A terminal building baggage claim facility health predictive maintenance method

Abstract

The present application relates to the technical field of machine vision image perception, and discloses a terminal building in-luggage extraction facility health degree predictive maintenance method, comprising the following steps: step S01: acquiring the timing acquisition image of the target object, step S02: carrying out double-branch image processing on the target image and target image sequence, step S03: carrying out gray level co-occurrence texture data processing on the calibrated target image, step S04: carrying out two-dimensional displacement anomaly prediction processing on the target image sequence, step S05: predicting the health degree of the target object based on the processing result, and step S06: receiving the judgment prediction result and outputting the predictive maintenance method, carrying out surface health detection on the target object, carrying out interframe two-dimensional displacement anomaly prediction on the surface texture block in each arc segment of the continuous effective frame, realizing early fault warning, and effectively improving the operation and maintenance intelligent level of the terminal building in-luggage extraction carousel equipment.

Description

Technical Field

[0001] This invention relates to the field of machine vision image perception technology, and more specifically to a method for predictive maintenance of the health status of baggage claim facilities in airport terminals. Background Technology

[0002] Baggage claim carousels in airport terminals are core infrastructure for passenger transport at civil aviation airports, undertaking the critical functions of baggage transfer, sorting, and conveying. The equipment operates continuously at high frequency around the clock, under complex load conditions, with frequent starts and stops, and experiences significant and continuous wear. Over long-term operation, it is highly susceptible to deterioration issues such as surface wear of the conveyor belts, aging of the texture, localized cracks, loosening of assembly gaps, accumulation of structural micro-deformations, and mechanical fatigue. With the advancement of intelligent and refined operation and maintenance upgrades in civil aviation airports, the traditional model of regular manual inspections and reactive repairs is no longer adequate to meet the high safety, high reliability, and zero-downtime operational requirements of airports.

[0003] Currently, the operation and maintenance system of domestic airport terminal facilities has gradually introduced intelligent visual inspection, condition monitoring, and information-based operation and maintenance platform technologies, gradually replacing the traditional purely manual inspection mode. Existing baggage claim carousel inspection technologies mostly rely on machine vision methods such as high-definition camera image acquisition, single-frame image visual analysis, edge detection, and pixel comparison to identify and screen visible faults such as surface damage, obvious stains, and large cracks on the conveyor belt. Meanwhile, some airports conduct periodic statistical analysis based on equipment operating years, operating hours, and fault log data to achieve simple equipment condition classification assessments. Existing technologies can meet the basic troubleshooting needs of obvious equipment faults and routine periodic operation and maintenance management, reducing the intensity of manual inspections to a certain extent, and improving the information and intelligent foundation level of terminal facility operation and maintenance. They are widely used in the daily operation and maintenance scenarios of electromechanical facilities in large, medium, and small airports.

[0004] However, the baggage carousel in the terminal is a highly dynamic open environment. Passengers bending over to retrieve their luggage and stacking suitcases periodically cover a large area of ​​the carousel surface. Such images cannot extract effective texture features or calculate effective motion information. If they are not removed, they will directly contaminate the monitoring data and lead to misjudgments. Surface appearance inspection and internal functional degradation inspection are disconnected from each other, lacking a failure mode verification mechanism to link the two, resulting in a high false alarm rate. Existing motion measurement solutions cannot locate anomalies to specific arc segments (angle positions), and maintenance personnel still need to check point by point. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a predictive maintenance method for the health of baggage claim facilities in airport terminals, so as to solve the problems existing in the background art.

[0006] This invention provides the following technical solution: a method for predictive maintenance of the health status of baggage claim facilities in airport terminals, comprising the following steps: Step S01: Acquire timed images of the target object: Use a fixed high-definition gimbal camera at the airport to timed acquire and update images of the baggage claim carousel in the terminal, and label the target image and target image sequence; Step S02: Perform dual-branch image processing on the target image and the target image sequence: perform preprocessing on the target image of a single frame, and perform temporal pixel processing on the target image sequence; Step S03: Perform gray-level co-occurrence texture data processing on the calibrated target image: perform surface health analysis on the calibrated single-frame target image based on the texture features extracted from the gray-level co-occurrence matrix; Step S04: Perform two-dimensional displacement anomaly prediction processing on the target image sequence: Divide the turntable annular working surface into multiple arc segments according to the arc length, and perform inter-frame two-dimensional displacement anomaly prediction on the surface texture blocks within each arc segment based on continuous effective frames. Step S05: Predict the health of the target object based on the processing results: Based on the processing results, perform surface health assessment and internal anomaly prediction on the target baggage claim carousel; Step S06: Receive the judgment and prediction results and output the predictive maintenance method: Based on the surface health judgment and internal anomaly prediction results, automatically output maintenance suggestions and push them to the terminal facility operation and maintenance management platform.

