data of the blower successively with a time window of 1 second; at T=10:05:32, the system finds an anomaly while processing the data of the "gearbox high-speed shaft bearing seat" node: an obvious asymmetric impact appears in the time-domain waveform of the vibration signal, and the vibration acceleration effective value (RMS) exceeds the dynamic threshold; at the same time, the spectral analysis of the noise data shows that there is a sideband with a spacing of the rotation frequency near the gear meshing frequency (500 Hz); although the temperature data only rises slightly at this time, the characteristics of "vibration impact" and "sideband" are highly coupled, which is consistent with the typical characteristics of bearing cage damage; the system determines that this is a significant abnormal event, and immediately generates a "data anomaly tag Tag-01" with the timestamp 10:05:32 and the node ID (Gearbox_HighSpeed).

[0069] According to the timestamp 10:05:32 and the node ID in "Tag-01", the system immediately performs traceability in the data cache; the system locates the data segment before and after this time, and finds that from 10:05:28, a weak low-frequency modulation component has appeared in the vibration signal, and by 10:05:32, this component has evolved into a severe impact; the system confirms that the abnormal source is the "gearbox high-speed shaft bearing seat" at the coordinates (1.25m, 0.80m, 1.50m) in the multi-modal framework.

[0070] The system packages all related data within the 10-second time window from 10:05:28 to 10:05:37; the collected content includes: the original waveform data of three-axis vibration acceleration, the corresponding FFT spectrum, the time series of noise sound pressure level, and the temperature sampling value at this time; the system encapsulates these data as a standard "abnormal data" object, labeled as {EventID:Tag-01, Source:Gearbox_HighSpeed, Waveform:[...], Features:{Kurtosis:5.2, Sideband_Level:High}}.

[0071] Further, based on the detection of each abnormal data, the corresponding abnormal component is determined, the first heavy abnormal identification content is determined according to the abnormal component corresponding to each abnormal data and the overall form of the blower, the second heavy abnormal identification content is determined according to each abnormal data and the current working mode of the blower, and the abnormal identification system is determined according to the first heavy abnormal identification content and the second heavy abnormal identification content. The overall consideration of the compatibility of the first heavy abnormal identification content and the second heavy abnormal identification content ensures the accuracy of the abnormal identification system.

[0072] At this time, the system deeply analyzes the collected abnormal data and locks the fault source. By analyzing the spatial coordinates of the abnormal data in the multi-modal framework, the system accurately maps it to a specific physical component. This process relies on the matching of the fault feature library: for example, a specific impact frequency is mapped to a bearing, a specific airflow noise is mapped to an impeller, and a specific temperature distribution is mapped to a winding or a lubrication system; the system finally determines the unique entity component that produces these abnormal data and outputs the "abnormal component ID".

[0073] After locking the abnormal component, the system performs first heavy abnormal identification in combination with the overall mechanical form and topological structure of the blower. This dimension focuses on the physical conduction and structural influence of the fault; the system calculates the position of the abnormal component in the overall form (such as whether it is a cantilever end or a core of the transmission chain), and analyzes whether its vibration mode is coupled with the surrounding components to resonate; for example, if the abnormal component is located at the base, the system will determine whether its vibration is being transmitted upwards to the motor end through the rack.

[0074] For the second heavy abnormal identification, the system analyzes the abnormal data in association with the current working mode of the blower (such as constant pressure control, variable frequency speed regulation, bypass adjustment, etc.). This dimension focuses on the authenticity of the fault, i.e., distinguishing whether it is a device intrinsic fault or a normal response caused by external working condition changes; the system uses mechanism models to determine whether the current abnormal parameter change violates physical laws under the current mode; for example, pressure fluctuations are normal during flow regulation, but if accompanied by specific high-frequency vibration, it is abnormal.

[0075] The system uses D-S evidence theory or Bayesian inference network to fuse the identification results of the first abnormality recognition (structure dimension) and the second abnormality recognition (working condition dimension); if the first abnormality recognition shows "serious structural resonance" and the second abnormality recognition shows "stable working condition not caused by adjustment", the system determines that it is a "serious structural failure"; if the first abnormality recognition shows "local slight abnormality" and the second abnormality recognition shows "in the process of rapid variable load adjustment", the system determines that it is a "working condition induced temporary abnormality". Through this cross verification, the system constructs an abnormality recognition system with high robustness. The abnormality recognition system is a comprehensive system that fuses the structure dimension and the working condition dimension of the blower to realize fault grading.

[0076] Specifically, the system has collected abnormal data at time T, and now needs to confirm the fault nature through the dual recognition system; the current blower is in "constant pressure variable frequency operation" mode, and the frequency is set to 45Hz; the system analyzes the abnormal data and finds that the dominant feature is broadband vibration (1X-3X rotation frequency) of medium and low frequency, and there is obvious airflow turbulence sound in the noise data; according to the spatial index (X: 1.25m, Y: 0.80m, Z: 1.50m) of the multi-modal framework and the feature library matching, the system excludes the motor and the gear box, and determines that the corresponding abnormal component is the volute assembly of the blower.

[0077] The system analyzes the overall morphology of the blower: the volute is fixed on the base by four bolts; the first abnormality recognition shows that the typical structural resonance frequency of the base does not appear in the frequency spectrum of the abnormal data, indicating that the vibration mainly comes from the fluid pulsation inside the volute or the insufficient stiffness of the volute itself, and there is no strong path to conduct to the base; the identification content conclusion is: "the abnormality is limited to the volute body, and has not caused the resonance conduction of the overall structure, but there is a risk of loose connection under the current morphology".

[0078] The system checks the working log, and the frequency of the frequency converter of the blower quickly rises from 40Hz to 45Hz (i.e. in the rising step response stage) from T-5 seconds to T when the abnormality occurs; according to the principle of fluid mechanics, the volute will indeed excite transient airflow excitation during rapid change of rotation speed; the second abnormality recognition content conclusion is: "the abnormal data characteristics are consistent with 'fluid transient response under rapid variable load working condition', which is an expected physical phenomenon under this working mode, not equipment damage".

[0079] The system integrates the above two contents: input A (first abnormality recognition): local vibration, but not spread, medium risk; input B (second abnormality recognition): clear working condition change inducement, non-ontology fault; fusion conclusion: the abnormality recognition system qualitatively determines this event as "working condition induced transient fluctuation", and sets the level as "attention"; the system automatically suppresses the serious alarm for "volute fault", but generates a preventive maintenance suggestion for "fatigue of connecting bolt under frequent load change", thereby realizing accurate identification.

