An intelligent integrated ship accessory surface treatment defect identification and scheduling method

Abstract

The application discloses a kind of intelligent integrated ship accessory surface treatment defect identification scheduling method, it is related to ship accessory surface detection technical field, including: collection parallel polarization image, cross polarization image, first illumination image and second illumination image, extract candidate abnormal area, and according to polarization brightness attenuation ratio, illumination characteristic distance, surface migration distance and mirror surface reflection angle judge reflection interference;According to surface dew point margin and three unit temperature rise data, condensation interference is judged;According to the compactness of region, the standard deviation of regional gray scale and the change of abnormal area before and after airflow action, loose attachment is judged;Film thickness detection is carried out to stable entity abnormal area, determine coating insufficient coverage, coating local accumulation and equipment common anomaly;According to each detection record generation and execution task, and update historical effective detection record.The application can exclude optical interference, environmental interference and loose attachment gradually, improve the accuracy of entity defect identification and task scheduling.

Description

Technical Field

[0001] This invention relates to the field of surface inspection technology for ship parts, specifically to an intelligent integrated method for identifying and scheduling surface treatment defects in ship parts. Background Technology

[0002] After ship fittings undergo painting, coating, and other surface treatments, it is necessary to inspect for insufficient coating coverage, localized coating buildup, and other surface anomalies. Existing inspection methods typically involve acquiring images of the fitting surface and determining the defect type through image feature extraction and classification models. However, due to the presence of reflectivity, condensation, and loose deposits on the surface of ship fittings, bright spots, dark spots, and irregular areas in the acquired images may have similar appearance characteristics to actual defects, easily affecting the defect identification results. Chinese invention patent application CN112950584A, published on June 11, 2021, discloses a deep learning-based method for identifying coating surface defects. This method extracts features from coating surface defect images using a feature extraction network and outputs defect identification results using a classifier, thus automatically identifying coating surface defects. However, this patent application primarily relies on appearance features in the acquired images for defect identification. When the imaging state of the inspected surface is affected by lighting conditions, the environment, and surface deposits, image features generated by non-quality defects may resemble actual defects, leading to inaccurate defect type identification. Furthermore, existing detection methods typically focus on outputting identification results, lacking a unified arrangement for detection results, subsequent processing, and production schedules, making it difficult to balance detection accuracy and timely processing.

[0003] Therefore, the present invention provides an intelligent integrated method for identifying and scheduling surface treatment defects in ship parts. Summary of the Invention

[0004] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an intelligent integrated method for identifying and scheduling surface treatment defects in ship parts. By sequentially judging polarization and illumination position changes, controlled heating, airflow disturbance, and film thickness, it distinguishes between reflective interference, condensation interference, loose attachments, and physical defects. Based on the identification results, it generates corresponding processing tasks and equipment verification tasks, thereby achieving accurate identification and task scheduling of surface treatment defects in ship parts.

[0005] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: an intelligent integrated method for identifying and scheduling surface treatment defects in ship parts, comprising: S1. Obtain accessory identification, 3D surface model, equipment operating parameters and historical valid test records, and acquire parallel polarization image, cross polarization image, first illumination image and second illumination image; S2. Extract candidate abnormal regions and form reflection interference records and environmental interference judgment records based on polarization brightness attenuation ratio, illumination characteristic distance, surface migration distance and specular reflection angle. S3. For the environmental interference judgment record, a condensation interference record and an adhesion state judgment record are formed based on the surface dew point balance and three unit temperature rise data. S4. For the record of the state to be attached, the records of loose attachments and stable entity anomalies are formed based on the area compactness, area gray standard deviation, abnormal center displacement, abnormal area reduction ratio and area overlap ratio. S5. For stable entity anomaly records, based on film thickness detection results, low area overlap ratio, high area overlap ratio and film thickness difference, form coating insufficient coverage record, coating local accumulation record and equipment common anomaly record. S6. Generate and execute corresponding tasks based on the records of reflective interference, condensation interference, loose attachments, insufficient coating coverage, local coating accumulation, and common equipment anomalies, and update the historical valid detection records.

[0006] Preferably, the following information is obtained for ship parts: part identification, 3D surface model, substrate material, coating type, coating color, surface treatment equipment identification, batch of treatment materials, coating treatment completion time, and start time of the next production process; Acquire the lateral focal length, camera longitudinal focal length, principal point x-coordinate, principal point y-coordinate, 3D coordinates and surface normals of each surface point of the 3D surface model in the camera coordinate system, effective measurement diameter of the film thickness detection probe, upper limit of image positioning error, maximum value of repeated measurement error of temperature sensor, upper limit of allowable detection temperature, and upper limit of allowable purging pressure. Obtain the equipment operating parameters corresponding to the ship parts. The equipment operating parameters include the material supply pressure from the start of outputting coating material to the stop of outputting coating material at the material output end, the material output per unit time, the distance between the material output end and the surface of the ship parts, and the moving speed of the material output end relative to the surface of the ship parts. Ship parts with the same base material, coating type and coating color are classified as similar parts; ship parts with the same batch of treated materials among similar parts are classified as parts from the same batch. The camera is controlled to sequentially photograph the surface of the ship parts, and the surface area of ​​the ship parts corresponding to each photograph is recorded as an image acquisition area. Each image acquisition area is set sequentially, with overlapping portions between adjacent areas, until all image acquisition areas cover the surface of the ship parts to be inspected. For each image acquisition area, keeping the position of the camera and the ship parts unchanged, the polarization direction of the light source is adjusted to be parallel to the transmission direction of the analyzer, and a parallel polarized image is acquired. Then, the polarization direction of the light source is adjusted to be perpendicular to the transmission direction of the analyzer, and a cross-polarized image is acquired. Keeping the polarization direction of the light source perpendicular to the transmission direction of the analyzer, the light source is moved to the first illumination position and the second illumination position, respectively. First and second illumination images are acquired for the same image acquisition area, and the three-dimensional coordinates of the light source center at the first and second illumination positions in the camera coordinate system are recorded. Simultaneously, the surface temperature and dew point temperature of the ship parts are acquired. For each pixel in each image, subtract the principal point's x-coordinate from the pixel's x-coordinate and then divide by the camera's lateral focal length to obtain the ray's lateral component; subtract the principal point's ordinate from the pixel's ordinate and then divide by the camera's longitudinal focal length to obtain the ray's longitudinal component; set the ray's forward component to one, and determine the ray's direction originating from the camera's optical center using the lateral, longitudinal, and forward components. Extending from the camera optical center along the ray direction to the three-dimensional surface model, the intersection point of the ray and the three-dimensional surface model that is closest to the camera optical center is determined as the surface point corresponding to the pixel, and the three-dimensional coordinates and surface normal of the surface point in the camera coordinate system are read as the three-dimensional surface position and surface normal corresponding to the pixel. For any region in the image, calculate the average of the three coordinate components of the three-dimensional coordinates of all pixels in the region. Select the surface point in the three-dimensional surface model that is closest to the position represented by the average of the three coordinate components and record this surface point as the center surface point of the region. Record the shortest path length between the center surface points of two image regions along the three-dimensional surface model as the surface migration distance. Obtain historical valid test records for similar parts; historical valid test records include part identification, substrate material, coating type, coating color, batch of treatment material, surface treatment equipment identification, coating treatment completion time, test images, equipment operating parameters, and test results.

[0007] Preferably, the cross-polarized image is input into the image segmentation model, and the image feature size is reduced sequentially to extract edge features, texture features and region shape features. When restoring the image feature size, shallow features and deep features of the same size are fused to obtain the final-level image features with the same size as the cross-polarized image. The brightness difference between parallel polarized and cross-polarized images at the same pixel position is calculated to form a polarization difference map. The minimum pixel value of the polarization difference map is subtracted from each pixel value, and then divided by the difference between the maximum and minimum pixel values ​​to obtain the polarization response value. When the maximum and minimum pixel values ​​are equal, the polarization response value is set to zero. The polarization response value is multiplied pixel by pixel with the final-level image features, and then added pixel by pixel with the final-level image features. The sum is input into the pixel classification layer to output candidate anomaly regions. Anomalies in the first illumination image are extracted using an image segmentation model. The anomaly region with the largest overlapping area with the projection position of the candidate anomaly region is selected as the first candidate anomaly region. If no overlapping region is obtained, a record to be optically verified is formed. Centered on the surface point of the first candidate anomaly region, the surface range whose shortest path length along the three-dimensional surface model between it and the surface point of the first candidate anomaly region is no greater than the maximum surface migration distance in the historical valid detection records where the detection result is reflective interference is taken as the reflective search area. Anomalies in the second illumination image are extracted using an image segmentation model, and anomalies located within the reflection search area are read. For each anomaly read, the incident direction is determined based on the second illumination position and the center surface point of the anomaly. The incident direction is mirrored with respect to the surface normal of the center surface point to obtain the predicted reflection direction. The specular reflection angle between the predicted reflection direction and the direction from the center surface point of the anomaly to the camera optical center is calculated. Anomalies with specular reflection angles greater than the maximum specular reflection angle in the historical valid detection records where the detection result indicates reflection interference are deleted.