[0007] Preferably, in step S01, the specific details of obtaining the timed image of the target object are as follows: High-definition PTZ cameras are deployed at fixed monitoring points in the baggage claim hall of the terminal building. The cameras periodically collect data and dynamically update data according to a preset fixed time interval. Acquire timed images of the target object, which is the baggage claim carousel in the terminal building; The acquired images are archived and stored in chronological order according to their timestamps to construct a target image sequence for the baggage claim carousel. The latest frame in the target image sequence is designated as the target image. The number of images in the target image sequence is a fixed value. The target image sequence is updated according to an automatic update rule.

[0008] Preferably, in step S02, the specific content of performing dual-branch image processing on the target image and the target image sequence is as follows: Based on semantic segmentation, the proportion of the area of ​​the turntable's circular working surface occluded by the foreground target in each frame image is calculated, and invalid frames with an occlusion area proportion exceeding the first threshold are removed. For the target image after removing invalid frames, perform single-frame image preprocessing: convert the RGB color image to an 8-bit grayscale image; use Gaussian smoothing filter to remove noise during the image acquisition process; and combine the preset luggage carousel monitoring area coordinates to cut and lock the conveyor belt working surface, splicing seam area and baffle bonding area. The calibrated target image sequence is updated and global inter-frame registration correction is performed. The key monitoring pixel positions in the whole domain are fixed, and the gray values ​​of the same spatial coordinate position at different times are extracted frame by frame.

[0009] Preferably, in step S03, the specific content of performing grayscale co-occurrence texture data processing on the calibrated target image is as follows: The original gray-level co-occurrence matrix values ​​are obtained using a fixed unidirectional, fixed step-size method. The original gray-level co-occurrence matrix is ​​a two-dimensional counting matrix with dimensions of [missing information]. Where L is the total number of gray levels in the image, and any element in the matrix... The value is: within the effective ROI area of ​​the baggage carousel, under the condition of 0° horizontal direction and step size d=1, the baseline pixel grayscale value is The grayscale value of the adjacent pixel on the right is The total number of times the pixel pairs appear; A gray-level co-occurrence original matrix is ​​constructed by statistically analyzing the gray-level values ​​of adjacent pixels within the effective region. The content of the gray-level co-occurrence original matrix construction is as follows: Using the pre-processed grayscale ROI image of the luggage carousel as input, the traversal direction is fixed at 0° to the right horizontally and the pixel interval step size d=1; Traverse the valid pixels of the image row by row and column by column, taking the current coordinate pixel as the reference pixel and taking the one pixel adjacent to the right horizontally as the matching neighbor pixel; Read the grayscale value of the reference pixel Neighboring pixel grayscale values For the corresponding positions of the original matrix Perform cumulative counting; The entire process is traversed, counting only horizontally adjacent pixel pairs in the same direction. The final output is a raw gray-level co-occurrence count matrix consisting solely of gray-level pairing frequencies. This matrix is ​​then globally normalized to eliminate numerical biases caused by differences in image size and total pixel count, resulting in a standardized gray-level co-occurrence probability matrix. , where the matrix values ​​represent the spatial co-occurrence probability values ​​of the corresponding grayscale pixel pairs; Based on the standardized gray-level co-occurrence probability matrix The texture entropy is calculated separately and used as the unique texture quantization output value; The standard texture entropy baseline value of the baggage claim carousel under brand new standard conditions is pre-stored. The baseline value is a constant reference value when the equipment is free of wear and defects and has a regular texture. A threshold range is set based on the constant reference value. The unique texture feature value calculated in real time is compared with the threshold range: when the unique texture feature value does not exceed the set threshold range, the surface of the carousel conveyor belt is determined to be normal; when the unique texture feature value calculated in real time exceeds the set threshold range, the surface of the carousel conveyor belt is determined to be abnormal.