[0080] Therefore, in the abnormality recognition system, a plurality of abnormal regions are determined based on the identification of the abnormality recognition system, and the corresponding abnormal influence level is determined according to the data combination, corresponding region position and corresponding component function of each abnormal region, the overall consideration of the data combination, corresponding region position and corresponding component function of each abnormal region is compatible, the accuracy of the corresponding abnormal influence level is ensured, at the same time, each abnormal data is further controlled, and the corresponding abnormal component is marked, the overall consideration of the plurality of abnormal data, the abnormal component corresponding to each abnormal data and the overall form of the blower is realized, and the accuracy of the abnormality recognition system is improved.

[0081] At this time, the system uses the abnormality recognition system constructed to accurately divide the physical range of fault influence, and this process is no longer a point positioning of a single component, but a regional mapping based on data correlation; the system analyzes the propagation path of abnormal data in a multi-modal framework, and identifies the data cluster affected by the fault through a graph theory algorithm (such as connected component analysis); the system circumscribes the components that are adjacent in space position, highly correlated in feature parameters and affected by the same fault source as one or more "abnormal regions"; for example, the failure of a bearing not only affects the bearing seat itself, but also affects the adjacent base through rigid connection, and the three are combined and defined as an abnormal region in space and logic.

[0082] For each determined abnormal region, the system uses a multi-factor weighting model to calculate its "abnormal influence level"; the model mainly quantifies and evaluates the following three core dimensions:

[0083] Deviation degree of data combination: calculate the deviation multiple and change trend of vibration RMS value, temperature gradient, sound pressure level and other data in the region relative to the baseline (ISO10816 standard or device historical baseline), and represent the intensity of the fault;

[0084] Key of region position: according to the topological structure, the position weight of the region in the device is evaluated; the core transmission chain region is usually higher than the edge support region, and the high pressure end close to the outlet is usually higher than the low pressure end close to the inlet;

[0085] Sensitivity of component function: assess the impact of the function carried by this area on the overall operation of the system; components related to power transmission, core sealing or rotor balance have extremely high sensitivity of function.

[0086] The system performs a weighted operation on the above three dimensions based on the analytic hierarchy process (AHP) or fuzzy logic, and finally outputs a quantitative abnormal influence level (usually divided into: I level - emergency / fatal, II level - important / serious, III level - general / warning).

[0087] Specifically, the blower is identified as having two different fault signs: one is the sudden impact of the gearbox high-speed shaft, and the other is the slight blockage of the fan inlet filter screen; now the influence level of the abnormal area defined by the two fault sources needs to be assessed.

[0088] Region one definition: the system detects high-frequency impact signals in the gearbox high-speed shaft bearing seat (coordinates 1.25m, 0.80m, 1.50m), and the signal characteristics show strong correlation with the adjacent coupling region (coordinates 1.10m, 0.80m, 1.50m); the identification system determines that these two physical positions have formed a force-connected whole, and encircles them as an abnormal area, named "power transmission core area".

[0089] Region two definition: the system detects weak fluctuations in negative pressure data at the inlet pipe, and detects low-frequency airflow sound (less than 200Hz) in the noise data; this signal does not spread to the inside of the body, but is limited to the vicinity of the air inlet; the system independently encircles it as another abnormal area, named "inlet airflow auxiliary area".

[0090] Evaluation of "power transmission core area": data combination: the peak value of vibration acceleration reaches 12mm / s (ISO standard warning value is 7.1mm / s), and is accompanied by a sharp rise in temperature (+2℃ per minute), with a high degree of deviation; area location: located between the motor and the load, at the most critical position in the transmission chain; component function: responsible for high-speed torque transmission, which will directly cause the blower to stop running, and even cause the blade to fly and cause a vicious accident; comprehensive evaluation: the system calculates the comprehensive risk index > 0.9, and determines the abnormal influence level of this area as Level I (emergency shutdown level).

[0091] Evaluation of "intake air flow auxiliary area": data combination: no abnormal vibration, only noise value increased by 5dB, negative pressure fluctuation within 3%, low deviation; Area location: located on the non-load-bearing structure at the edge of the device; Component function: only responsible for air filtration and flow guide, functional failure only leads to efficiency decline, will not immediately damage the device; Comprehensive evaluation: the system calculates the comprehensive risk index <0.3, and determines the abnormal influence level of the area as Level III (planned maintenance level); The system visually displays the two areas on the human-machine interface (HMI) of the blower: the gear box area is displayed in red (Level I), and the intake area is displayed in yellow (Level III).

[0092] Reference Figure 5 In step S14, the specific steps are:

[0093] S141: Real-time monitoring of each abnormal area, and marking of the abnormal time period of each abnormal area, dynamic identification of the abnormal time period of each abnormal area, and determination of a plurality of working contents of the abnormal area in the abnormal time period in the identification process, determination of a corresponding abnormal event according to the area form of the abnormal area, the corresponding plurality of working contents and the corresponding component priority;

[0094] S142: Collecting real-time working data of the blower, determining a first major influence content according to each abnormal event and the real-time working data of the blower, determining a second major influence content according to the real-time working data of the blower and the abnormal identification system, and determining a plurality of abnormal influence factors based on the first major influence content and the second major influence content;

[0095] S143: Marking of the influence components of each abnormal influence factor, determination of a plurality of sub-influence routes according to each influence component, the corresponding abnormal influence factor and the current working mode of the blower, and formation of a corresponding abnormal influence route based on the plurality of sub-influence routes, the current working route of the blower and the corresponding data transmission route.

[0096] In the embodiment of the present application, each abnormal area is monitored in real time, and the abnormal time period of each abnormal area is marked. The abnormal time period of each abnormal area is dynamically identified, and a plurality of working contents of the abnormal area in the abnormal time period are determined in the identification process. The corresponding abnormal event is determined according to the area form of the abnormal area, the corresponding plurality of working contents and the corresponding component priority, thereby improving the accuracy of the abnormal event.