[0008] Preferably, the surface migration distance between each remaining abnormal region and the first candidate abnormal region is calculated, and the abnormal region with the smallest absolute value of the difference between the surface migration distance and the median surface migration distance in the historical valid detection records with the detection result of reflective interference is selected as the second candidate abnormal region; if the absolute values ​​are the same, the abnormal region with the smallest specular reflection angle is selected; if there are no remaining abnormal regions, a record to be optically verified is formed. Images of the first and second candidate anomaly regions are cropped respectively. After adjusting the two images to the same size, they are input into the feature extraction part of the image segmentation model. The image features output before restoring the image feature size are read. The two image features are converted into feature vectors of the same length, and the Euclidean distance between the two feature vectors is calculated to obtain the illumination feature distance. The polarization brightness attenuation ratio is obtained by subtracting the average brightness of the same position in the cross-polarized image from the average brightness of the first candidate anomalous region in the parallel polarized image, and then dividing by the average brightness of the first candidate anomalous region in the parallel polarized image. The minimum values ​​of polarization brightness attenuation ratio, minimum illumination feature distance, and minimum surface migration distance in the historical valid detection records with the detection result of reflective interference are respectively taken as the lower limit of reflective attenuation ratio, lower limit of reflective feature distance, and lower limit of reflective migration distance. The maximum value of specular reflection angle is taken as the upper limit of reflective angle. The maximum value of illumination feature distance in the historical valid detection records with the detection result of solid defect is taken as the upper limit of solid feature distance. When the polarization brightness attenuation ratio is not less than the lower limit of the reflection attenuation ratio, the illumination feature distance is not less than the lower limit of the reflection feature distance, the surface migration distance is not less than the lower limit of the reflection migration distance, and the specular reflection angle is not greater than the upper limit of the reflection angle, a reflection interference record is formed; when the polarization brightness attenuation ratio is not less than the lower limit of the reflection attenuation ratio, the illumination feature distance is not less than the lower limit of the reflection feature distance, the surface migration distance is not less than the lower limit of the reflection migration distance, the specular reflection angle is not greater than the upper limit of the reflection angle, the illumination feature distance is not greater than the upper limit of the entity feature distance, and the surface migration distance is not greater than the upper limit of the image positioning error, a record pending environmental interference judgment is formed; all other cases form a record pending optical verification.

[0009] Preferably, for the record to be judged by environmental interference, the boundary of the candidate abnormal area is extended outward along the three-dimensional surface by the effective measurement diameter of the film thickness detection probe, the part that overlaps with other candidate abnormal areas is deleted, and the remaining area is regarded as the neighboring normal area; if there is no remaining area, the surface area that is closest to the candidate abnormal area and has not been identified as abnormal is selected; if there is still no usable area, a record to be optically verified is formed. The surface dew point margin is obtained by subtracting the dew point temperature from the surface temperature of the ship fittings. When the surface dew point margin is between the minimum and maximum values ​​of the surface dew point margin in the historical valid test records where the test results show condensation interference, controlled heating is performed. When it is not between the two limits, a record of the pending attachment state is formed. When performing controlled heating, the median of the heating amplitude and holding time after heating in the historical valid test records with qualified heating test results are used as the heating amplitude and holding time for each stage, respectively. The lower of the upper limit of the allowable test temperature and the highest surface temperature value in the historical valid test records is used as the heating stop temperature. After each stage of heating, the positions of the camera, ship accessories and lighting are kept unchanged, and the heating response image is acquired when the difference between the maximum and minimum values ​​of three consecutive surface temperature readings is not greater than the maximum value of the repeated measurement error of the temperature sensor. The candidate abnormal regions before heating are projected onto the heating response image, and the abnormal region with the largest overlapping area is taken as the candidate abnormal region after heating. The reduction ratio of the abnormal area is obtained by subtracting the area of ​​the candidate abnormal area after the heating from the area of ​​the candidate abnormal area before the heating, and then dividing by the area of ​​the candidate abnormal area before the heating. The shortest surface distance from each point on the boundary of the candidate abnormal area after the heating to the boundary of the candidate abnormal area before the heating is calculated. The distance inside the candidate abnormal area before the heating is recorded as a positive value, and the distance outside is recorded as a negative value. The average value of the corresponding distances at each boundary point is taken to obtain the boundary inward shift. The reduction of the brightness difference is obtained by subtracting the corresponding brightness difference after the heating from the brightness difference between the abnormal area before the heating and the adjacent normal area. Divide the abnormal area reduction ratio, boundary inward shift, and brightness difference reduction by the current temperature rise value to obtain the area reduction ratio per unit temperature rise, boundary inward shift per unit temperature rise, and brightness difference reduction per unit temperature rise. When all three unit temperature rise data are between the minimum and maximum values ​​of the corresponding data in the historical valid test records where the test result is condensation interference, a condensation interference record is formed and a delayed retest task is generated; if the three conditions are not met simultaneously and the surface temperature is lower than the heating stop temperature, the heating continues; if the three conditions are still not met simultaneously after reaching the heating stop temperature, the heating stops and waits for the temperature to recover. If the difference between the current surface temperature and the surface temperature before heating is not greater than the maximum value of the repeated measurement error of the temperature sensor, keep the camera and illumination positions unchanged and re-acquire cross-polarized images; input the re-acquired cross-polarized images into the image segmentation model to extract abnormal regions; if there are extracted abnormal regions within the three-dimensional surface range corresponding to the candidate abnormal regions before heating, a pending attachment state judgment record is formed; if there are no extracted abnormal regions, a delayed re-inspection task is generated; if no abnormality is found during the delayed re-inspection, condensation interference is confirmed, and a pending attachment state judgment record is formed when an abnormality reappears.

[0010] Preferably, for the record of the state to be attached, the ratio of the area of ​​the candidate abnormal region to the area of ​​the minimum circumscribed convex hull of the candidate abnormal region is calculated to obtain the region compactness; the standard deviation of the gray value of each pixel in the candidate abnormal region is calculated to obtain the region gray standard deviation. When the area compactness is between the minimum and maximum values ​​of the area compactness in the historical valid test records where the test result is loose attachment, and the area grayscale standard deviation is between the minimum and maximum values ​​of the area grayscale standard deviation in the historical valid test records, airflow disturbance is performed; if neither condition is met simultaneously, a stable entity anomaly record is formed. When performing airflow disturbance, the median of the distance from the airflow nozzle to the surface of the ship's fittings, the airflow action time, and the pressure difference between two adjacent airflow stages in the historical valid test records with the test result of passing the purging test are respectively used as the current nozzle distance, the single action time, and the pressure increase per stage. The minimum airflow pressure that can cause displacement of the abnormal area in the historical valid test records with the test result of loose attachments is used as the initial airflow pressure. Orient the airflow nozzle toward the center surface point of the candidate anomaly area, and make the angle between the airflow direction and the surface normal of the center surface point of the area 90 degrees. Apply directional airflow to the candidate anomaly area. After each airflow stops, acquire the airflow response image. The airflow pressure does not exceed the upper limit of the allowable purging pressure. Candidate anomaly regions before airflow action are projected onto the airflow response image, and the anomaly region with the largest overlapping area is taken as the candidate anomaly region after airflow action; if no overlapping area is obtained, a loose attachment record is formed. When there are candidate anomaly regions after the airflow effect, the surface migration distance between the two candidate anomaly regions before and after the airflow effect is taken as the displacement of the anomaly center; the area of ​​the candidate anomaly region before the airflow effect is subtracted from the area of ​​the candidate anomaly region after the airflow effect, and then divided by the area of ​​the candidate anomaly region before the airflow effect to obtain the anomaly area reduction ratio; the overlapping area of ​​the two candidate anomaly regions is divided by the area of ​​the candidate anomaly region before the airflow effect to obtain the region overlap ratio. The minimum value of abnormal center displacement, the minimum value of abnormal area reduction ratio, and the maximum value of region overlap ratio in the historical valid detection records of loose attachments are respectively taken as the lower limit of loose displacement, the lower limit of loose area reduction ratio, and the upper limit of loose overlap ratio; the maximum value of abnormal center displacement, the maximum value of abnormal area reduction ratio, and the minimum value of region overlap ratio in the historical valid detection records of solid defects are respectively taken as the upper limit of solid displacement, the upper limit of solid area reduction ratio, and the lower limit of solid overlap ratio. A record of loose attachments is formed when at least one of the following conditions is met: the abnormal center displacement is not less than the lower limit of loose displacement and the abnormal area reduction ratio is not less than the lower limit of loose area reduction ratio, and the area overlap ratio is not greater than the upper limit of loose overlap ratio. A stable entity anomaly record is formed when no loose attachment record is formed, and the displacement of the abnormal center is not greater than the upper limit of the entity displacement, the reduction ratio of the abnormal area is not greater than the upper limit of the entity area reduction ratio, and the regional overlap ratio is not less than the lower limit of the entity overlap ratio. When the regional overlap ratio is greater than the upper limit of loose overlap ratio but less than the lower limit of solid overlap ratio, the portion obtained by subtracting the overlapping region from the candidate anomaly region before the airflow action is recorded as loose attachment, and the overlapping region is recorded as a stable solid anomaly; in other cases, a pending attachment status verification record is formed.