[0010] Preferably, in step S04, the specific content of performing two-dimensional displacement anomaly prediction processing on the target image sequence is as follows: Based on the continuously valid frames updated in step S02, the inter-frame two-dimensional displacement of the surface texture blocks in each arc segment of the turntable's annular working surface is tracked. The segmented velocities of each arc segment are obtained from the tangential components of the two-dimensional displacement, forming a segmented velocity field, and the segmented velocity consistency index is determined based on the dispersion of the velocities of each arc segment at the same moment. The surface runout energy of each arc segment is determined by the normal component of the two-dimensional displacement; When the segment speed consistency index is lower than the first judgment threshold, or the surface runout energy of any arc segment exceeds the second judgment threshold, the operation abnormality will be located to the corresponding arc segment.

[0011] Preferably, in step S05, the specific content of predicting the health status of the target object based on the processing result is as follows: The system receives the surface health status of the carousel conveyor belt output in step S03. When the surface of the carousel conveyor belt is normal, the result is that the target baggage claim carousel is in a healthy surface state. When the surface of the carousel conveyor belt is abnormal, the result is that the target baggage claim carousel is in a deteriorated surface state. The system outputs the above status to the interactive system connected to the terminal facility operation and maintenance management platform. Receive the operation abnormality positioning arc segment output in step S04, and output the positioning arc segment to the interactive system connected to the terminal facility operation and maintenance management platform.

[0012] Preferably, in step S06, the specific content of receiving the judgment prediction result and outputting the predictive maintenance method is as follows: Based on the current surface health assessment and anomaly location prediction results, maintenance recommendations are automatically generated and pushed to the terminal facility operation and maintenance management platform: When the conveyor belt surface is healthy and no abnormal positioning is predicted, the overall operating status of the equipment is determined to be excellent with no risk of failure. Routine maintenance recommendations are output: instruct maintenance personnel to maintain the frequency of inspections and record the steady-state operating data of the equipment. When the conveyor belt surface deteriorates and no abnormal positioning is predicted, it is determined that only the surface structure of the equipment is damaged, and there is no cumulative deterioration in the internal assembly structure and mechanical clearance. The overall structural stability of the equipment is normal, and the system outputs a special surface maintenance suggestion: to guide maintenance personnel to complete the local repair of the conveyor belt without disassembling and inspecting the internal structure. When the conveyor belt surface is healthy but an abnormal location is predicted, it is judged as a high potential fault risk state, and the system will remind the operation and maintenance personnel to carry out equipment disassembly and inspection in advance. When the conveyor belt surface deteriorates and abnormal positioning is predicted, it is determined that the overall equipment is severely aged and the structural stability has dropped significantly, indicating that it is in a high-risk operating state. The system outputs emergency operation and maintenance rectification suggestions: pushes high-risk alarm information, and instructs operation and maintenance personnel to shut down the machine to carry out comprehensive inspection, component replacement and structural calibration.