[0097] At this time, the system starts high-frequency sampling monitoring flow on the locked abnormal area, and uses state machine model or change point detection algorithm to accurately define the starting time and ending time of the abnormal occurrence. This process not only records a single time point, but also marks a continuous "abnormal time period"; within this time window, the system performs stream processing on multi-modal data to capture the dynamic evolution trend of signal characteristics; for example, distinguish whether the abnormality is a "transient pulse" (such as a transient impact) or a "persistent drift" (such as a slow temperature rise); the system uses these marked time periods as the basis for subsequent behavior analysis.

[0098] The system uses pattern recognition algorithms to analyze the dynamic behavior characteristics of the area and converts them into specific "work content". Here, "work content" does not refer to the normal operation of the device, but to the physical behavior patterns exhibited by the faulty component at the microscopic level; the system extracts feature sequences through time-frequency analysis (such as short-time Fourier transform STFT or wavelet transform) to identify specific behavior patterns, such as: "sub-synchronous frequency components appear and gradually increase" (showing friction behavior), "amplitude shows quadratic curve growth with rotational speed" (showing imbalance behavior), "high-frequency envelope demodulation spectrum densely emerges" (showing peeling behavior), these behavior descriptions constitute the "work content" of the abnormal area.

[0099] The system semantically fuses the identified "work content" with the "area morphology" and "component priority" of the area to define the final "abnormal event"; assess the impact of the physical structure on the behavior; for example, "cantilever structure" amplifies vibration amplitude, "closed cavity" causes noise standing wave; combine P0, P1 levels to qualitatively describe the nature of the event; minor behaviors of high-priority components are defined as serious events; the system maps [work content + area morphology + component priority] to standardized abnormal event labels through a rule engine or expert system; for example, "high-frequency impact behavior" + "high-speed rotating morphology" + "P0-level component" is defined as "high-speed shaft bearing catastrophic failure precursor event".

[0100] Specifically, the system has marked the "gearbox high-speed shaft area" of the blower as an abnormal area (Level I), and the system maintains real-time monitoring of the "gearbox high-speed shaft area" at a sampling rate of 20k per second; at 10:15:30, the vibration RMS value exceeds the threshold, and the system starts the timer; in subsequent monitoring, the signal does not fall back, and is accompanied by intermittent pulse peaks, until 10:17:00 the system triggers safety logic shutdown; the system accurately marks the abnormal time period of the area as [10:15:30, 10:17:00], with a duration of 90 seconds; the system extracts all vibration waveforms, temperature curves, and audio clips within this time period.

[0101] During this 90-second abnormal period, the system performs dynamic time-frequency analysis on the collected data: it identifies that there is a clear modulation phenomenon in the vibration signal, that is, amplitude modulation with the rotation frequency as the carrier and the fault frequency as the sideband; it identifies the non-stationary impact characteristics, the impact interval is relatively regular at the beginning and then becomes chaotic; it identifies the high-frequency energy release, and there is a continuous resonance demodulation spectrum in the 2000Hz-4000Hz range; based on the above analysis, the system determines the working content of this area during the abnormal period as: "irregular mechanical impact and rolling body slip friction behavior occurs".

[0102] The system further combines the attributes of this area to make a comprehensive judgment: area morphology: this area belongs to a high-speed heavy-duty gear transmission structure and is located at a rigidly connected transmission chain node, any small instability will be amplified by the structure; component priority: this component is a P0-level core component (high-speed shaft bearing), and failure will directly lead to shutdown; event generation: the system puts "irregular impact and slip behavior" into "high-speed heavy-duty form" and superimposes the high-risk weight of "P0 level"; the system determines the performance during this abnormal period as an abnormal event: "catastrophic friction event caused by high-speed bearing retainer fracture", which contains richer physical meaning and maintenance guidance value than the simple "vibration is large".

[0103] Further, real-time working data of the air blower is collected, the first major impact content is determined according to each abnormal event and the real-time working data of the air blower, the second major impact content is determined according to the real-time working data of the air blower and the abnormal identification system, and the multiple abnormal impact factors are determined based on the first major impact content and the second major impact content, which compatiblely considers the overall consideration of the first major impact content and the second major impact content, and ensures the accuracy of the multiple abnormal impact factors.

[0104] At this time, the system collects real-time working data of the air blower through an industrial bus at a high frequency, including current, voltage, inlet and outlet pressure, flow, valve opening, etc.; the system uses Pearson correlation coefficient or Granger causality test to perform point-by-point comparison between the "abnormal event" characteristics (such as vibration peak value, mutation time) determined in the previous steps and the time series of these real-time working data, which aims to find the synchronicity between the abnormal event and the external working condition change; for example, whether the outbreak of the abnormal event is synchronized with the sudden increase of the current; whether the change of the vibration frequency is consistent with the pace of the flow regulation; if there is strong synchronicity, the first major impact content will point to the "external working condition induced" factor; if the working data remains stable when the abnormal event occurs, the first major impact content will point to the "device internal spontaneous" factor.

[0105] The system inputs the collected real-time working data into the baseline model or digital twin in the "abnormality identification system"; the model stores the physical coupling relationship between the parameters of the equipment in the healthy state (such as the pressure-flow curve, the speed-vibration threshold curve).

[0106] The system calculates the residual error between the real-time data and the model output, and determines whether the current working state deviates from the design working condition or the health baseline by analyzing the residual error distribution; for example, whether the current power consumption is abnormally high under the same load; whether the vibration spectrum violates the specific area limit in the ISO10816 standard; the second major impact content mainly answers "whether the current working state triggers the failure mode defined by the system".

[0107] The system uses decision-level fusion logic (such as D-S evidence theory or fuzzy logic reasoning) to cross-verify and synthesize the first major impact content and the second major impact content; if the first major shows "stable working condition" and the second major shows "serious deviation from the health model", the determining factor is "component performance degradation"; if the first major shows "severe working condition fluctuation" and the second major shows "model deviation within the allowed range", the determining factor is "external process disturbance"; if both show abnormalities, the determining factor is "complex coupled failure", through this logic to exclude interference items (such as coincidental working condition fluctuations), the system finally outputs the "abnormal impact factor" that is reliable and has physical meaning.

[0108] Specifically, the blower experienced a "catastrophic friction event caused by high-speed bearing cage fracture" at 10:15:30; the system extracts the real-time working data in the abnormal time period [10:15:30, 10:17:00]; data analysis: shows that during this period, the motor current of the blower remains around 120A (rated value 125A), with a fluctuation range of less than ±1%; the outlet air pressure is maintained at 45kPa, with no obvious pressure pulse; the inlet valve opening is locked at 75% and does not move; associated conclusion: when the abnormal event occurs, the external load and process parameters are extremely stable, and there is no overload or fluid transient impact; first major impact content: determines that the abnormality is not related to the external working condition, and points to "device internal spontaneous mechanical failure".

