Workpiece repairing method and system based on data analysis

Abstract

The invention discloses a workpiece repairing method and system based on data analysis, and the method comprises the steps: S1, obtaining a point cloud data set through the scanning of a visual sensor by a repairing robot, and transmitting the point cloud data set and the deflection angle information of the visual sensor to an intelligent manufacturing cloud platform; s2, a data separation module divides the point cloud data set into a background layer point cloud data set and a foreground layer point cloud data set; s3, a relation construction module obtains an adjacent point data set of each workpiece scanning point; s4, the defect analysis module performs analysis to obtain position information, damage depth and damage area of the damaged area of the defective workpiece; s5, a grade evaluation module performs evaluation to obtain the damage grade of the corresponding defective workpiece; and S6, a decision execution module executes a corresponding workpiece repair strategy according to the damage level of the defective workpiece. According to the invention, the workload of operators can be reduced, the labor cost is reduced, and the detection and repair efficiency of defective workpieces is improved.

Description

Technical Field

[0001] This invention relates to workpiece repair, and in particular to a workpiece repair method and system based on data analysis. Background Technology

[0002] With the development of information and communication technologies, big data is no longer just a concept, but is gradually being integrated into all aspects of people's production and life, and society is showing a trend of interconnection of everything.

[0003] With the advancement of digital technology, computer algorithms have become increasingly complex, stable, and scientific, revolutionizing the ways data is generated, transmitted, and processed, profoundly impacting people's lifestyles. The foundational technology of big data is cloud computing for storing, managing, mining, and analyzing data. Core technologies include data acquisition, machine learning, data preprocessing, and databases.

[0004] Manufacturing is a major pillar of the national economy. Remanufacturing, as an emerging manufacturing method, plays an increasingly important role in the manufacturing sector. Remanufacturing is an industry that uses high technology to repair and renovate used products. It targets damaged or near-obsolescence components, conducting remanufacturing design based on performance failure and lifespan assessment, and employing a series of related advanced manufacturing technologies to ensure that the quality of remanufactured products reaches or exceeds that of new products. Remanufacturing extends the manufacturing industry chain, upgrading the entire life cycle of equipment from an open-loop system of "development-use-obsolescence" to a closed-loop system of "development-use-obsolescence-regeneration."

[0005] With the advancement of remanufacturing technology, manufacturing companies have begun to remanufacture key components of engineering machinery equipment that are nearing the end of their service life, thereby extending the equipment's lifespan, allowing it to continue to realize its value, and reducing resource waste to some extent.

[0006] Currently, traditional design models do not consider the subsequent remanufacturing of equipment parts, making it difficult to determine the timing of remanufacturing of equipment parts. In other words, it is difficult to determine the damage status of each part during operation in real time. The time cost of manually inspecting whether equipment parts are damaged is too high and the efficiency is low. If the damaged parts of equipment parts cannot be detected and repaired in time, it will accelerate the wear and tear of equipment parts and affect the production progress and quality of products. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a workpiece repair method and system based on data analysis, which enables repair robots to perform workpiece repair operations more intelligently, reduces the workload of operators, lowers labor costs, and improves the detection and repair efficiency of damaged workpieces.

[0008] The objective of this invention is achieved through the following technical solution: a workpiece repair method based on data analysis, comprising the following steps: Step S1. The repair robot scans the damaged workpiece in the target area using a vision sensor to obtain a point cloud dataset, and sends the point cloud dataset and the deflection angle information of the vision sensor to the intelligent manufacturing cloud platform. Step S2. In the intelligent manufacturing cloud platform, the data separation module performs depth-level division on all workpiece scanning points to divide the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset. Step S3. In the intelligent manufacturing cloud platform, the relationship building module removes discrete workpiece scanning points whose deviation curvature exceeds the surface curvature threshold as noise points from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. Then, based on the scanning perspective information of the vision sensor, it obtains the adjacent point dataset of each workpiece scanning point in the smooth point cloud dataset in three-dimensional space. Step S4. In the intelligent manufacturing cloud platform, the defect analysis module reconstructs the model based on the adjacent point dataset of each workpiece scanning point, the three-dimensional coordinates and depth values ​​of each workpiece scanning point to obtain the three-dimensional workpiece model of the target detected defective workpiece, and compares the three-dimensional workpiece model with the standard workpiece model to obtain the location information, damage depth and damage area of ​​the damaged area of ​​the defective workpiece. Step S5. In the intelligent manufacturing cloud platform, the level assessment module assesses the degree of damage of the missing workpiece based on the depth and area of ​​the damage to obtain the corresponding damage level of the missing workpiece. The damage level includes minor damage, moderate damage and severe damage. Furthermore, this application uses the product of the damaged area and the damaged depth of the defective workpiece as an evaluation index for the degree of damage, and sets a first threshold and a second threshold (where the first threshold is less than the second threshold) to evaluate the damage level; when the evaluation index is less than or equal to the first threshold, the damage level is considered to be minor damage; when the evaluation index is greater than the first threshold and less than or equal to the second threshold, the damage level is considered to be moderate damage; when the evaluation index is greater than the second threshold, the damage level is considered to be severe damage.