[0013] The technical effects and advantages of this invention are as follows: This invention eliminates invalid frames by removing the proportion of foreground occlusion area, fundamentally ensuring the data quality for subsequent surface and motion detection. It solves the problem of image data pollution caused by passenger occlusion in open and dynamic scenes, enabling machine vision solutions to be applied to highly dynamic open scenes such as airport baggage carousels. The surface detection branch quantifies the surface wear state of the band based on the gray-level co-occurrence matrix; the internal function detection branch constructs a segmented velocity field and calculates the segmented velocity consistency index based on the tangential component of the inter-frame two-dimensional displacement, calculates the band surface jump energy based on the normal component, and realizes arc-level positioning of abnormal operation through dual threshold judgment. The turntable's annular working surface is divided into multiple arc segments. Any operational abnormality can be located to a specific angle and a location mark can be output. Maintenance personnel can then directly access the fault point for repair, significantly shortening the troubleshooting time. The use of surface wear status and internal functional indicators to jointly determine the fault type significantly reduces the false alarm rate and improves the accuracy of maintenance recommendations. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating a predictive maintenance method for the health of baggage claim facilities within an airport terminal. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. In addition, the forms of the various structures described in the following embodiments are merely illustrative. The predictive maintenance method for the health status of baggage claim facilities in airport terminals involved in the present invention is not limited to the structures described in the following embodiments. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] like Figure 1As shown, this invention provides a method for predictive maintenance of the health status of baggage claim facilities in airport terminals, comprising the following steps: Step S01: Acquire timed images of the target object: Use a fixed high-definition gimbal camera at the airport to timed acquire and update images of the baggage claim carousel in the terminal, and label the target image and target image sequence; Step S02: Perform dual-branch image processing on the target image and the target image sequence: perform preprocessing on the target image of a single frame, and perform temporal pixel processing on the target image sequence; Step S03: Perform gray-level co-occurrence texture data processing on the calibrated target image: perform surface health analysis on the calibrated single-frame target image based on the texture features extracted from the gray-level co-occurrence matrix; Step S04: Perform two-dimensional displacement anomaly prediction processing on the target image sequence: Divide the turntable annular working surface into multiple arc segments according to the arc length, and perform inter-frame two-dimensional displacement anomaly prediction on the surface texture blocks within each arc segment based on continuous effective frames. Step S05: Predict the health of the target object based on the processing results: Based on the processing results, perform surface health assessment and internal anomaly prediction on the target baggage claim carousel; Step S06: Receive the judgment and prediction results and output the predictive maintenance method: Based on the surface health judgment and internal anomaly prediction results, automatically output maintenance suggestions and push them to the terminal facility operation and maintenance management platform.

[0017] In this embodiment, it should be specifically noted that the specific content of obtaining the timed image of the target object in step S01 is as follows: High-definition PTZ cameras are deployed at fixed monitoring points in the baggage claim hall of the terminal building. The cameras periodically collect data and dynamically update data according to a preset fixed time interval. Acquire timed images of the target object, which is the baggage claim carousel in the terminal building; The acquired images are archived and stored in chronological order according to their timestamps to construct a target image sequence for the baggage claim carousel. The latest frame in the target image sequence is designated as the target image. The number of images in the target image sequence is a fixed value. The target image sequence is updated according to an automatic update rule. The automatic update rule is as follows: after each round of new image acquisition, the earliest acquired single frame image in the image sequence is removed from the archive, and the latest acquired single frame image is stored at the end of the sequence, always maintaining the number of images in the target image sequence at a set fixed value.

[0018] In this embodiment, it should be specifically explained that the specific content of the dual-branch image processing of the target image and the target image sequence in step S02 is as follows: Based on semantic segmentation, the proportion of the area of ​​the turntable's annular working surface occluded by the foreground target in each frame of the image is calculated, and invalid frames with an occlusion area proportion exceeding a first threshold are removed. The calculation of the occlusion area proportion includes: pre-establishing a background model containing only the turntable's annular working surface; performing semantic segmentation on each frame of the image and the background model to obtain a foreground target mask; and calculating the proportion of the intersection area of ​​the foreground target mask and the turntable's annular working surface region to the total area of ​​the annular working surface as the occlusion area proportion. The first threshold is set to 15% to 30%, and in this embodiment, it is set to 25%. It can be adjusted according to the camera resolution and the number of arcs through ROC calibration. For the target image after removing invalid frames, perform single-frame image preprocessing: convert the RGB color image to an 8-bit grayscale image; use Gaussian smoothing filter to remove noise during the image acquisition process; and combine the preset luggage carousel monitoring area coordinates to cut and lock the conveyor belt working surface, splicing seam area and baffle bonding area. The calibrated target image sequence is updated and global inter-frame registration correction is performed. The key monitoring pixel positions in the whole domain are fixed, and the gray values ​​of the same spatial coordinate position at different times are extracted frame by frame. The update content is to remove invalid frames in the target image sequence.