[0109] The system compares the real-time vibration data with the "gearbox health baseline model" in the abnormality identification system; model operation: at the current speed, the normal vibration RMS threshold of the bearing seat should be 2.5mm / s; however, the actually collected data shows that the vibration value has soared to more than 8.0mm / s, and the high-frequency demodulation spectrum amplitude has exceeded the baseline by 400%; system determination: this serious deviation exceeds the "aging" or "wear" interval allowed by the model, and enters the "sudden fracture" characteristic region; second major impact content: determines that the current state triggers the "catastrophic structural failure mode" in the system.

[0110] The system makes a fusion decision on the two content: input A (first influence content): the external environment is stable, no inducement > exclude "misoperation" or "overload"; input B (second influence content): serious deviation from the health model, the characteristics are consistent with the fracture > confirm "internal failure"; comprehensive judgment: the system combines the running time of the bearing (which has reached the end of the design life) and the vibration characteristics (random impact of the cage fracture), and finally determines that the abnormal influence factor is: "cage structural fracture caused by high-speed shaft bearing fatigue life depletion", which provides a direct basis for subsequent development of maintenance scheme.

[0111] Therefore, the influence components of each abnormal influence factor are marked, a plurality of sub-influence routes are determined according to each influence component, the corresponding abnormal influence factor and the current working mode of the blower, and a corresponding abnormal influence route is formed based on the plurality of sub-influence routes, the current working route of the blower and the corresponding data transmission route, which is compatible with the overall consideration of each influence component, the corresponding abnormal influence factor and the current working mode of the blower, and ensures the accuracy of the plurality of sub-influence routes.

[0112] At this time, the system performs fault tree analysis (FTA) or topological mapping on the determined abnormal influence factor, and accurately locks the specific component where the fault occurs at the physical level, that is, the "influence component". This process not only marks the directly failed components (such as the source component), but also identifies secondary components that are directly associated or force-transmitted according to the mechanical structure of the equipment; for example, if the factor is "lubricating oil pollution", the influence component is not only the "oil pump", but also the "gear" and "bearing" that are lubricated; the system highlights these components in the topological graph as a node set for subsequent route construction.

[0113] The system analyzes the transmission logic of fault energy or abnormal signals between components in combination with the "current working mode" of the blower (such as variable frequency regulation mode, power frequency constant pressure mode, bypass return mode); for physical faults: analyze how mechanical vibration, thermal stress or fluid impact is transmitted from one component to another (for example: rotor imbalance > bearing > bearing seat > foundation); for working condition influence: analyze how the feedback loop of the control system is affected under a certain mode (for example: surge occurs > flow decreases > PLC issues an instruction to increase the frequency of the frequency converter > motor overload); the system defines these local and one-way transmission paths as "sub-influence routes", and each route represents an independent causal dimension.

[0114] The system spatiotemporally aligns and fuses the "sub-influence route" in the physical layer with the "data transmission route" in the information layer (sensor > gateway > cloud platform) and the "current working route" in the control layer (given value > controller > actuator); the system constructs a multi-dimensional directed graph, in which the nodes represent components or data points, and the edges represent influence relationships; the finally formed "abnormal influence route" serves as a panoramic multi-dimensional fault mapping graph, which not only shows where the fault is, but also more completely presents the whole process from the physical occurrence of the fault to the system response.

[0115] Specifically, the system has determined that the abnormal influencing factor of the blower is "high-speed shaft bearing retainer fracture"; the current working mode is "constant pressure variable frequency control mode"; the system performs fault tracing and propagation analysis in the digital twin model of the blower according to the influencing factor "retainer fracture"; direct marking: locking the core influencing component as "high-speed shaft bearing (component ID: B-002)"; correlation marking: analysis finds that the intense vibration caused by the fracture will be directly transmitted to "gearbox body (component ID: G-001)", and conducted to "motor driving end (component ID: M-001)" through the coupling; the system marks these three components as key influencing nodes of the current abnormal event.

[0116] The system deduces two key sub-influence routes in combination with the "constant pressure variable frequency control" mode: sub-influence route A (mechanical physical chain): high-speed shaft bearing (fracture occurs) > generates high-frequency impact > causes gearbox body to vibrate violently > transmits to motor shaft through coupling > induces abnormal vibration of motor rotor; sub-influence route B (working condition control chain): impeller generates dynamic imbalance due to bearing damage > causes air outlet pressure fluctuation > pressure sensor detects pressure drop > PID controller (in constant pressure mode) outputs an increase in frequency command > frequency converter increases motor speed > vibration is further intensified (positive feedback vicious circle).

[0117] The system fuses the above sub-routes with data flow to form a panoramic abnormal influence route: [physical propagation layer]: high-speed shaft bearing (B-002)—(vibration impact)—>gearbox (G-001)—(torque fluctuation)—>motor (M-001); [data perception layer]: vibration sensor (VS-03)—(high-frequency signal)—>edge gateway—(data packet)—>abnormal identification engine; [control response layer]: pressure sensor (PT-01)—(low pressure signal)—>PLC controller—(frequency increase command)—>frequency converter (Inv-01)—(high-frequency carrier)—>motor winding; [fusion conclusion]: the abnormal influence route graph clearly shows that the bearing fracture causes physical vibration, which on the one hand triggers an alarm (data chain) captured by the sensor, and on the other hand interferes with the constant pressure control, causing the system to incorrectly increase the frequency (control chain), ultimately forming a vicious closed loop of continuously expanding vibration.

[0118] Reference Figure 6 In step S15, the specific steps are as follows:

[0119] S151: In the abnormality-affected route, a plurality of abnormality-affected road segments are determined according to the identification of the abnormality-affected route, each abnormality-affected road segment corresponding to at least one abnormal component in the blower, an abnormality-affected node corresponding to each abnormality-affected road segment is determined according to the road segment form of each abnormality-affected road segment and the corresponding abnormal component, and a plurality of abnormality-affected nodes are collected;

[0120] S152: A node position of each abnormality-affected node is determined according to the detection of each abnormality-affected node, and a node form of each abnormality-affected node is marked, a to-be-maintained area is determined according to the node position of each abnormality-affected node and the use scene of the blower, and to-be-maintained content is determined according to the to-be-maintained area and the node form of each abnormality-affected node;

[0121] S153: A corresponding maintenance system is determined based on the traceability of a past maintenance event of the blower, a first re-maintenance coefficient is determined according to the maintenance system and the to-be-maintained content, a second re-maintenance coefficient is determined according to the maintenance system and the corresponding abnormal component, and a corresponding abnormal maintenance measure is determined based on a mapping relationship table of the first re-maintenance coefficient, the second re-maintenance coefficient, and the abnormal maintenance measure.