[0011] Preferably, for stable entity anomaly records, film thickness detection points are arranged in the stable entity anomaly area and the adjacent normal area, and the distance between adjacent detection points along the three-dimensional surface is no greater than half of the effective measurement diameter of the film thickness detection probe; with each detection point as the center, a film thickness detection coverage area is generated with half of the effective measurement diameter as the surface distance radius; The minimum and maximum values ​​of the film thickness test results in the historical valid test records with qualified film thickness are respectively used as the lower limit and upper limit of the qualified film thickness range; the overlapping coverage areas where the film thickness is all below the lower limit of the qualified film thickness range are merged, the overlapping coverage areas where the film thickness is all above the upper limit of the qualified film thickness range are merged, and the overlapping coverage areas where the film thickness is all within the qualified film thickness range are merged. Areas with film thickness below the lower limit of the qualified film thickness range are merged into areas with low film thickness, and areas with film thickness above the upper limit of the qualified film thickness range are merged into areas with high film thickness. The overlap ratio of the low film thickness area is obtained by dividing the overlapping area of ​​the stable entity abnormal area and the low film thickness area by the area of ​​the stable entity abnormal area. The overlap ratio of the high film thickness area is obtained by dividing the overlapping area of ​​the stable entity abnormal area and the high film thickness area by the area of ​​the stable entity abnormal area. The film thickness difference is obtained by subtracting the median film thickness of the adjacent normal area from the median film thickness of the stable entity abnormal area. The minimum overlap ratio of low-value areas in historical valid inspection records where the inspection result is a solid defect and the solid defect type is insufficient coating coverage is taken as the lower limit of the overlap ratio for insufficient coverage; the minimum overlap ratio of high-value areas in historical valid inspection records where the inspection result is a solid defect and the solid defect type is local coating accumulation is taken as the lower limit of the overlap ratio for local accumulation. When the overlap ratio of the low-coverage area is not less than the lower limit of the overlap ratio for insufficient coverage and the film thickness difference is less than zero, a record of insufficient coating coverage is formed; when the overlap ratio of the high-coverage area is not less than the lower limit of the overlap ratio for local accumulation and the film thickness difference is greater than zero, a record of local coating accumulation is formed; when all film thickness test results within the abnormal area of ​​a stable entity are within the acceptable film thickness range, a record of entity defect pending verification is formed; in other cases, a record of film thickness verification pending verification is formed and film thickness test points are added. Standard samples refer to samples whose substrate material, coating type, and coating color are the same as those of ship fittings, and which have been confirmed to be qualified by appearance inspection and film thickness inspection. The same batch of accessories processed by the same surface treatment equipment after the most recent standard sample film thickness test passed the test are arranged according to the coating treatment completion time to form the same batch test sequence; when adjacent records show the same defect type and the same positive or negative sign of the film thickness difference, the number of consecutive records is accumulated; when there are qualified records, records with different defect types, or records with different positive or negative signs of film thickness difference, the count is reset. The minimum number of consecutive records in the historical valid inspection records where the equipment abnormality was confirmed is taken as the lower limit of the number of consecutive records of equipment abnormalities. When the current number of consecutive records reaches the lower limit of the number of consecutive records of equipment abnormalities, a common equipment abnormality record is formed. If it does not reach the lower limit, a defect record of entity to be associated is formed. When there is no historical valid inspection record where the equipment abnormality is confirmed, a defect record of entity to be associated is formed directly. The latest sample verification time is obtained by subtracting the average processing time of standard samples and the average film thickness detection time in the historical valid inspection records from the start time of the next production process of the current ship parts. If the subsequent inspection records are still insufficient to make the number of consecutive records reach the lower limit of the number of consecutive equipment abnormalities when the latest sample verification time is reached, the standard sample is processed using the current batch of processed materials and the current equipment operating parameters. When standard samples exhibit the same defect type and have the same positive or negative sign for the film thickness difference, a common equipment anomaly record is generated. When no samples exhibit the same defect type, a single-sample entity defect record is generated. When samples exhibit the same defect type but have different positive or negative signs for the film thickness difference, a record awaiting equipment verification is generated.

[0012] Preferably, an illumination adjustment task and a supplementary acquisition task are generated based on the reflection interference record; when the anomaly disappears after supplementary acquisition, the reflection interference is confirmed; when the surface migration distance between the two abnormal areas before and after supplementary acquisition is not greater than the upper limit of the image positioning error, it is passed to S3. An optical verification task is generated based on the optical verification record; the task ends when the reflection is confirmed during verification; if the surface migration distance between the two abnormal areas before and after verification is not greater than the upper limit of the image positioning error, it is passed to S3; if it cannot be confirmed, the corresponding ship parts are prohibited from entering the next production process. Based on the condensation interference record, a delayed re-inspection task is generated. When the abnormality disappears after the re-inspection, the condensation interference is confirmed. If the abnormality still exists, the cross-polarization image is re-acquired and transmitted to S4. Surface cleaning tasks and post-cleaning data collection tasks are generated based on the loose attachment records; when the anomaly disappears after cleaning, the loose attachments are confirmed; if the anomaly still exists, a stable entity anomaly record is formed and transmitted to S5. Generate an attachment status verification task based on the attachment status verification record; perform a surface cleaning task when loose attachments are confirmed; if the surface migration distance between the two abnormal areas before and after verification is not greater than the upper limit of the image positioning error, transfer it to S5; if it cannot be confirmed, prohibit the corresponding ship parts from entering the next production process. Based on records of insufficient coating coverage, localized coating buildup, and single-piece defects, local supplementary treatment tasks, localized repair tasks, and localized rework tasks are generated, and film thickness retesting is performed after processing. Detection points are added and retested based on the film thickness verification records. Quality verification tasks are generated based on the physical defect records. When the quality verification confirms that there are no physical defects, a release record is generated. When physical defects are confirmed, local supplementary treatment tasks, localized repair tasks, or localized rework tasks are generated based on the confirmed defect type. All parts from the same batch processed by the same equipment after the most recent standard sample film thickness test passed are recorded as related parts; an equipment verification task is generated based on the common equipment anomaly record, and the equipment is suspended from receiving subsequent parts from the same batch. For each piece of equipment operating parameter, if the operating parameter is less than the minimum value of the corresponding operating parameter in the historical valid test records where the test result indicates that the equipment is operating qualified, the operating parameter of the equipment is adjusted to the median of the corresponding operating parameter; if the operating parameter is greater than the maximum value of the corresponding operating parameter, the operating parameter of the equipment is adjusted to the median of the corresponding operating parameter; the adjusted operating parameters are then used to process the standard sample. If the appearance inspection results and film thickness inspection results of the standard sample are both qualified, the equipment operation is resumed and the associated accessories are re-inspected; if the standard sample is unqualified and the associated accessories continue to show the same common equipment abnormality, the equipment is kept paused and an equipment maintenance task is generated. When the film thickness is retested and found to be qualified, a release record for the next production process is generated; if it is still not qualified, a quality review task is generated. When the quality review confirms that rework can continue, a corresponding local processing task is regenerated; when it is confirmed that rework cannot continue, a non-conforming product disposal task is generated. Tasks are executed in the following order: lighting adjustment before supplementary data collection, delayed waiting before re-inspection, surface cleaning before post-cleaning data collection, local treatment before film thickness re-measurement, and equipment verification before release of related accessories. Tasks in which all preceding tasks are completed are executed from earliest to latest according to the start time of the next production process. During the current batch of processed materials, the lower limit, upper limit, and median of judgment determined in S2 to S5 based on historical valid test records remain unchanged; after the current batch of processed materials ends, the current test record with complete corresponding test data, which has obtained one of the following: final verification result, film thickness retest result, or equipment verification result, will be added to the historical valid test record. When newly added test data is less than the current minimum value or greater than the current maximum value, it will be added to the historical valid test record after manual verification confirms that both the test data and the test results are valid.

[0013] (III) Beneficial Effects This invention provides an intelligent integrated method for identifying and scheduling surface treatment defects in ship components, which has the following advantages: 1. Candidate abnormal regions are extracted by parallel polarization images, cross polarization images, first illumination images and second illumination images. The polarization brightness attenuation ratio, illumination feature distance, surface migration distance and specular reflection angle are combined to judge the reflection interference, thereby reducing the situation where reflective areas are identified as physical defects.

[0014] 2. Dew point margin and three unit temperature rise data are used to determine condensation interference, and regional compactness, regional gray scale standard deviation and abnormal regional changes before and after airflow are used to determine loose attachments, so that condensation interference, loose attachments and stable entity anomalies can be distinguished in turn.

[0015] 3. Perform film thickness detection on stable entities with abnormal areas. Based on the overlap ratio of low-area and high-area areas and the film thickness difference, determine insufficient coating coverage and local coating accumulation. Combine the number of continuous records and standard samples to determine common equipment anomalies, and then generate corresponding processing tasks, equipment verification tasks and release records to achieve the connection between defect identification results and task execution. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the process for a smart integrated method for identifying and scheduling surface treatment defects in ship parts according to the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Please see Figure 1 This invention provides an intelligent integrated method for identifying and scheduling surface treatment defects in ship parts, comprising the following steps: S1. Obtain accessory identification, 3D surface model, equipment operating parameters and historical valid test records, and acquire parallel polarization image, cross polarization image, first illumination image and second illumination image.

[0019] Obtain the component identification, 3D surface model, substrate material, coating type, coating color, surface treatment equipment identification, treatment material batch, coating completion time, and start time of the next production process for ship components; Acquire the lateral focal length, camera longitudinal focal length, principal point x-coordinate, principal point y-coordinate, 3D coordinates and surface normals of each surface point of the 3D surface model in the camera coordinate system, effective measurement diameter of the film thickness detection probe, upper limit of image positioning error, maximum value of repeated measurement error of temperature sensor, upper limit of allowable detection temperature, and upper limit of allowable purging pressure. Obtain the equipment operating parameters corresponding to the ship parts. The equipment operating parameters include the material supply pressure from the start of outputting coating material to the stop of outputting coating material at the material output end, the material output per unit time, the distance between the material output end and the surface of the ship parts, and the moving speed of the material output end relative to the surface of the ship parts. Ship parts with the same base material, coating type and coating color are classified as similar parts; ship parts with the same batch of treated materials among similar parts are classified as parts from the same batch. The camera is controlled to sequentially photograph the surface of the ship parts, and the surface area of ​​the ship parts corresponding to each photograph is recorded as an image acquisition area. Each image acquisition area is set sequentially, with overlapping portions between adjacent areas, until all image acquisition areas cover the surface of the ship parts to be inspected. For each image acquisition area, keeping the position of the camera and the ship parts unchanged, the polarization direction of the light source is adjusted to be parallel to the transmission direction of the analyzer, and a parallel polarized image is acquired. Then, the polarization direction of the light source is adjusted to be perpendicular to the transmission direction of the analyzer, and a cross-polarized image is acquired. Keeping the polarization direction of the light source perpendicular to the transmission direction of the analyzer, the light source is moved to the first illumination position and the second illumination position, respectively. First and second illumination images are acquired for the same image acquisition area, and the three-dimensional coordinates of the light source center at the first and second illumination positions in the camera coordinate system are recorded. Simultaneously, the surface temperature and dew point temperature of the ship parts are acquired. For each pixel in each image, subtract the principal point's x-coordinate from the pixel's x-coordinate and then divide by the camera's lateral focal length to obtain the ray's lateral component; subtract the principal point's ordinate from the pixel's ordinate and then divide by the camera's longitudinal focal length to obtain the ray's longitudinal component; set the ray's forward component to one, and determine the ray's direction originating from the camera's optical center using the lateral, longitudinal, and forward components. Extending from the camera optical center along the ray direction to the three-dimensional surface model, the intersection point of the ray and the three-dimensional surface model that is closest to the camera optical center is determined as the surface point corresponding to the pixel, and the three-dimensional coordinates and surface normal of the surface point in the camera coordinate system are read as the three-dimensional surface position and surface normal corresponding to the pixel. For any region in the image, calculate the average of the three coordinate components of the three-dimensional coordinates of all pixels in the region. Select the surface point in the three-dimensional surface model that is closest to the position represented by the average of the three coordinate components and record this surface point as the center surface point of the region. Record the shortest path length between the center surface points of two image regions along the three-dimensional surface model as the surface migration distance. Obtain historical valid test records corresponding to similar accessories; historical valid test records refer to records in which one of the following has been obtained: final verification result, film thickness retest result, and equipment verification result, and the corresponding test data is complete; Historical valid inspection records include accessory identification, substrate material, coating type, coating color, batch of treatment materials, surface treatment equipment identification, coating treatment completion time, inspection images, equipment operating parameters, and inspection results; they also include the polarization brightness attenuation ratio, illumination characteristic distance, specular reflection angle, surface migration distance, surface dew point margin, temperature rise, holding time after temperature rise, highest surface temperature, reduction ratio of area per unit temperature rise, inward displacement of boundary per unit temperature rise, reduction of brightness difference per unit temperature rise, regional compactness, regional grayscale standard deviation, distance from airflow nozzle to the surface of the ship accessory, airflow pressure at each level, airflow action time, displacement of abnormal center, reduction ratio of abnormal area, regional overlap ratio, film thickness inspection results, film thickness difference, overlap ratio of low-area areas, overlap ratio of high-area areas, number of consecutive records before equipment abnormality was confirmed, standard sample processing time, film thickness inspection time, and equipment verification results.