[0019] In this embodiment, it should be specifically explained that the specific content of grayscale co-occurrence texture data processing of the calibrated target image in step S03 is as follows: The original gray-level co-occurrence matrix values ​​are obtained using a fixed unidirectional, fixed step-size method. Specifically, a step-size d=1 and a horizontal 0° direction are selected for pixel gray-level pair traversal and statistics. The original gray-level co-occurrence matrix is ​​a two-dimensional counting matrix with dimensions of [missing information]. Where L is the total number of gray levels in the image, and any element in the matrix... The value is: within the effective ROI area of ​​the baggage carousel, under the condition of 0° horizontal direction and step size d=1, the baseline pixel grayscale value is The grayscale value of the adjacent pixel on the right is The total number of times the pixel pairs appear; A gray-level co-occurrence original matrix is ​​constructed by statistically analyzing the gray-level values ​​of adjacent pixels within the effective region. The content of the gray-level co-occurrence original matrix construction is as follows: Using the pre-processed grayscale ROI image of the luggage carousel as input, the traversal direction is fixed at 0° to the right horizontally and the pixel interval step size d=1; Traverse the valid pixels of the image row by row and column by column, taking the current coordinate pixel as the reference pixel and taking the one pixel adjacent to the right horizontally as the matching neighbor pixel; Read the grayscale value of the reference pixel Neighboring pixel grayscale values For the corresponding positions of the original matrix Perform cumulative counting; The entire process is traversed, counting only horizontally adjacent pixel pairs in the same direction. The final output is a raw gray-level co-occurrence count matrix consisting solely of gray-level pairing frequencies. This matrix is ​​then globally normalized to eliminate numerical biases caused by differences in image size and total pixel count, resulting in a standardized gray-level co-occurrence probability matrix. The matrix value represents the spatial co-occurrence probability value of the corresponding grayscale pixel pair. The specific calculation content of the spatial co-occurrence probability value is as follows: ,in This represents the total statistical frequency of all valid grayscale pixel pairs; By normalizing the probabilities of the frequencies of each element in the original matrix, a standardized gray-level co-occurrence probability matrix is ​​obtained: ,in Represents the standardized gray-level co-occurrence probability matrix; Based on the standardized gray-level co-occurrence probability matrix The texture entropy is calculated separately and used as the unique texture quantization output value. The expression for the texture entropy is: ,in This represents the unique texture feature value corresponding to the current target image. Represents the standardized gray-level co-occurrence probability matrix; The standard texture entropy baseline value of the baggage claim carousel under brand new standard conditions is pre-stored. The baseline value is a constant reference value when the equipment is free of wear and defects and has a regular texture. A threshold range is set based on the constant reference value. The unique texture feature value calculated in real time is compared with the threshold range: when the unique texture feature value does not exceed the set threshold range, the surface of the carousel conveyor belt is determined to be normal; when the unique texture feature value calculated in real time exceeds the set threshold range, the surface of the carousel conveyor belt is determined to be abnormal.

[0020] In this embodiment, it should be specifically explained that the specific content of the two-dimensional displacement anomaly prediction processing of the target image sequence in step S04 is as follows: Based on the continuously valid frames updated in step S02, the inter-frame two-dimensional displacement of the surface texture blocks within each arc segment of the turntable's annular working surface is tracked: An ellipse fit is performed on the outer ring of the turntable to obtain the turntable's center and radius parameters. Using a pre-calibrated homography matrix H (solved using the turntable plane as the reference plane and the corner points of the calibration object), the image is transformed into a top-down bird's-eye view. A polar coordinate system with the turntable center as the origin is then established in the top-down plane. ; The annular working surface is divided into M arc segments according to the arc length. In this embodiment, M=72, each segment is 5°, and each arc segment x has a central angle. Within each arc segment, three circumferential orbital lines (radius) are defined: inner, middle, and outer. (located at 25%, 50%, and 75% of the bandwidth), texture blocks are selected along the track; In the polar coordinate system viewed from above, the x-center angle of the arc segment The tangential unit vector at point is The normal unit vector is The two-dimensional displacement vector of texture block k Decomposed into: .

[0021] The segmented velocities of each arc segment are obtained from the tangential components of the two-dimensional displacement, forming a segmented velocity field. The segmented velocity consistency index is then determined based on the velocities of each arc segment at the same moment. The segmented velocity of arc segment x at time t is taken as the median of the effective block tangential components within the arc segment, and the segmented velocities of all arc segments constitute the segmented velocity field. based on Determine the segmented speed consistency index, where This represents the piecewise velocity consistency index of arc segment i at time t. and This represents the standard deviation and mean of the segment velocities of all effective arc segments at time t.