[0122] In the embodiment of the present application, the system performs logical segmentation of the constructed “abnormality-affected route” at the graph theory level; since the affected route usually covers a continuous process of energy flow, mechanical transmission or fluid transmission, the system cuts the entire route into a plurality of “abnormality-affected road segments” according to the discontinuity of physical connection or the logical level of signal transmission; each road segment represents a transmission interval of fault energy between two key components; the division principle follows the change of transmission medium or the change of connection mode, for example, from “rotor” to “bearing” as a segment (mechanical contact), and from “bearing seat” to “foundation” as another segment (rigid connection), in this way, the complex long-chain fault is decoupled into a plurality of easily managed local road segments, and each road segment is necessarily mapped to at least one specific “abnormal component” in the blower physical structure.

[0123] For each divided abnormal influence section, the system performs multi-dimensional feature analysis to determine the monitoring node; section morphology analysis: evaluate the physical connection characteristics of the section, such as rigid connection, flexible connection, or fluid coupling; rigidly connected sections usually need to monitor vibration transmission rate, while flexibly connected sections need to monitor displacement or centering conditions; abnormal component attribute analysis: combined with the material properties (such as metal, rubber) and motion properties (such as rotation, static) of the abnormal components corresponding to the section; the system integrates these two dimensions to determine the "abnormal influence node" on the geometric boundary or sensitive point of the section; the abnormal influence node is usually defined as the most critical physical location for maintenance operations, such as bolt fastening surface, sealing groove, shaft centerline, or sensor mounting base.

[0124] After determining all abnormal influence nodes, the system starts targeted data collection mechanism, which is no longer full-device general collection, but "focused collection" for nodes; the system dynamically calls adaptive sensor resources or adjusts sampling parameters (such as increasing sampling rate, turning on anti-aliasing filter) according to the physical environment (such as high temperature, high pressure, strong electromagnetic interference) where the node is located; the collected content includes the physical state parameters (displacement, stress, temperature) of the node, environmental parameters, and relative position data between nodes; the system encapsulates these data as the digital twin mirror image of the node, providing quantitative basis for subsequent precise maintenance.

[0125] Specifically, the "abnormal influence route" of the blower has been determined as: high-speed shaft bearing (fault source) > gearbox housing > coupling > motor shaft; the system divides the entire influence route into two main abnormal influence sections according to the mechanical connection relationship: section one: high-speed shaft bearing to gearbox lower housing, this section belongs to internal transmission structure, the corresponding abnormal components include "high-speed shaft bearing" and "gearbox lower housing"; section two: gearbox output shaft flange to motor shaft flange, this section belongs to external transmission connection, the corresponding abnormal components include "diaphragm coupling" and "motor input shaft".

[0126] For section one: the system analyzes its section morphology as "high-speed heavy-load closed rigid connection", and the abnormal component is a precision bearing; in order to capture internal damage impact, the system determines the "outer vertical radial surface of the gearbox high-speed shaft bearing seat" as the abnormal influence node N-01, which is the "key sensitive point" on the vibration transmission path and best reflects the internal state of the bearing; for section two: the system analyzes its section morphology as "flexible torque transmission", and the abnormal component includes the coupling; in order to evaluate the impact of the fault on transmission accuracy, the system determines the "horizontal centering reference surface of the coupling guard and the motor base" as the abnormal influence node N-02, which is the key position for detecting whether the coaxiality is invalid due to vibration.

[0127] Node N-01 acquisition: the system instructs the piezoelectric vibration sensor at this location to collect acceleration waveform at a high sampling rate of 20 kHz, focusing on capturing the high-frequency envelope signal of bearing failure, while collecting temperature data at this point; Node N-02 acquisition: the system uses a laser collimator or displacement sensor to collect the radial runout and axial displacement of this node, quantifying the deformation degree of section two; the system obtains node N-01 (high-frequency impact data) and node N-02 (deformation data).

[0128] Further, the corresponding node position is determined according to the detection of each abnormal influence node, and the node form of each abnormal influence node is marked, the to-be-maintained area is determined according to the node position of each abnormal influence node and the use scene of the blower, and the to-be-maintained content is determined according to the to-be-maintained area and the node form of each abnormal influence node, which is compatible with the overall consideration of the to-be-maintained area and the node form of each abnormal influence node, and ensures the accuracy of the to-be-maintained content.

[0129] At this time, the system maps each abstract abnormal influence node to the physical space coordinate system through the three-dimensional digital twin model of the device, determines its accurate "node position", and this process involves analyzing the installation coordinates, hierarchical relationship (such as being located at the bottom layer or the top layer) and the envelope boundary of the node; the system marks the node according to the geometric characteristics and physical properties of the components at the node; the form marking is not limited to a simple name, and the node form of the abnormal influence node includes form topological characteristics, such as: "rotary body fitting surface", "cantilever support structure", "flange connection interface" or "sealed fluid cavity". At this time, the node position of the abnormal influence node covers the corresponding installation coordinates, hierarchical relationship and envelope boundary of the surrounding.

[0130] The system introduces the "use scene of the blower" parameter, including personnel accessibility, operation safety distance, environmental interference factors (such as high temperature area, high pressure pipeline area) and disassembly and assembly logic (such as whether the upper cover needs to be lifted); the system uses a spatial clustering algorithm to merge abnormal influence nodes with adjacent spatial positions and consistent maintenance scene attributes (for example, both need to open the same access hole, or both need the same protective equipment); the spatial set after merging is defined as the "to-be-maintained area".