[0020] S2. Extract candidate abnormal regions and form reflection interference records and environmental interference judgment records based on polarization brightness attenuation ratio, illumination characteristic distance, surface migration distance and specular reflection angle.

[0021] The cross-polarized image is input into the image segmentation model, and the image feature size is reduced sequentially to extract edge features, texture features and region shape features. When restoring the image feature size, shallow features and deep features of the same size are fused to obtain the final-level image features with the same size as the cross-polarized image. The brightness difference between parallel polarized and cross-polarized images at the same pixel position is calculated to form a polarization difference map. The minimum pixel value of the polarization difference map is subtracted from each pixel value, and then divided by the difference between the maximum and minimum pixel values ​​to obtain the polarization response value. When the maximum and minimum pixel values ​​are equal, the polarization response value is set to zero. The polarization response value is multiplied pixel by pixel with the final-level image features, and then added pixel by pixel with the final-level image features. The sum is input into the pixel classification layer to output candidate anomaly regions. In an optional embodiment, OpenCV is used to read and process parallel polarized and cross-polarized images; the cross-polarized image with manually annotated abnormal regions is divided into training and validation images, and the U-Net segmentation model is trained using the training images; the cross-polarized image is adjusted to the input size of the U-Net segmentation model and then input into the trained U-Net segmentation model. The encoding part progressively reduces the image feature size and extracts the edge, texture, and shape features of the abnormal regions, and the decoding part progressively restores the image feature size. The image features output by the decoding part are then compared with the images of the same size in the encoding part. The features are stitched together to obtain a final-level image feature with the same size as the cross-polarized image. OpenCV is used to read the grayscale values ​​at the same pixel positions in the parallel polarization and cross-polarization images. The grayscale values ​​of the parallel polarization image are subtracted from the grayscale values ​​of the cross-polarization image to obtain a polarization difference map. Each pixel value in the polarization difference map is subtracted from the minimum pixel value, and then divided by the difference between the maximum and minimum pixel values ​​to obtain a polarization response map with values ​​ranging from zero to one. When the maximum and minimum pixel values ​​in the polarization difference map are equal, each pixel value in the polarization response map is set to zero. The final-level image features are then processed according to their characteristics. The polarization response map is replicated using multiple channels, ensuring it has the same size as the final-level image features. The replicated polarization response map is then multiplied pixel-by-pixel with the final-level image features, and the result is added pixel-by-pixel to obtain the polarization-enhanced image features. These features are then input into the single-channel convolutional and activation layers at the end of the U-Net segmentation model to obtain the probability that each pixel belongs to an abnormal region. The different probability values ​​output by the model in the validation image are sequentially used as candidate judgment values. For each candidate judgment value, pixels with a probability not less than that value are marked as abnormal pixels. The abnormal pixels identified by the model are then compared with... The overlap ratio of abnormal regions is obtained by dividing twice the number of pixels obtained by the intersection of manually labeled abnormal pixels by the sum of the number of abnormal pixels divided by the model and the number of manually labeled abnormal pixels. The candidate judgment value with the largest overlap ratio is used as the abnormal probability judgment value. When there are two or more candidate judgment values ​​corresponding to the same maximum abnormal region overlap ratio, the candidate judgment value with the largest value is selected as the abnormal probability judgment value. Pixels with a probability not less than the abnormal probability judgment value in the cross-polarized image to be detected are recorded as abnormal pixels. Adjacent abnormal pixels are merged, and each connected region formed after merging is used as a candidate abnormal region. Anomalies in the first illumination image are extracted using an image segmentation model. The anomaly region with the largest overlapping area with the projection position of the candidate anomaly region is selected as the first candidate anomaly region. If no overlapping region is obtained, a record to be optically verified is formed. Centered on the surface point of the first candidate anomaly region, the surface range whose shortest path length along the three-dimensional surface model between it and the surface point of the first candidate anomaly region is no greater than the maximum surface migration distance in the historical valid detection records where the detection result is reflective interference is taken as the reflective search area. Anomalies in the second illumination image are extracted using an image segmentation model, and anomalies located within the reflection search area are read. For each anomaly read, the incident direction is determined based on the second illumination position and the center surface point of the anomaly. The incident direction is mirrored with respect to the surface normal of the center surface point to obtain the predicted reflection direction. The specular reflection angle between the predicted reflection direction and the direction from the center surface point of the anomaly to the camera optical center is calculated. Anomalies with specular reflection angles greater than the maximum specular reflection angle in the historical valid detection records where the detection result indicates reflection interference are deleted. Calculate the surface migration distance between each remaining abnormal region and the first candidate abnormal region. The abnormal region with the smallest absolute value of the difference between the surface migration distance and the median surface migration distance in the historical valid detection records where the detection result is reflective interference is selected as the second candidate abnormal region. If the absolute values ​​are the same, select the abnormal region with the smallest specular reflection angle. If there are no remaining abnormal regions, form a record to be optically verified. Images of the first and second candidate anomaly regions are cropped respectively. After adjusting the two images to the same size, they are input into the feature extraction part of the image segmentation model. The image features output before restoring the image feature size are read. The two image features are converted into feature vectors of the same length, and the Euclidean distance between the two feature vectors is calculated to obtain the illumination feature distance. The polarization brightness attenuation ratio is obtained by subtracting the average brightness of the same position in the cross-polarized image from the average brightness of the first candidate anomalous region in the parallel polarized image, and then dividing by the average brightness of the first candidate anomalous region in the parallel polarized image. The minimum values ​​of polarization brightness attenuation ratio, minimum illumination feature distance, and minimum surface migration distance in the historical valid detection records with the detection result of reflective interference are respectively taken as the lower limit of reflective attenuation ratio, lower limit of reflective feature distance, and lower limit of reflective migration distance. The maximum value of specular reflection angle is taken as the upper limit of reflective angle. The maximum value of illumination feature distance in the historical valid detection records with the detection result of solid defect is taken as the upper limit of solid feature distance. When the polarization brightness attenuation ratio is not less than the lower limit of the reflection attenuation ratio, the illumination feature distance is not less than the lower limit of the reflection feature distance, the surface migration distance is not less than the lower limit of the reflection migration distance, and the specular reflection angle is not greater than the upper limit of the reflection angle, a reflection interference record is formed; when the polarization brightness attenuation ratio is not less than the lower limit of the reflection attenuation ratio, the illumination feature distance is not less than the lower limit of the reflection feature distance, the surface migration distance is not less than the lower limit of the reflection migration distance, the specular reflection angle is not greater than the upper limit of the reflection angle, the illumination feature distance is not greater than the upper limit of the entity feature distance, and the surface migration distance is not greater than the upper limit of the image positioning error, a record pending environmental interference judgment is formed; all other cases form a record pending optical verification.

[0022] S3. For the environmental interference judgment record, a condensation interference record and an adhesion status judgment record are formed based on the surface dew point balance and three unit temperature rise data.