[0022] The surface runout energy of each arc segment is determined by the normal component of the two-dimensional displacement: The sum of squares of the normal components of all valid texture blocks in consecutive frames within arc segment i is calculated. The result is divided by the product of the number of valid texture blocks and the number of consecutive frames to obtain the band surface jump energy of the arc segment. The larger the energy value, the more violent the band surface structure normal fluctuation at the corresponding position, and the higher the probability of hidden degradation.

[0023] When the segment speed consistency index is lower than the first judgment threshold, or the surface runout energy of any arc segment exceeds the second judgment threshold, the operation abnormality will be located to the corresponding arc segment.

[0024] In this embodiment, it should be specifically explained that the specific content of predicting the health of the target object based on the processing result in step S05 is as follows: The system receives the surface health status of the carousel conveyor belt output in step S03. When the surface of the carousel conveyor belt is normal, the result is that the target baggage claim carousel is in a healthy surface state. When the surface of the carousel conveyor belt is abnormal, the result is that the target baggage claim carousel is in a deteriorated surface state. The system outputs the above status to the interactive system connected to the terminal facility operation and maintenance management platform. Receive the operation abnormality positioning arc segment output in step S04, and output the positioning arc segment to the interactive system connected to the terminal facility operation and maintenance management platform.

[0025] In this embodiment, it should be specifically explained that the specific content of receiving the judgment prediction result and outputting the predictive maintenance method in step S06 is as follows: Based on the current surface health assessment and anomaly location prediction results, maintenance recommendations are automatically generated and pushed to the terminal facility operation and maintenance management platform: When the conveyor belt surface is healthy and no abnormal positioning is predicted, the overall operating status of the equipment is determined to be excellent with no risk of failure. Routine maintenance recommendations are output: instruct maintenance personnel to maintain the frequency of inspections and record the steady-state operating data of the equipment. When the conveyor belt surface deteriorates and no abnormal positioning is predicted, it is determined that only the surface structure of the equipment is damaged, and there is no cumulative deterioration in the internal assembly structure and mechanical clearance. The overall structural stability of the equipment is normal, and the system outputs a special surface maintenance suggestion: to guide maintenance personnel to complete the local repair of the conveyor belt without disassembling and inspecting the internal structure. When the conveyor belt surface is healthy but an abnormal location is predicted, it is judged as a high potential fault risk state, and the system will remind the operation and maintenance personnel to carry out equipment disassembly and inspection in advance. When the conveyor belt surface deteriorates and abnormal positioning is predicted, it is determined that the overall equipment is severely aged and the structural stability has dropped significantly, indicating that it is in a high-risk operating state. The system outputs emergency operation and maintenance rectification suggestions: pushes high-risk alarm information, and instructs operation and maintenance personnel to shut down the machine to carry out comprehensive inspection, component replacement and structural calibration.

[0026] The main difference between this implementation and the prior art is that this embodiment removes invalid frames by eliminating the proportion of foreground occlusion area, which fundamentally ensures the data quality for subsequent surface detection and motion detection. This solves the problem of image data pollution caused by passenger occlusion in open and dynamic scenes, enabling the machine vision solution to be applied to highly dynamic open scenes such as airport baggage carousels. The surface detection branch quantifies the surface wear state of the band based on the gray-level co-occurrence matrix; the internal function detection branch constructs a segmented velocity field and calculates the segmented velocity consistency index based on the tangential component of the inter-frame two-dimensional displacement, calculates the band surface jump energy based on the normal component, and realizes arc-level positioning of abnormal operation through dual threshold judgment. The turntable's annular working surface is divided into multiple arc segments. Any operational abnormality can be located to a specific angle and a location mark can be output. Maintenance personnel can then directly access the fault point for repair, significantly shortening the troubleshooting time. The use of surface wear status and internal functional indicators to jointly determine the fault type significantly reduces the false alarm rate and improves the accuracy of maintenance recommendations.