[0131] For each determined maintenance area, the system analyzes the "node morphology" of all nodes in the area, and combines the fault type (such as wear, fracture, misalignment) to derive the specific "maintenance content"; the content definition follows the standardized job language, for example: for the "wear morphology" of the rotary body node, the content definition is "cooperation surface size measurement and repair"; for the "fracture morphology" of the fastener node, the content definition is "removal of debris and re-threading of threaded hole"; for the "offset morphology" of the connection node, the content definition is "geometric centering verification (laser centering)"; the maintenance content forms a detailed work order for the area, at this time, the maintenance content refers to the specific maintenance task determined to solve the fault; the maintenance system includes the mapping of fault mode and repair scheme.

[0132] Specifically, the system has locked two key abnormal nodes of the blower: N-01 (gearbox high-speed shaft bearing seat) and N-02 (coupling guard side); node N-01: the system locates its coordinates in the digital model as the inside of the gearbox, the vertical of the high-speed shaft input end; according to its characteristics of bearing high-speed rotating parts and being wrapped by the box, it is marked as "precise rotary support cavity (closed type)" in node morphology; node N-02: the system locates its coordinates between the gearbox output end and the motor, the outer circumference of the coupling flange; according to its characteristics of connecting two shafts and being exposed, it is marked as "open coaxial connection end face" in node morphology.

[0133] Scene analysis: the blower is currently in a shutdown state for repair, the ambient temperature is 50°C; N-01 is located inside the gearbox and must be removed to be accessed, and involves precision components, requiring a dust-free environment; N-02 is located externally and can be directly accessed, but the space is narrow and limited by the surrounding pipes; area division: considering the closedness and disassembly difficulty (hoisting) of N-01, the system divides N-01 and its associated gearbox upper cover peripheral area (radius 1 meter range) as "maintenance area A: core transmission precision area"; considering the external location of N-02 and the centering operation requirement, the system divides the coupling peripheral area as "maintenance area B: transmission centering operation area".

[0134] For area A (node N-01): based on the node morphology of "closed precision cavity" and the fault of bearing fracture, the system determines the maintenance content as: "perform gearbox opening operation, clean the cavity metal debris, replace the high-speed shaft bearing assembly"; for area B (node N-02): based on the node morphology of "open connection end face" and the offset affected by vibration, the system determines the maintenance content as: "remove the coupling guard, clean the flange, perform laser centering correction, check the elastomer column pin wear condition".

[0135] Therefore, based on the traceability of the previous maintenance event of the air blower, the corresponding maintenance system is determined, the first heavy maintenance coefficient is determined according to the maintenance system and the to-be-maintained content, the second heavy maintenance coefficient is determined according to the maintenance system and the corresponding abnormal component, the corresponding abnormal maintenance measure is determined based on the mapping relationship table of the first heavy maintenance coefficient, the second heavy maintenance coefficient and the abnormal maintenance measure, the overall consideration of the first heavy maintenance coefficient, the second heavy maintenance coefficient and the mapping relationship table of the abnormal maintenance measure is compatible, the accuracy of the corresponding abnormal maintenance measure is guaranteed, and meanwhile, the overall consideration of the to-be-maintained content, the current working mode of the air blower and the corresponding maintenance system is realized, and the accuracy of the abnormal maintenance measure of the air blower is improved.

[0136] At this time, the system accesses the equipment full life cycle management database (PLM / EAM), deeply traces the past maintenance records of the air blower and the same family equipment; the system uses natural language processing (NLP) and association rule mining technology to extract key entities and relationships from historical work orders, fault reports and maintenance logs, and constructs a dynamic "maintenance system"; the system not only contains the mapping of fault modes and repair schemes, but also contains the historical success rate, average repair time (MTTR), spare part consumption law and "man-machine-material-method-environment" coupling characteristics. Through this system, the system can identify the optimal disposal path of a specific fault in different scenarios, and provide data support for the calculation of subsequent coefficients.

[0137] The system inputs the "to-be-maintained content" (such as "replace bearing", "laser centering", "dynamic balance check") determined in the previous step into the maintenance system; the system evaluates the technical complexity of these contents, which involves the precision level of the required tools, the topological complexity of the disassembly steps and the requirements for the skill qualification of the maintenance personnel; the system calculates the "first heavy maintenance coefficient" combined with the operation time fluctuation rate and the one-time repair rate of such contents in the historical records; the coefficient mainly reflects the execution difficulty and resource consumption intensity of the maintenance operation itself; for example, the coefficient of a precision assembly operation which needs to be heated and disassembled and cold assembled will be significantly higher than that of a conventional bolt fastening operation.

[0138] The system inputs the "abnormal component" (such as P0 level high-speed shaft, P1 level impeller) into the maintenance system; the system evaluates the functional criticality of the component in the system, the severity of the failure consequences (SafetyCriticality) and the economic value based on the BOM (bill of materials) level and reliability data of the component; combined with the average failure interval time (MTBF) and the spare part supply cycle of the component in the historical records, the system calculates the "second heavy maintenance coefficient"; the coefficient mainly reflects the risk weight of the component failure on production and the urgency of replacement / repair; for example, the coefficient of a core bearing component failure which causes the whole machine to stop is higher than that of an auxiliary cooling system component.

[0139] The system establishes a high-dimensional "abnormal maintenance measure mapping table", which divides the maintenance measures into multiple levels, such as "after-maintenance (BM), preventive maintenance (PM), predictive maintenance (PdM), improvement maintenance (IM) or active reset (AR)"; the system uses a weighted decision algorithm (such as AHP or fuzzy comprehensive evaluation) to fuse and calculate the "first maintenance coefficient" (execution difficulty) and the "second maintenance coefficient" (component importance) to generate a comprehensive decision value; in the mapping table, the entry that best matches the decision value is searched and outputted as the final "abnormal maintenance measure"; the measure not only contains action instructions (such as "immediate shutdown"), but also contains auxiliary resource suggestions (such as "300-ton crane needs to be dispatched" and "spare part A-002 needs to be prepared").

[0140] Specifically, the blower has determined the maintenance content to be "gearbox cover opening and replacement of high-speed shaft bearing (high difficulty)" and "coupling centering (medium difficulty)", and the abnormal component is "high-speed shaft and bearing (P0 core component)"; the system searches the maintenance records of the blower in the past 5 years and finds that there have been 12 historical events for "gearbox high-speed shaft" failure; the system analysis finds that 8 of them have adopted the "replace bearing only" scheme, of which 3 have failed again within half a year due to gear damage; 4 of them have adopted the "replace bearing + gear flaw detection + shaft system repair" scheme, and the average operation life has been extended by 200%; based on this, the maintenance system confirms that for this part failure, simply replacing is a "high-risk low-benefit" strategy, while systematic repair is a "high-reliability strategy".