[0023] For records to be judged by environmental interference, the boundary of the candidate abnormal area is extended outward along the three-dimensional surface by the effective measurement diameter of the film thickness detection probe. The part that overlaps with other candidate abnormal areas is deleted, and the remaining area is regarded as the neighboring normal area. If there is no remaining area, the surface area that is closest to the candidate abnormal area and has not been identified as abnormal is selected. If there is still no usable area, a record to be optically verified is formed. The surface dew point margin is obtained by subtracting the dew point temperature from the surface temperature of the ship fittings. When the surface dew point margin is between the minimum and maximum values ​​of the surface dew point margin in the historical valid test records where the test results show condensation interference, controlled heating is performed. When it is not between the two limits, a record of the pending attachment state is formed. When performing controlled heating, the median of the heating amplitude and holding time after heating in the historical valid test records with qualified heating test results are used as the heating amplitude and holding time for each stage, respectively. The lower of the upper limit of the allowable test temperature and the highest surface temperature value in the historical valid test records is used as the heating stop temperature. After each stage of heating, the positions of the camera, ship accessories and lighting are kept unchanged, and the heating response image is acquired when the difference between the maximum and minimum values ​​of three consecutive surface temperature readings is not greater than the maximum value of the repeated measurement error of the temperature sensor. The candidate abnormal regions before heating are projected onto the heating response image, and the abnormal region with the largest overlapping area is taken as the candidate abnormal region after heating. The reduction ratio of the abnormal area is obtained by subtracting the area of ​​the candidate abnormal area after the heating from the area of ​​the candidate abnormal area before the heating, and then dividing by the area of ​​the candidate abnormal area before the heating. The shortest surface distance from each point on the boundary of the candidate abnormal area after the heating to the boundary of the candidate abnormal area before the heating is calculated. The distance inside the candidate abnormal area before the heating is recorded as a positive value, and the distance outside is recorded as a negative value. The average value of the corresponding distances at each boundary point is taken to obtain the boundary inward shift. The reduction of the brightness difference is obtained by subtracting the corresponding brightness difference after the heating from the brightness difference between the abnormal area before the heating and the adjacent normal area. Divide the abnormal area reduction ratio, boundary inward shift, and brightness difference reduction by the current temperature rise value to obtain the area reduction ratio per unit temperature rise, boundary inward shift per unit temperature rise, and brightness difference reduction per unit temperature rise. When all three unit temperature rise data are between the minimum and maximum values ​​of the corresponding data in the historical valid test records where the test result is condensation interference, a condensation interference record is formed and a delayed retest task is generated; if the three conditions are not met simultaneously and the surface temperature is lower than the heating stop temperature, the heating continues; if the three conditions are still not met simultaneously after reaching the heating stop temperature, the heating stops and waits for the temperature to recover. If the difference between the current surface temperature and the surface temperature before heating is not greater than the maximum value of the repeated measurement error of the temperature sensor, keep the camera and illumination positions unchanged and re-acquire cross-polarized images; input the re-acquired cross-polarized images into the image segmentation model to extract abnormal regions; if there are extracted abnormal regions within the three-dimensional surface range corresponding to the candidate abnormal regions before heating, a pending attachment state judgment record is formed; if there are no extracted abnormal regions, a delayed re-inspection task is generated; if no abnormality is found during the delayed re-inspection, condensation interference is confirmed, and a pending attachment state judgment record is formed when an abnormality reappears.

[0024] S4. For the record of the state to be attached, the records of loose attachments and stable entities are formed based on the regional compactness, regional gray standard deviation, abnormal center displacement, abnormal area reduction ratio and regional overlap ratio.

[0025] For the record of the state to be attached, calculate the ratio of the area of ​​the candidate abnormal region to the area of ​​the minimum circumscribed convex hull of the candidate abnormal region to obtain the region compactness; calculate the standard deviation of the gray value of each pixel in the candidate abnormal region to obtain the region gray standard deviation. When the area compactness is between the minimum and maximum values ​​of the area compactness in the historical valid test records where the test result is loose attachment, and the area grayscale standard deviation is between the minimum and maximum values ​​of the area grayscale standard deviation in the historical valid test records, airflow disturbance is performed; if neither condition is met simultaneously, a stable entity anomaly record is formed. When performing airflow disturbance, the median of the distance from the airflow nozzle to the surface of the ship's fittings, the airflow action time, and the pressure difference between two adjacent airflow stages in the historical valid test records with the test result of passing the purging test are respectively used as the current nozzle distance, the single action time, and the pressure increase per stage. The minimum airflow pressure that can cause displacement of the abnormal area in the historical valid test records with the test result of loose attachments is used as the initial airflow pressure. Orient the airflow nozzle toward the center surface point of the candidate anomaly area, and make the angle between the airflow direction and the surface normal of the center surface point of the area 90 degrees. Apply directional airflow to the candidate anomaly area. After each airflow stops, acquire the airflow response image. The airflow pressure does not exceed the upper limit of the allowable purging pressure. Candidate anomaly regions before airflow action are projected onto the airflow response image, and the anomaly region with the largest overlapping area is taken as the candidate anomaly region after airflow action; if no overlapping area is obtained, a loose attachment record is formed. When there are candidate anomaly regions after the airflow effect, the surface migration distance between the two candidate anomaly regions before and after the airflow effect is taken as the displacement of the anomaly center; the area of ​​the candidate anomaly region before the airflow effect is subtracted from the area of ​​the candidate anomaly region after the airflow effect, and then divided by the area of ​​the candidate anomaly region before the airflow effect to obtain the anomaly area reduction ratio; the overlapping area of ​​the two candidate anomaly regions is divided by the area of ​​the candidate anomaly region before the airflow effect to obtain the region overlap ratio. The minimum value of abnormal center displacement, the minimum value of abnormal area reduction ratio, and the maximum value of region overlap ratio in the historical valid detection records of loose attachments are respectively taken as the lower limit of loose displacement, the lower limit of loose area reduction ratio, and the upper limit of loose overlap ratio; the maximum value of abnormal center displacement, the maximum value of abnormal area reduction ratio, and the minimum value of region overlap ratio in the historical valid detection records of solid defects are respectively taken as the upper limit of solid displacement, the upper limit of solid area reduction ratio, and the lower limit of solid overlap ratio. A record of loose attachments is formed when at least one of the following conditions is met: the abnormal center displacement is not less than the lower limit of loose displacement and the abnormal area reduction ratio is not less than the lower limit of loose area reduction ratio, and the area overlap ratio is not greater than the upper limit of loose overlap ratio. A stable entity anomaly record is formed when no loose attachment record is formed, and the displacement of the abnormal center is not greater than the upper limit of the entity displacement, the reduction ratio of the abnormal area is not greater than the upper limit of the entity area reduction ratio, and the regional overlap ratio is not less than the lower limit of the entity overlap ratio. When the regional overlap ratio is greater than the upper limit of loose overlap ratio but less than the lower limit of solid overlap ratio, the portion obtained by subtracting the overlapping region from the candidate anomaly region before the airflow action is recorded as loose attachment, and the overlapping region is recorded as a stable solid anomaly; in other cases, a pending attachment status verification record is formed.

[0026] S5. For stable entity anomaly records, based on film thickness detection results, low overlap ratio, high overlap ratio, and film thickness difference, form records of insufficient coating coverage, local coating accumulation, and common equipment anomaly records.

[0027] For stable entity anomaly records, film thickness detection points are arranged in the stable entity anomaly area and the adjacent normal area. The distance between adjacent detection points along the three-dimensional surface is no greater than half of the effective measurement diameter of the film thickness detection probe. With each detection point as the center, a film thickness detection coverage area is generated with half of the effective measurement diameter as the surface distance radius. The minimum and maximum values ​​of the film thickness test results in the historical valid test records with qualified film thickness are respectively used as the lower limit and upper limit of the qualified film thickness range; the overlapping coverage areas where the film thickness is all below the lower limit of the qualified film thickness range are merged, the overlapping coverage areas where the film thickness is all above the upper limit of the qualified film thickness range are merged, and the overlapping coverage areas where the film thickness is all within the qualified film thickness range are merged. Areas with film thickness below the lower limit of the qualified film thickness range are merged into areas with low film thickness, and areas with film thickness above the upper limit of the qualified film thickness range are merged into areas with high film thickness. The overlap ratio of the low film thickness area is obtained by dividing the overlapping area of ​​the stable entity abnormal area and the low film thickness area by the area of ​​the stable entity abnormal area. The overlap ratio of the high film thickness area is obtained by dividing the overlapping area of ​​the stable entity abnormal area and the high film thickness area by the area of ​​the stable entity abnormal area. The film thickness difference is obtained by subtracting the median film thickness of the adjacent normal area from the median film thickness of the stable entity abnormal area. The minimum overlap ratio of low-value areas in historical valid inspection records where the inspection result is a solid defect and the solid defect type is insufficient coating coverage is taken as the lower limit of the overlap ratio for insufficient coverage; the minimum overlap ratio of high-value areas in historical valid inspection records where the inspection result is a solid defect and the solid defect type is local coating accumulation is taken as the lower limit of the overlap ratio for local accumulation. When the overlap ratio of the low-coverage area is not less than the lower limit of the overlap ratio for insufficient coverage and the film thickness difference is less than zero, a record of insufficient coating coverage is formed; when the overlap ratio of the high-coverage area is not less than the lower limit of the overlap ratio for local accumulation and the film thickness difference is greater than zero, a record of local coating accumulation is formed; when all film thickness test results within the abnormal area of ​​a stable entity are within the acceptable film thickness range, a record of entity defect pending verification is formed; in other cases, a record of film thickness verification pending verification is formed and film thickness test points are added. Standard samples refer to samples whose substrate material, coating type, and coating color are the same as those of ship fittings, and which have been confirmed to be qualified by appearance inspection and film thickness inspection. The same batch of accessories processed by the same surface treatment equipment after the most recent standard sample film thickness test passed the test are arranged according to the coating treatment completion time to form the same batch test sequence; when adjacent records show the same defect type and the same positive or negative sign of the film thickness difference, the number of consecutive records is accumulated; when there are qualified records, records with different defect types, or records with different positive or negative signs of film thickness difference, the count is reset. The minimum number of consecutive records in the historical valid inspection records where the equipment abnormality was confirmed is taken as the lower limit of the number of consecutive records of equipment abnormalities. When the current number of consecutive records reaches the lower limit of the number of consecutive records of equipment abnormalities, a common equipment abnormality record is formed. If it does not reach the lower limit, a defect record of entity to be associated is formed. When there is no historical valid inspection record where the equipment abnormality is confirmed, a defect record of entity to be associated is formed directly. The latest sample verification time is obtained by subtracting the average processing time of standard samples and the average film thickness detection time in the historical valid inspection records from the start time of the next production process of the current ship parts. If the subsequent inspection records are still insufficient to make the number of consecutive records reach the lower limit of the number of consecutive equipment abnormalities when the latest sample verification time is reached, the standard sample is processed using the current batch of processed materials and the current equipment operating parameters. When standard samples exhibit the same defect type and have the same positive or negative sign for the film thickness difference, a common equipment anomaly record is generated. When no samples exhibit the same defect type, a single-sample entity defect record is generated. When samples exhibit the same defect type but have different positive or negative signs for the film thickness difference, a record awaiting equipment verification is generated.