[0027] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0028] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predictive maintenance of the health status of baggage claim facilities in an airport terminal, characterized in that: Includes the following steps: Step S01: Acquire timed images of the target object: Use a fixed high-definition gimbal camera at the airport to timed acquire and update images of the baggage claim carousel in the terminal, and label the target image and target image sequence; Step S02: Perform dual-branch image processing on the target image and the target image sequence: perform preprocessing on the target image of a single frame, and perform temporal pixel processing on the target image sequence; Step S03: Perform gray-level co-occurrence texture data processing on the calibrated target image: perform surface health analysis on the calibrated single-frame target image based on the texture features extracted from the gray-level co-occurrence matrix; Step S04: Perform two-dimensional displacement anomaly prediction processing on the target image sequence: Divide the turntable annular working surface into multiple arc segments according to the arc length, and perform inter-frame two-dimensional displacement anomaly prediction on the surface texture blocks within each arc segment based on continuous effective frames. Step S05: Predict the health of the target object based on the processing results: Based on the processing results, perform surface health assessment and internal anomaly prediction on the target baggage claim carousel; Step S06: Receive the judgment and prediction results and output the predictive maintenance method: Based on the surface health judgment and internal anomaly prediction results, automatically output maintenance suggestions and push them to the terminal facility operation and maintenance management platform.

2. The method for predictive maintenance of the health status of baggage claim facilities in an airport terminal according to claim 1, characterized in that: In step S01, the specific details of obtaining the timed image of the target object are as follows: High-definition PTZ cameras are deployed at fixed monitoring points in the baggage claim hall of the terminal building. The cameras periodically collect data and dynamically update data according to a preset fixed time interval. Acquire timed images of the target object, which is the baggage claim carousel in the terminal building; The acquired images are archived and stored in chronological order according to their timestamps to construct a target image sequence for the baggage claim carousel. The latest frame in the target image sequence is designated as the target image. The number of images in the target image sequence is a fixed value. The target image sequence is updated according to an automatic update rule.

3. The method for predictive maintenance of the health status of baggage claim facilities in an airport terminal according to claim 1, characterized in that: In step S02, the specific details of performing dual-branch image processing on the target image and the target image sequence are as follows: Based on semantic segmentation, the proportion of the area of ​​the turntable's circular working surface occluded by the foreground target in each frame image is calculated, and invalid frames with an occlusion area proportion exceeding the first threshold are removed. For the target image after removing invalid frames, perform single-frame image preprocessing: convert the RGB color image to an 8-bit grayscale image; Gaussian smoothing filter is used to remove noise during image acquisition. Combined with the preset coordinates of the baggage carousel monitoring area, the working surface of the conveyor belt, the splicing seam area and the baffle bonding area are clipped and locked. The calibrated target image sequence is updated and global inter-frame registration correction is performed. The key monitoring pixel positions in the whole domain are fixed, and the gray values ​​of the same spatial coordinate position at different times are extracted frame by frame.

4. The method for predictive maintenance of the health status of baggage claim facilities in an airport terminal according to claim 1, characterized in that: In step S03, the specific details of performing grayscale co-occurrence texture data processing on the calibrated target image are as follows: The original gray-level co-occurrence matrix values ​​are obtained using a fixed unidirectional, fixed step-size method. The original gray-level co-occurrence matrix is ​​a two-dimensional counting matrix with dimensions of [missing information]. Where L is the total number of gray levels in the image, and any element in the matrix... The value is: within the effective ROI area of ​​the baggage carousel, under the condition of 0° horizontal direction and step size d=1, the baseline pixel grayscale value is The grayscale value of the adjacent pixel on the right is The total number of times the pixel pairs appear; A gray-level co-occurrence original matrix is ​​constructed by statistically analyzing the gray-level values ​​of adjacent pixels within the effective region. The content of the gray-level co-occurrence original matrix construction is as follows: Using the pre-processed grayscale ROI image of the luggage carousel as input, the traversal direction is fixed at 0° to the right horizontally and the pixel interval step size d=1; Traverse the valid pixels of the image row by row and column by column, taking the current coordinate pixel as the reference pixel and taking the one pixel adjacent to the right horizontally as the matching neighbor pixel; Read the grayscale value of the reference pixel Neighboring pixel grayscale values For the corresponding positions of the original matrix Perform cumulative counting; The entire process is traversed, counting only horizontally adjacent pixel pairs in the same direction. The final output is a raw gray-level co-occurrence count matrix consisting solely of gray-level pairing frequencies. This matrix is ​​then globally normalized to eliminate numerical biases caused by differences in image size and total pixel count, resulting in a standardized gray-level co-occurrence probability matrix. , where the matrix values ​​represent the spatial co-occurrence probability values ​​of the corresponding grayscale pixel pairs; Based on the standardized gray-level co-occurrence probability matrix The texture entropy is calculated separately and used as the unique texture quantization output value; The standard texture entropy baseline value of the baggage claim carousel under brand new standard conditions is pre-stored. The baseline value is a constant reference value when the equipment is free of wear and defects and has a regular texture. A threshold range is set based on the constant reference value. The unique texture feature value calculated in real time is compared with the threshold range: when the unique texture feature value does not exceed the set threshold range, the surface of the carousel conveyor belt is determined to be normal; when the unique texture feature value calculated in real time exceeds the set threshold range, the surface of the carousel conveyor belt is determined to be abnormal.