[0009] Step S6. The decision execution module executes the corresponding workpiece repair strategy according to the damage level of the defective workpiece.

[0010] A workpiece repair system based on data analysis includes a management terminal, a repair robot, and an intelligent manufacturing cloud platform; the intelligent manufacturing cloud platform has communication connections with both the management terminal and the repair robot. The intelligent manufacturing cloud platform includes a data separation module, a relationship building module, a defect analysis module, a level assessment module, and a decision execution module; The data separation module is used to perform depth-level division of all workpiece scanning points according to the depth information of each workpiece scanning point in the point cloud dataset and the standard structural information of the defective workpiece detected by the target, so as to divide the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset. The relationship construction module is used to construct a neighboring surface centered on each workpiece scanning point based on the three-dimensional coordinates of each workpiece scanning point in the three-dimensional space in the foreground layer point cloud dataset, and to obtain the deviation curvature of each discrete workpiece scanning point based on the curvature of all neighboring surfaces. Discrete workpiece scanning points with deviation curvature exceeding the curvature threshold are treated as noise points and removed from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. Then, the adjacent point dataset of each workpiece scanning point in the three-dimensional space is obtained based on the scanning perspective information of the visual sensor. The defect analysis module is used to reconstruct the model based on the adjacent point dataset of each workpiece scanning point, the three-dimensional coordinates and depth values ​​of each workpiece scanning point to obtain the three-dimensional workpiece model of the target detected defective workpiece, and compare the three-dimensional workpiece model with the standard workpiece model to obtain the location information, damage depth and damage area of ​​the damaged area of ​​the defective workpiece. The grade assessment module is used to assess the degree of damage to the missing workpiece based on the depth and area of ​​the damage to obtain the corresponding damage grade of the missing workpiece. The damage grades include minor damage, moderate damage, and severe damage. The decision execution module is used to execute corresponding workpiece repair strategies based on the damage level of the defective workpiece.

[0011] The beneficial effects of this invention are as follows: By rapidly scanning and measuring each surface of a workpiece, the three-dimensional workpiece model formed by the scan is compared with a standard workpiece model to identify the location information, depth, and area of ​​the workpiece's defective areas. This allows for timely detection of defects on the workpiece surface. Based on the defective area and depth, the degree of workpiece damage is assessed, and corresponding repair strategies are adopted. This enables the repair robot to perform workpiece repair operations more intelligently, reducing the workload of operators, lowering labor costs, and improving the efficiency of detecting and repairing defective workpieces. Attached Figure Description

[0012] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the system principle of the present invention. Detailed Implementation

[0013] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following description.

[0014] like Figure 1 As shown in the embodiments of this application, the workpiece repair method based on data analysis of the present invention includes the following steps: S1. The repair robot responds to the workpiece inspection request sent by the management terminal, scans the defective workpiece in the target area through its peripheral vision sensor to obtain a point cloud dataset of several geometric structural surfaces of the defective workpiece, and sends the point cloud dataset and the deflection angle information of the vision sensor to the intelligent manufacturing cloud platform.

[0015] Optionally, the workpiece inspection request includes a robot identifier, a target inspection area, and workpiece list information; the workpiece list information includes the workpiece number, workpiece name, workpiece model, standard structure information, and standard workpiece model for each workpiece to be inspected.

[0016] Optionally, the robot identifier is used to uniquely identify the robot; the deflection angle information is used to indicate the changing angle of the vision sensor during the scanning of various geometric surfaces of the damaged workpiece.

[0017] Optionally, the standard structural information is used to characterize the initial structural information of the damaged workpiece, that is, the shape structure in the undamaged state; the standard workpiece model is the three-dimensional structural model of the damaged workpiece in the undamaged state.