[0028] S6. Generate and execute corresponding tasks based on the records of reflective interference, condensation interference, loose attachments, insufficient coating coverage, local coating accumulation, and common equipment anomalies, and update the historical valid detection records.

[0029] Based on the reflection interference record, an illumination adjustment task and a supplementary acquisition task are generated; when the anomaly disappears after supplementary acquisition, the reflection interference is confirmed; when the surface migration distance between the two abnormal areas before and after supplementary acquisition is not greater than the upper limit of the image positioning error, it is passed to S3. An optical verification task is generated based on the optical verification record; the task ends when the reflection is confirmed during verification; if the surface migration distance between the two abnormal areas before and after verification is not greater than the upper limit of the image positioning error, it is passed to S3; if it cannot be confirmed, the corresponding ship parts are prohibited from entering the next production process. Based on the condensation interference record, a delayed re-inspection task is generated. When the abnormality disappears after the re-inspection, the condensation interference is confirmed. If the abnormality still exists, the cross-polarization image is re-acquired and transmitted to S4. Surface cleaning tasks and post-cleaning data collection tasks are generated based on the loose attachment records; when the anomaly disappears after cleaning, the loose attachments are confirmed; if the anomaly still exists, a stable entity anomaly record is formed and transmitted to S5. Generate an attachment status verification task based on the attachment status verification record; perform a surface cleaning task when loose attachments are confirmed; if the surface migration distance between the two abnormal areas before and after verification is not greater than the upper limit of the image positioning error, transfer it to S5; if it cannot be confirmed, prohibit the corresponding ship parts from entering the next production process. Based on records of insufficient coating coverage, localized coating buildup, and single-piece defects, local supplementary treatment tasks, localized repair tasks, and localized rework tasks are generated, and film thickness retesting is performed after processing. Detection points are added and retested based on the film thickness verification records. Quality verification tasks are generated based on the physical defect records. When the quality verification confirms that there are no physical defects, a release record is generated. When physical defects are confirmed, local supplementary treatment tasks, localized repair tasks, or localized rework tasks are generated based on the confirmed defect type. All parts from the same batch processed by the same equipment after the most recent standard sample film thickness test passed are recorded as related parts; an equipment verification task is generated based on the common equipment anomaly record, and the equipment is suspended from receiving subsequent parts from the same batch. For each piece of equipment operating parameter, if the operating parameter is less than the minimum value of the corresponding operating parameter in the historical valid test records where the test result indicates that the equipment is operating qualified, the operating parameter of the equipment is adjusted to the median of the corresponding operating parameter; if the operating parameter is greater than the maximum value of the corresponding operating parameter, the operating parameter of the equipment is adjusted to the median of the corresponding operating parameter; the adjusted operating parameters are then used to process the standard sample. If the appearance inspection results and film thickness inspection results of the standard sample are both qualified, the equipment operation is resumed and the associated accessories are re-inspected; if the standard sample is unqualified and the associated accessories continue to show the same common equipment abnormality, the equipment is kept paused and an equipment maintenance task is generated. When the film thickness is retested and found to be qualified, a release record for the next production process is generated; if it is still not qualified, a quality review task is generated. When the quality review confirms that rework can continue, a corresponding local processing task is regenerated; when it is confirmed that rework cannot continue, a non-conforming product disposal task is generated. Tasks are executed in the following order: lighting adjustment before supplementary data collection, delayed waiting before re-inspection, surface cleaning before post-cleaning data collection, local treatment before film thickness re-measurement, and equipment verification before release of related accessories. Tasks in which all preceding tasks are completed are executed from earliest to latest according to the start time of the next production process. During the current batch of processed materials, the lower limit, upper limit, and median of judgment determined in S2 to S5 based on historical valid test records remain unchanged; after the current batch of processed materials ends, the current test record with complete corresponding test data, which has obtained one of the following: final verification result, film thickness retest result, or equipment verification result, will be added to the historical valid test record. When newly added test data is less than the current corresponding minimum or greater than the current corresponding maximum, it is added to the historical valid test record after manual verification to confirm that the test data and test results are valid. The updated historical valid test records are summarized according to the test results, and the minimum, maximum and median of each test data are re-determined. The re-determined minimum, maximum and median are used for the next batch of materials to be processed.

[0030] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0031] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0032] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes 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.

Claims

1. A smart integrated method for identifying and scheduling surface treatment defects in ship components, characterized in that, include: S1. Obtain accessory identification, 3D surface model, equipment operating parameters and historical valid test records, and acquire parallel polarization image, cross polarization image, first illumination image and second illumination image; S2. Extract candidate abnormal regions and form reflection interference records and environmental interference judgment records based on polarization brightness attenuation ratio, illumination characteristic distance, surface migration distance and specular reflection angle. S3. For the environmental interference judgment record, a condensation interference record and an adhesion state judgment record are formed based on the surface dew point balance and three unit temperature rise data. S4. For the record of the state to be attached, the records of loose attachments and stable entity anomalies are formed based on the area compactness, area gray standard deviation, abnormal center displacement, abnormal area reduction ratio and area overlap ratio. S5. For stable entity anomaly records, based on film thickness detection results, low area overlap ratio, high area overlap ratio and film thickness difference, form coating insufficient coverage record, coating local accumulation record and equipment common anomaly record. S6. Generate and execute corresponding tasks based on the records of reflective interference, condensation interference, loose attachments, insufficient coating coverage, local coating accumulation, and common equipment anomalies, and update the historical valid detection records.

2. The intelligent integrated method for identifying and scheduling surface treatment defects in ship components according to claim 1, characterized in that: In S1, accessory identification, 3D surface model, equipment operating parameters, and historical valid detection records are acquired. Parallel polarization images, cross polarization images, first illumination images, and second illumination images are also collected, as follows: Obtain the component identification, 3D surface model, substrate material, coating type, coating color, surface treatment equipment identification, treatment material batch, coating completion time, and start time of the next production process for ship components; Acquire the lateral focal length, camera longitudinal focal length, principal point x-coordinate, principal point y-coordinate, 3D coordinates and surface normals of each surface point of the 3D surface model in the camera coordinate system, effective measurement diameter of the film thickness detection probe, upper limit of image positioning error, maximum value of repeated measurement error of temperature sensor, upper limit of allowable detection temperature, and upper limit of allowable purging pressure. Obtain the equipment operating parameters corresponding to the ship parts. The equipment operating parameters include the material supply pressure from the start of outputting coating material to the stop of outputting coating material at the material output end, the material output per unit time, the distance between the material output end and the surface of the ship parts, and the moving speed of the material output end relative to the surface of the ship parts. Ship parts with the same base material, coating type and coating color are classified as similar parts; ship parts with the same batch of treated materials among similar parts are classified as parts from the same batch. The camera is controlled to take pictures of the surface of the ship parts in sequence, and the surface area of ​​the ship parts corresponding to each picture is recorded as an image acquisition area; Each image acquisition area is set sequentially, with overlapping portions between adjacent areas, until all image acquisition areas cover the surface of the ship fitting to be inspected. For each image acquisition area, keeping the position of the camera and the ship fitting unchanged, the polarization direction of the light source is adjusted to be parallel to the transmission direction of the analyzer, and a parallel polarized image is acquired. The polarization direction of the light source is then adjusted to be perpendicular to the transmission direction of the analyzer, and a cross-polarized image is acquired. Keeping the polarization direction of the light source perpendicular to the transmission direction of the analyzer, the light source is moved to the first illumination position and the second illumination position, respectively. The first illumination image and the second illumination image are acquired for the same image acquisition area, and the three-dimensional coordinates of the light source center at the first illumination position and the second illumination position in the camera coordinate system are recorded. The surface temperature and dew point temperature of the ship fitting are acquired simultaneously. For each pixel in each image, subtract the principal point's x-coordinate from the pixel's x-coordinate and then divide by the camera's lateral focal length to obtain the ray's lateral component; subtract the principal point's ordinate from the pixel's ordinate and then divide by the camera's longitudinal focal length to obtain the ray's longitudinal component; set the ray's forward component to one, and determine the ray's direction originating from the camera's optical center using the lateral, longitudinal, and forward components. Extending from the camera optical center along the ray direction to the three-dimensional surface model, the intersection point of the ray and the three-dimensional surface model that is closest to the camera optical center is determined as the surface point corresponding to the pixel, and the three-dimensional coordinates and surface normal of the surface point in the camera coordinate system are read as the three-dimensional surface position and surface normal corresponding to the pixel. For any region in the image, calculate the average of the three coordinate components of the three-dimensional coordinates of all pixels in the region. Select the surface point in the three-dimensional surface model that is closest to the position represented by the average of the three coordinate components and record this surface point as the center surface point of the region. Record the shortest path length between the center surface points of two image regions along the three-dimensional surface model as the surface migration distance. Obtain historical valid test records for similar parts; historical valid test records include part identification, substrate material, coating type, coating color, batch of treatment material, surface treatment equipment identification, coating treatment completion time, test images, equipment operating parameters, and test results.