5. The method for predictive maintenance of the health status of baggage claim facilities in an airport terminal according to claim 1, characterized in that: In step S04, the specific details of performing two-dimensional displacement anomaly prediction processing on the target image sequence are as follows: Based on the continuously valid frames updated in step S02, the inter-frame two-dimensional displacement of the surface texture blocks in each arc segment of the turntable's annular working surface is tracked. The segmented velocities of each arc segment are obtained from the tangential components of the two-dimensional displacement, forming a segmented velocity field, and the segmented velocity consistency index is determined based on the dispersion of the velocities of each arc segment at the same moment. The surface runout energy of each arc segment is determined by the normal component of the two-dimensional displacement; When the segment speed consistency index is lower than the first judgment threshold, or the surface runout energy of any arc segment exceeds the second judgment threshold, the operation abnormality will be located to the corresponding arc segment.

6. The method for predictive maintenance of the health status of baggage claim facilities in an airport terminal according to claim 1, characterized in that: In step S05, the specific content of predicting the health status of the target object based on the processing results is as follows: The system receives the surface health status of the carousel conveyor belt output in step S03. When the surface of the carousel conveyor belt is normal, the result is that the target baggage claim carousel is in a healthy surface state. When the surface of the carousel conveyor belt is abnormal, the result is that the target baggage claim carousel is in a deteriorated surface state. The system outputs the above status to the interactive system connected to the terminal facility operation and maintenance management platform. Receive the operation abnormality positioning arc segment output in step S04, and output the positioning arc segment to the interactive system connected to the terminal facility operation and maintenance management platform.

7. The method for predictive maintenance of the health status of baggage claim facilities in an airport terminal according to claim 1, characterized in that: In step S06, the specific content of receiving the judgment prediction result and outputting the predictive maintenance method is as follows: Based on the current surface health assessment and anomaly location prediction results, maintenance recommendations are automatically generated and pushed to the terminal facility operation and maintenance management platform: When the conveyor belt surface is healthy and no abnormal positioning is predicted, the overall operating status of the equipment is determined to be excellent with no risk of failure. Routine maintenance recommendations are output: instruct maintenance personnel to maintain the frequency of inspections and record the steady-state operating data of the equipment. When the conveyor belt surface deteriorates and no abnormal positioning is predicted, it is determined that only the surface structure of the equipment is damaged, and there is no cumulative deterioration in the internal assembly structure and mechanical clearance. The overall structural stability of the equipment is normal, and the system outputs a special surface maintenance suggestion: to guide maintenance personnel to complete the local repair of the conveyor belt without disassembling and inspecting the internal structure. When the conveyor belt surface is healthy but an abnormal location is predicted, it is judged as a high potential fault risk state, and the system will remind the operation and maintenance personnel to carry out equipment disassembly and inspection in advance. When the conveyor belt surface deteriorates and abnormal positioning is predicted, it is determined that the overall equipment is severely aged and the structural stability has dropped significantly, indicating that it is in a high-risk operating state. The system outputs emergency operation and maintenance rectification suggestions: pushes high-risk alarm information, and instructs operation and maintenance personnel to shut down the machine to carry out comprehensive inspection, component replacement and structural calibration.

Images