[0141] The system analyzes the "maintenance content": the content includes "gearbox cover opening" and "precision bearing replacement", and historical data shows that such operation takes an average of 24 hours and requires a constant temperature environment and hydraulic puller tools; according to the complexity evaluation model of the maintenance system (reference value is 0.0-1.0), such operation is judged to be of extremely high complexity; the system calculates the first maintenance coefficient K1=0.9 (close to 1, indicating that the operation difficulty is extremely great and the resource consumption is high).

[0142] The system analyzes the "abnormal component": the high-speed shaft bearing is a P0-level core component, and its direct failure will cause the production line to stop production, resulting in a loss of hundreds of thousands of yuan per hour; the spare part is a long-cycle imported part (with a procurement cycle of 8 weeks); according to the FMEA (Failure Mode and Effects Analysis) model of the maintenance system, the risk priority number (RPN) of such component failure is the highest; the system calculates the second maintenance coefficient K2=1.0 (the maximum value, indicating that the component is extremely critical and the failure consequence is extremely serious).

[0143] The system calculates the comprehensive decision value: V = a K1 + b K2 (assuming the weights are each 0.5, V = 0.5 x 0.9 + 0.5 x 1.0 = 0.95); the system queries the "abnormal maintenance measure mapping relationship table" and finds that the interval [0.85, 1.0] corresponds to the measure "Level I emergency overhaul (complete recovery)"; the final output is: the system determines the abnormal maintenance measure as "immediately perform shutdown overhaul: lift the entire gear box unit, return to the factory or perform open cover complete disassembly inspection by an expert group, replace the entire high-speed shaft system components, and perform magnetic powder detection on all related gears, and strictly prohibit partial replacement of bearings".

[0144] Please refer to Figure 7 , the abnormal identification system of the blower based on multi-modal data comprises:

[0145] A vibration data module 21 is configured to determine a real-time image of the blower based on visual detection of the blower when the blower is in a working state, determine a plurality of component regions based on recognition of the real-time image of the blower, and determine a corresponding vibration data combination according to vibration detection of each component region.

[0146] A multi-modal data module 22 is configured to determine a multi-modal framework of the blower based on the region position of each component region and the corresponding vibration data combination, and construct multi-modal data corresponding to the blower according to the multi-modal framework of the blower, noise data and temperature data of the blower.

[0147] An abnormal identification module 23 is configured to determine a plurality of abnormal data according to detection of the multi-modal data corresponding to the blower, mark abnormal components corresponding to each abnormal data, determine an abnormal identification system according to the plurality of abnormal data, the abnormal components corresponding to each abnormal data and the overall form of the blower, and mark a plurality of abnormal regions of different abnormal influence levels.

[0148] An abnormal influence route module 24 is configured to determine a corresponding abnormal event in each abnormal region according to dynamic detection of the abnormal region, determine a plurality of abnormal influence factors based on each abnormal event, real-time working data of the blower and the abnormal identification system, and form a corresponding abnormal influence route.

[0149] An abnormal maintenance module 25 is configured to determine a plurality of abnormal influence nodes based on identification of the abnormal influence route, determine maintenance content according to the node position of each abnormal influence node, the corresponding node form and the use scene of the blower, and determine corresponding abnormal maintenance measures based on the maintenance content, the current working mode of the blower and the corresponding maintenance system.

[0150] Any technical features of the above embodiments can be combined in any manner, and for brevity, not all combinations of the technical features described above are described, however, any combination of the technical features should be considered as within the scope of the present specification, as long as the combination does not result in a contradiction.