3. The intelligent integrated method for identifying and scheduling surface treatment defects in ship components according to claim 1, characterized in that: In S2, candidate anomaly regions are extracted as follows: The cross-polarized image is input into the image segmentation model, and the image feature size is reduced sequentially to extract edge features, texture features and region shape features. When restoring the image feature size, shallow features and deep features of the same size are fused to obtain the final-level image features with the same size as the cross-polarized image. The brightness difference between parallel polarized and cross-polarized images at the same pixel position is calculated to form a polarization difference map. The minimum pixel value of the polarization difference map is subtracted from each pixel value, and then divided by the difference between the maximum and minimum pixel values ​​to obtain the polarization response value. When the maximum and minimum pixel values ​​are equal, the polarization response value is set to zero. The polarization response value is multiplied pixel by pixel with the final-level image features, and then added pixel by pixel with the final-level image features. The sum is input into the pixel classification layer to output candidate anomaly regions. Anomalies in the first illumination image are extracted using an image segmentation model. The anomaly region with the largest overlapping area with the projection position of the candidate anomaly region is selected as the first candidate anomaly region. If no overlapping region is obtained, a record to be optically verified is formed. Centered on the surface point of the first candidate anomaly region, the surface range whose shortest path length along the three-dimensional surface model between it and the surface point of the first candidate anomaly region is no greater than the maximum surface migration distance in the historical valid detection records where the detection result is reflective interference is taken as the reflective search area. Anomalies in the second illumination image are extracted using an image segmentation model, and the anomalies located within the reflection search area are read. For each abnormal region read, the incident direction is determined based on the second illumination position and the center surface point of the abnormal region. The incident direction is mirrored with respect to the surface normal of the center surface point to obtain the predicted reflection direction. The specular reflection angle between the predicted reflection direction and the direction from the center surface point of the region to the camera optical center is calculated. Abnormal regions with specular reflection angles greater than the maximum specular reflection angle in the historical valid detection records where the detection result is a reflection interference are deleted.

4. The intelligent integrated method for identifying and scheduling surface treatment defects in ship parts according to claim 3, characterized in that: In S2, based on the polarization brightness attenuation ratio, illumination characteristic distance, surface migration distance, and specular reflection angle, a reflection interference record and an environmental interference judgment record are generated, as follows: Calculate the surface migration distance between each remaining abnormal region and the first candidate abnormal region. The abnormal region with the smallest absolute value of the difference between the surface migration distance and the median surface migration distance in the historical valid detection records where the detection result is reflective interference is selected as the second candidate abnormal region. If the absolute values ​​are the same, select the abnormal region with the smallest specular reflection angle. If there are no remaining abnormal regions, form a record to be optically verified. Images of the first and second candidate anomaly regions are cropped respectively. After adjusting the two images to the same size, they are input into the feature extraction part of the image segmentation model. The image features output before restoring the image feature size are read. The two image features are converted into feature vectors of the same length, and the Euclidean distance between the two feature vectors is calculated to obtain the illumination feature distance. The polarization brightness attenuation ratio is obtained by subtracting the average brightness of the same position in the cross-polarized image from the average brightness of the first candidate anomalous region in the parallel polarized image, and then dividing by the average brightness of the first candidate anomalous region in the parallel polarized image. The minimum polarization brightness attenuation ratio, minimum illumination feature distance, and minimum surface migration distance in the historical valid detection records with the detection result of reflective interference are respectively taken as the lower limit of reflective attenuation ratio, lower limit of reflective feature distance, and lower limit of reflective migration distance, and the maximum value of specular reflection angle is taken as the upper limit of reflective angle. The maximum distance of the lighting feature in the historical valid detection records where the detection result is a physical defect is taken as the upper limit of the physical feature distance; When the polarization brightness attenuation ratio is not less than the lower limit of the reflection attenuation ratio, the illumination feature distance is not less than the lower limit of the reflection feature distance, the surface migration distance is not less than the lower limit of the reflection migration distance, and the specular reflection angle is not greater than the upper limit of the reflection angle, a reflection interference record is formed; when the polarization brightness attenuation ratio is not less than the lower limit of the reflection attenuation ratio, the illumination feature distance is not less than the lower limit of the reflection feature distance, the surface migration distance is not less than the lower limit of the reflection migration distance, the specular reflection angle is not greater than the upper limit of the reflection angle, the illumination feature distance is not greater than the upper limit of the entity feature distance, and the surface migration distance is not greater than the upper limit of the image positioning error, a record pending environmental interference judgment is formed; all other cases form a record pending optical verification.

5. The intelligent integrated method for identifying and scheduling surface treatment defects in ship parts according to claim 1, characterized in that: In S3, for the environmental interference judgment record, a condensation interference record and an adhesion state judgment record are formed based on the surface dew point balance and three unit temperature rise data, as follows: For records to be judged by environmental interference, the boundary of the candidate abnormal area is extended outward along the three-dimensional surface by the effective measurement diameter of the film thickness detection probe. The part that overlaps with other candidate abnormal areas is deleted, and the remaining area is regarded as the neighboring normal area. If there is no remaining area, the surface area that is closest to the candidate abnormal area and has not been identified as abnormal is selected. If there is still no usable area, a record to be optically verified is formed. The surface dew point margin is obtained by subtracting the dew point temperature from the surface temperature of the ship fittings. When the surface dew point margin is between the minimum and maximum values ​​of the surface dew point margin in the historical valid test records where the test results show condensation interference, controlled heating is performed. When it is not between the two limits, a record of the pending attachment state is formed. When performing controlled heating, the median of the heating amplitude and holding time after heating in the historical valid test records with qualified heating test results are used as the heating amplitude and holding time for each stage, respectively. The lower of the upper limit of the allowable test temperature and the highest surface temperature value in the historical valid test records is used as the heating stop temperature. After each stage of heating, the positions of the camera, ship accessories and lighting are kept unchanged, and the heating response image is acquired when the difference between the maximum and minimum values ​​of three consecutive surface temperature readings is not greater than the maximum value of the repeated measurement error of the temperature sensor. The candidate abnormal regions before heating are projected onto the heating response image, and the abnormal region with the largest overlapping area is taken as the candidate abnormal region after heating. The reduction ratio of the abnormal area is obtained by subtracting the area of ​​the candidate abnormal area after the heating from the area of ​​the candidate abnormal area before the heating, and then dividing by the area of ​​the candidate abnormal area before the heating. The shortest surface distance from each point on the boundary of the candidate abnormal area after the heating to the boundary of the candidate abnormal area before the heating is calculated. The distance inside the candidate abnormal area before the heating is recorded as a positive value, and the distance outside is recorded as a negative value. The average value of the corresponding distances at each boundary point is taken to obtain the boundary inward shift. The reduction of the brightness difference is obtained by subtracting the corresponding brightness difference after the heating from the brightness difference between the abnormal area before the heating and the adjacent normal area. Divide the abnormal area reduction ratio, boundary inward shift, and brightness difference reduction by the current temperature rise value to obtain the area reduction ratio per unit temperature rise, boundary inward shift per unit temperature rise, and brightness difference reduction per unit temperature rise. When all three unit temperature rise data are between the minimum and maximum values ​​of the corresponding data in the historical valid test records where the test result is condensation interference, a condensation interference record is formed and a delayed retest task is generated; if the three conditions are not met simultaneously and the surface temperature is lower than the heating stop temperature, the heating continues; if the three conditions are still not met simultaneously after reaching the heating stop temperature, the heating stops and waits for the temperature to recover. If the difference between the current surface temperature and the surface temperature before heating is not greater than the maximum value of the repeated measurement error of the temperature sensor, keep the camera and illumination positions unchanged and re-acquire cross-polarized images; input the re-acquired cross-polarized images into the image segmentation model to extract abnormal regions; if there are extracted abnormal regions within the three-dimensional surface range corresponding to the candidate abnormal regions before heating, a pending attachment state judgment record is formed; if there are no extracted abnormal regions, a delayed re-inspection task is generated; if no abnormality is found during the delayed re-inspection, condensation interference is confirmed, and a pending attachment state judgment record is formed when an abnormality reappears.

6. The intelligent integrated method for identifying and scheduling surface treatment defects in ship parts according to claim 1, characterized in that: In S4, for the record of the state to be attached, loose attachment records and stable entity anomaly records are formed based on the area compactness, area grayscale standard deviation, anomaly center displacement, anomaly area reduction ratio, and area overlap ratio, as follows: For the record of the state to be attached, calculate the ratio of the area of ​​the candidate abnormal region to the area of ​​the minimum circumscribed convex hull of the candidate abnormal region to obtain the region compactness; calculate the standard deviation of the gray value of each pixel in the candidate abnormal region to obtain the region gray standard deviation. When the area compactness is between the minimum and maximum values ​​of the area compactness in the historical valid test records where the test result is loose attachment, and the area grayscale standard deviation is between the minimum and maximum values ​​of the area grayscale standard deviation in the historical valid test records, airflow disturbance is performed; if neither condition is met simultaneously, a stable entity anomaly record is formed. When performing airflow disturbance, the median of the distance from the airflow nozzle to the surface of the ship's fittings, the airflow action time, and the pressure difference between two adjacent airflow stages in the historical valid test records with the test result of passing the purging test are respectively used as the current nozzle distance, the single action time, and the pressure increase per stage. The minimum airflow pressure that can cause displacement of the abnormal area in the historical valid test records with the test result of loose attachments is used as the initial airflow pressure. Orient the airflow nozzle toward the center surface point of the candidate anomaly area, and make the angle between the airflow direction and the surface normal of the center surface point of the area 90 degrees. Apply directional airflow to the candidate anomaly area. After each airflow stops, acquire the airflow response image. The airflow pressure does not exceed the upper limit of the allowable purging pressure. Candidate anomaly regions before airflow action are projected onto the airflow response image, and the anomaly region with the largest overlapping area is taken as the candidate anomaly region after airflow action; if no overlapping area is obtained, a loose attachment record is formed. When there are candidate anomaly regions after the airflow effect, the surface migration distance between the two candidate anomaly regions before and after the airflow effect is taken as the displacement of the anomaly center; the area of ​​the candidate anomaly region before the airflow effect is subtracted from the area of ​​the candidate anomaly region after the airflow effect, and then divided by the area of ​​the candidate anomaly region before the airflow effect to obtain the anomaly area reduction ratio; the overlapping area of ​​the two candidate anomaly regions is divided by the area of ​​the candidate anomaly region before the airflow effect to obtain the region overlap ratio. The minimum value of abnormal center displacement, the minimum value of abnormal area reduction ratio, and the maximum value of region overlap ratio in the historical valid detection records of loose attachments are respectively taken as the lower limit of loose displacement, the lower limit of loose area reduction ratio, and the upper limit of loose overlap ratio; the maximum value of abnormal center displacement, the maximum value of abnormal area reduction ratio, and the minimum value of region overlap ratio in the historical valid detection records of solid defects are respectively taken as the upper limit of solid displacement, the upper limit of solid area reduction ratio, and the lower limit of solid overlap ratio. A record of loose attachments is formed when at least one of the following conditions is met: the abnormal center displacement is not less than the lower limit of loose displacement and the abnormal area reduction ratio is not less than the lower limit of loose area reduction ratio, and the area overlap ratio is not greater than the upper limit of loose overlap ratio. A stable entity anomaly record is formed when no loose attachment record is formed, and the displacement of the abnormal center is not greater than the upper limit of the entity displacement, the reduction ratio of the abnormal area is not greater than the upper limit of the entity area reduction ratio, and the regional overlap ratio is not less than the lower limit of the entity overlap ratio. When the regional overlap ratio is greater than the upper limit of loose overlap ratio but less than the lower limit of solid overlap ratio, the portion obtained by subtracting the overlapping region from the candidate anomaly region before the airflow action is recorded as loose attachment, and the overlapping region is recorded as a stable solid anomaly; in other cases, a pending attachment status verification record is formed.