Claims

1. A method for identifying an anomaly of a blower based on multi-modal data, the method comprising: The method comprises the following steps: determining a real-time image of the blower based on visual detection of the blower when the blower is in operation, determining a plurality of component regions based on recognition of the real-time image of the blower, and determining a corresponding vibration data combination according to vibration detection of each component region; determining a multi-modal framework of the blower based on the region position of each component region and the corresponding vibration data combination, and constructing multi-modal data corresponding to the blower according to the multi-modal framework of the blower, noise data and temperature data of the blower; the multi-modal framework not only contains the current vibration state, but also implies the three-dimensional geometric structure and fault importance information of the equipment; the multi-modal data contains spatial position, vibration characteristics, thermodynamic characteristics, acoustic characteristics and scene semantics; determining a plurality of abnormal data according to detection of the multi-modal data corresponding to the blower, and marking an abnormal component corresponding to each abnormal data, determining an abnormal recognition system according to the plurality of abnormal data, the abnormal component corresponding to each abnormal data and the overall form of the blower, and marking a plurality of abnormal regions of different abnormal influence levels, the abnormal recognition system being a comprehensive system for fault grading by double fusion of the structural dimension and the working condition dimension of the blower; in each abnormal region, determining a corresponding abnormal event according to dynamic detection of the abnormal region, determining a plurality of abnormal influence factors based on each abnormal event, real-time working data of the blower and the abnormal recognition system, and forming a corresponding abnormal influence route, the abnormal influence route being a panoramic multi-dimensional fault mapping graph; determining a plurality of abnormal influence nodes based on recognition of the abnormal influence route, determining a maintenance content according to the node position, the corresponding node form and the use scene of the blower of each abnormal influence node, determining a corresponding abnormal maintenance measure based on the maintenance content, the current working mode of the blower and the corresponding maintenance system, and the abnormal influence node being defined as the most critical physical position of the maintenance operation; the node position of the abnormal influence node covers the corresponding installation coordinates, hierarchical relationship and surrounding envelope boundary; the node form of the abnormal influence node contains the corresponding form topological characteristics; the maintenance content refers to a specific maintenance task determined to solve the fault; the maintenance system contains the mapping of the fault mode and the repair scheme. 2.The method of claim 1, wherein, The method comprises the following steps: marking the position of the blower, collecting a plurality of working data of the blower, determining the working state of the blower according to the plurality of working data of the blower and the corresponding overall form, controlling the working state of the blower, triggering the positioning shooting of a plurality of cameras around the position of the blower to visually detect the blower, and collecting a plurality of sub-images of the blower in different dimensions, and constructing the real-time image of the blower according to the synthesis of the plurality of sub-images; The real-time image of the air blower is dynamically identified, and a plurality of component marks are determined in the identification process. The corresponding component area is determined according to the traceability of each component mark, so as to trigger the vibration detection of each component area. At the same time, a plurality of vibration data are determined according to the vibration detection of each component area. Based on the plurality of vibration data, the corresponding vibration detection position and the corresponding component area, the corresponding vibration data combination is determined. 3.The method of claim 1, wherein, The multi-modal framework of the air blower is determined based on the area position of each component area and the corresponding vibration data combination. The multi-modal data corresponding to the air blower is constructed according to the multi-modal framework of the air blower, the noise data and the temperature data of the air blower, including: In the plurality of component areas, the corresponding area position is determined based on the detection of each component area. The multi-modal framework of the air blower is determined according to the area position of each component area, the corresponding area priority and the corresponding vibration data combination. A plurality of environmental data are determined based on the detection of the location of the air blower. The noise data and the temperature data of the air blower are determined according to the screening of the plurality of environmental data, and the use scene of the air blower is marked. The primary data is determined according to the multi-modal framework of the air blower and the use scene of the air blower, and the multi-modal data corresponding to the air blower is constructed based on the primary data, the noise data and the temperature data of the air blower. 4.The method of claim 1, wherein, A plurality of abnormal data are determined according to the detection of the multi-modal data corresponding to the air blower, and the abnormal components corresponding to each abnormal data are marked. The abnormal recognition system is determined according to the plurality of abnormal data, the abnormal components corresponding to each abnormal data and the overall form of the air blower, and a plurality of abnormal areas of different abnormal influence levels are marked, including: The multi-modal data corresponding to the air blower is dynamically detected, and a plurality of data abnormality marks are determined in the detection process. The corresponding abnormal data is determined according to the traceability of each data abnormality mark, so as to collect a plurality of abnormal data. 5.The method of claim 4, wherein, A plurality of abnormal data are determined according to the detection of the multi-modal data corresponding to the air blower, and the abnormal components corresponding to each abnormal data are marked. The abnormal recognition system is determined according to the plurality of abnormal data, the abnormal components corresponding to each abnormal data and the overall form of the air blower, and a plurality of abnormal areas of different abnormal influence levels are marked, including: The corresponding abnormal component is determined based on the detection of each abnormal data. The first abnormal recognition content is determined according to the abnormal component corresponding to each abnormal data and the overall form of the air blower. The second abnormal recognition content is determined according to each abnormal data and the current working mode of the air blower. The abnormal recognition system is determined according to the first abnormal recognition content and the second abnormal recognition content. In the abnormal recognition system, a plurality of abnormal areas are determined based on the identification of the abnormal recognition system, and the corresponding abnormal influence level is determined according to the data combination of each abnormal area, the corresponding area position and the corresponding component function. 6.The method of identifying an abnormality of a blower based on multi-modal data according to claim 1, wherein, In each abnormal area, the corresponding abnormal event is determined according to the dynamic detection of the abnormal area. A plurality of abnormal influence factors are determined based on each abnormal event, the real-time working data of the air blower and the abnormal recognition system, and the corresponding abnormal influence route is formed, including: Real-time monitoring of each abnormal area, and marking the abnormal time period of each abnormal area, dynamically identifying the abnormal time period of each abnormal area, and determining the working content of the abnormal area in the abnormal time period in the identification process, according to the area form of the abnormal area, the corresponding working content and the corresponding component priority, determine the corresponding abnormal event, wherein the working content refers to the physical behavior pattern of the fault component in the micro level. 7.The method of claim 6, wherein, According to the dynamic detection of the abnormal area, the corresponding abnormal event is determined, and based on the abnormal event, the real-time working data of the blower and the abnormal identification system, a plurality of abnormal influence factors are determined, and the corresponding abnormal influence route is formed, which further comprises: Collecting real-time working data of the blower, determining the first heavy influence content according to the abnormal event and the real-time working data of the blower, determining the second heavy influence content according to the real-time working data of the blower and the abnormal identification system, and determining the plurality of abnormal influence factors based on the first heavy influence content and the second heavy influence content; Marking the influence component of each abnormal influence factor, determining a plurality of sub-influence routes according to each influence component, the corresponding abnormal influence factor and the current working mode of the blower, and forming the corresponding abnormal influence route based on the plurality of sub-influence routes, the current working route of the blower and the corresponding data transmission route. 8.The method of identifying an abnormality of a blower based on multi-modal data according to claim 1, wherein, According to the identification of the abnormal influence route, a plurality of abnormal influence nodes are determined, the maintenance content is determined according to the node position of each abnormal influence node, the corresponding node form and the use scene of the blower, and the corresponding abnormal maintenance measure is determined based on the maintenance content, the current working mode of the blower and the corresponding maintenance system, which comprises: In the abnormal influence route, a plurality of abnormal influence sections are determined according to the identification of the abnormal influence route, each abnormal influence section corresponds to at least one abnormal component in the blower, the corresponding abnormal influence node is determined according to the section form of each abnormal influence section and the corresponding abnormal component, and a plurality of abnormal influence nodes are collected; According to the detection of each abnormal influence node, the node position of each abnormal influence node is determined, and the node form of each abnormal influence node is marked, the maintenance area is determined according to the node position of each abnormal influence node and the use scene of the blower, and the maintenance content is determined according to the maintenance area and the node form of each abnormal influence node. 9.The method of claim 8, wherein, According to the identification of the abnormal influence route, a plurality of abnormal influence nodes are determined, the maintenance content is determined according to the node position of each abnormal influence node, the corresponding node form and the use scene of the blower, and the corresponding abnormal maintenance measure is determined based on the maintenance content, the current working mode of the blower and the corresponding maintenance system, which further comprises: Based on the traceability of the past maintenance event of the blower, the corresponding maintenance system is determined, the first heavy maintenance coefficient is determined according to the maintenance system and the maintenance content, the second heavy maintenance coefficient is determined according to the maintenance system and the corresponding abnormal component, and the corresponding abnormal maintenance measure is determined based on the mapping relationship table of the first heavy maintenance coefficient, the second heavy maintenance coefficient and the abnormal maintenance measure.

10. A blower abnormality identification system based on multi-modal data, characterized by, The abnormality recognition system of the blower based on multi-modal data is applied to the abnormality recognition method of the blower based on multi-modal data as claimed in any one of claims 1-9.

Images