7. The intelligent integrated method for identifying and scheduling surface treatment defects in ship parts according to claim 1, characterized in that: In S5, for stable entity anomaly records, based on film thickness detection results, the overlap ratio of low-area areas, the overlap ratio of high-area areas, and film thickness differences, incomplete coating coverage records, localized coating accumulation records, and common equipment anomaly records are generated, as follows: For stable entity anomaly records, film thickness detection points are arranged in the stable entity anomaly area and the adjacent normal area. The distance between adjacent detection points along the three-dimensional surface is no greater than half of the effective measurement diameter of the film thickness detection probe. With each detection point as the center, a film thickness detection coverage area is generated with half of the effective measurement diameter as the surface distance radius. The minimum and maximum values ​​of the film thickness test results in the historical valid test records with qualified film thickness are respectively used as the lower limit and upper limit of the qualified film thickness range; the overlapping coverage areas where the film thickness is all below the lower limit of the qualified film thickness range are merged, the overlapping coverage areas where the film thickness is all above the upper limit of the qualified film thickness range are merged, and the overlapping coverage areas where the film thickness is all within the qualified film thickness range are merged. Areas with film thickness below the lower limit of the qualified film thickness range are merged into areas with low film thickness, and areas with film thickness above the upper limit of the qualified film thickness range are merged into areas with high film thickness. The overlap ratio of the low film thickness region is obtained by dividing the overlapping area of ​​the stable entity anomaly region and the low film thickness region by the area of ​​the stable entity anomaly region. The overlap ratio of the high film thickness region is obtained by dividing the overlapping area of ​​the stable entity anomaly region and the region with high film thickness by the area of ​​the stable entity anomaly region. The difference in membrane thickness is obtained by subtracting the median membrane thickness of the adjacent normal region from the median membrane thickness of the abnormal region of the stable entity. The minimum overlap ratio of low-value areas in historical valid inspection records where the inspection result is a solid defect and the solid defect type is insufficient coating coverage is taken as the lower limit of the overlap ratio for insufficient coverage; the minimum overlap ratio of high-value areas in historical valid inspection records where the inspection result is a solid defect and the solid defect type is local coating accumulation is taken as the lower limit of the overlap ratio for local accumulation. When the overlap ratio of the low-coverage area is not less than the lower limit of the overlap ratio for insufficient coverage and the film thickness difference is less than zero, a record of insufficient coating coverage is formed; when the overlap ratio of the high-coverage area is not less than the lower limit of the overlap ratio for local accumulation and the film thickness difference is greater than zero, a record of local coating accumulation is formed; when all film thickness test results within the abnormal area of ​​a stable entity are within the acceptable film thickness range, a record of entity defect pending verification is formed; in other cases, a record of film thickness verification pending verification is formed and film thickness test points are added. Standard samples refer to samples whose substrate material, coating type, and coating color are the same as those of ship fittings, and which have been confirmed to be qualified by appearance inspection and film thickness inspection. The same batch of accessories processed by the same surface treatment equipment after the most recent standard sample film thickness test passed the test are arranged according to the coating treatment completion time to form the same batch test sequence; when adjacent records show the same defect type and the same positive or negative sign of the film thickness difference, the number of consecutive records is accumulated; when there are qualified records, records with different defect types, or records with different positive or negative signs of film thickness difference, the count is reset. The minimum number of consecutive records in the historical valid inspection records where the equipment abnormality was confirmed is taken as the lower limit of the number of consecutive records of equipment abnormalities. When the current number of consecutive records reaches the lower limit of the number of consecutive records of equipment abnormalities, a common equipment abnormality record is formed. If it does not reach the lower limit, a defect record of entity to be associated is formed. When there is no historical valid inspection record where the equipment abnormality is confirmed, a defect record of entity to be associated is formed directly. The latest sample verification time is obtained by subtracting the average processing time of standard samples and the average film thickness detection time in the historical valid inspection records from the start time of the next production process of the current ship parts. If the subsequent inspection records are still insufficient to make the number of consecutive records reach the lower limit of the number of consecutive equipment abnormalities when the latest sample verification time is reached, the standard sample is processed using the current batch of processed materials and the current equipment operating parameters. When standard samples exhibit the same defect type and have the same positive or negative sign for the film thickness difference, a common equipment anomaly record is generated. When no samples exhibit the same defect type, a single-sample entity defect record is generated. When samples exhibit the same defect type but have different positive or negative signs for the film thickness difference, a record awaiting equipment verification is generated.

8. The intelligent integrated method for identifying and scheduling surface treatment defects in ship parts according to claim 1, characterized in that: In S6, based on records of reflective interference, condensation interference, loose attachments, insufficient coating coverage, localized coating buildup, and common equipment anomalies, corresponding tasks are generated and executed, and historical valid detection records are updated, as follows: Based on the reflection interference record, an illumination adjustment task and a supplementary acquisition task are generated; when the anomaly disappears after supplementary acquisition, the reflection interference is confirmed; when the surface migration distance between the two abnormal areas before and after supplementary acquisition is not greater than the upper limit of the image positioning error, it is passed to S3. An optical verification task is generated based on the optical verification record; the task ends when the reflection is confirmed during verification; if the surface migration distance between the two abnormal areas before and after verification is not greater than the upper limit of the image positioning error, it is passed to S3; if it cannot be confirmed, the corresponding ship parts are prohibited from entering the next production process. Based on the condensation interference record, a delayed re-inspection task is generated. When the abnormality disappears after the re-inspection, the condensation interference is confirmed. If the abnormality still exists, the cross-polarization image is re-acquired and transmitted to S4. Surface cleaning tasks and post-cleaning sampling tasks are generated based on records of loose attachments; When the abnormality disappears after cleaning, confirm the loose attachments; if the abnormality still exists, form a stable entity abnormality record and transfer it to S5. Generate an attachment status verification task based on the attachment status verification record; perform a surface cleaning task when loose attachments are confirmed; if the surface migration distance between the two abnormal areas before and after verification is not greater than the upper limit of the image positioning error, transfer it to S5; if it cannot be confirmed, prohibit the corresponding ship parts from entering the next production process. Based on records of insufficient coating coverage, localized coating buildup, and single-piece defects, local supplementary treatment tasks, localized repair tasks, and localized rework tasks are generated, and film thickness retesting is performed after processing. Detection points are added and retested based on the film thickness verification records. Quality verification tasks are generated based on the physical defect records. When the quality verification confirms that there are no physical defects, a release record is generated. When physical defects are confirmed, local supplementary treatment tasks, localized repair tasks, or localized rework tasks are generated based on the confirmed defect type. All parts from the same batch processed by the same equipment after the most recent standard sample film thickness test passed are recorded as related parts; an equipment verification task is generated based on the common equipment anomaly record, and the equipment is suspended from receiving subsequent parts from the same batch. For each piece of equipment's operating parameters, if the operating parameter is less than the minimum value of the corresponding equipment's operating parameter in the historical valid test records where the test result indicates that the equipment is operating qualified, the operating parameter of that equipment will be adjusted to the median value of the corresponding equipment's operating parameters; if the operating parameter is greater than the maximum value of the corresponding equipment's operating parameters, the operating parameter of that equipment will be adjusted to the median value of the corresponding equipment's operating parameters. Standard samples were processed using the adjusted equipment operating parameters; If the appearance inspection results and film thickness inspection results of the standard sample are both qualified, the equipment operation is resumed and the associated accessories are re-inspected; if the standard sample is unqualified and the associated accessories continue to show the same common equipment abnormality, the equipment is kept paused and an equipment maintenance task is generated. When the film thickness is retested and found to be qualified, a release record for the next production process is generated; if it is still not qualified, a quality review task is generated. When the quality review confirms that rework can continue, a corresponding local processing task is regenerated; when it is confirmed that rework cannot continue, a non-conforming product disposal task is generated. Tasks are executed in the following order: lighting adjustment before supplementary data collection, delayed waiting before re-inspection, surface cleaning before post-cleaning data collection, local treatment before film thickness re-measurement, and equipment verification before release of related accessories. Tasks in which all preceding tasks are completed are executed from earliest to latest according to the start time of the next production process. During the current batch of processed materials, the lower limit, upper limit, and median of judgment determined in S2 to S5 based on historical valid test records remain unchanged; After the current batch of materials is processed, the current test record with complete corresponding test data, which has obtained one of the following: final verification result, film thickness retest result, or equipment verification result, will be added to the historical valid test record. When newly added test data is less than the current minimum value or greater than the current maximum value, it will be added to the historical valid test record after manual verification confirms that both the test data and the test results are valid.

Images