[0018] Optionally, the point cloud dataset includes several workpiece scanning points of each geometric surface of the damaged workpiece.

[0019] Optionally, the visual sensor includes a camera and a laser, the laser being used to emit structured light of a specific pattern onto the surface of an object; The specific modes include: point structured light mode, line structured light mode, multi-line structured light mode, and grid structured light mode.

[0020] S2. The data separation module of the intelligent manufacturing cloud platform performs depth-level division on all workpiece scanning points according to the depth information of each workpiece scanning point in the point cloud dataset and the standard structural information of the defective workpiece detected by the target, so as to divide the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset.

[0021] Optionally, the depth information is used to characterize the relative distance between the object surface corresponding to each workpiece scanning point in the actual scene and the laser of the vision sensor.

[0022] Specifically, the data separation module performs depth-level partitioning of all workpiece scanning points based on the depth information of each workpiece scanning point in the point cloud dataset and the standard structural information of the defective workpiece detected by the target, thereby dividing the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset, including: The data separation module analyzes the standard structural information of the defective workpiece to obtain the spatial size range of the defective workpiece, and divides all workpiece scanning points in the point cloud data into depth regions within the corresponding depth range based on the spatial size range and the depth information of each workpiece scanning point in the point cloud data. The data separation module performs depth statistics on the depth values ​​of the workpiece scanning points in each depth region based on the depth information of the workpiece scanning points, and takes the depth region containing the most workpiece scanning points as the main depth region. The depth value is used to characterize the relative distance between the corresponding workpiece scanning point and the vision sensor. The data analysis module obtains the depth values ​​of all workpiece scanning points in the neighboring depth regions whose depth range is smaller than that of the main depth region to obtain the minimum depth value of the neighboring depth region. Then, based on the maximum depth value of the main depth region and the minimum depth value of the neighboring depth region, the point cloud data is divided into a background layer point cloud dataset and a foreground layer point cloud dataset. The neighboring depth region is the previous depth region and the next depth region that are connected to the depth range of the corresponding main depth region.

[0023] Optionally, the spatial size range refers to the spatial volume occupied by the corresponding damaged workpiece in the actual scene.

[0024] S3. The relationship construction module constructs a neighboring surface centered on each workpiece scanning point based on the three-dimensional coordinates of each workpiece scanning point in the three-dimensional space in the foreground layer point cloud dataset. It then obtains the deviation curvature of each discrete workpiece scanning point based on the curvature of all neighboring surfaces. Discrete workpiece scanning points with deviation curvature exceeding the surface curvature threshold are treated as noise points and removed from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. Finally, it obtains the adjacent point dataset of each workpiece scanning point in the three-dimensional space based on the scanning perspective information of the vision sensor.

[0025] Optionally, the surface curvature threshold is a value preset by the system to determine whether the workpiece scanning point falls on an adjacent surface; the scanning angle information is the angle information when the vision sensor scans the corresponding geometric surface in the defective workpiece; the discrete workpiece scanning point is the workpiece scanning point that does not fall on an adjacent surface.

[0026] Specifically, the relation construction module constructs a neighboring surface centered on each workpiece scanning point based on the 3D coordinates of each workpiece scanning point in the foreground layer point cloud dataset. The relationship building module extracts several workpiece scanning points whose spatial distance to the corresponding workpiece scanning point is within a preset distance based on the three-dimensional coordinates of each workpiece scanning point, so as to obtain the neighbor point set of the workpiece scanning point of the target analysis; The relation building module constructs a neighboring surface centered on the workpiece scanning point of the target analysis, based on the three-dimensional coordinates of each workpiece scanning point in the neighboring point set and the normal vector between each workpiece scanning point.

[0027] Optionally, the relation building module, which constructs a neighboring surface centered on each workpiece scanning point based on the 3D coordinates of each workpiece scanning point in the foreground layer point cloud dataset, further includes: The data separation module obtains the translation and angular offset of the visual sensor during the scanning process based on the deflection angle information of the visual sensor, and analyzes the roll angle, pitch angle and yaw angle of the visual sensor under different scanning viewpoints based on the translation and angular offset of the visual sensor during the scanning process. The data analysis module constructs a rotation matrix and translation transformation vector for the point cloud dataset during spatial transformation based on the roll angle, pitch angle, and yaw angle of the view sensor. Then, based on the rotation matrix and the translation transformation vector, the workpiece scanning points of each geometric surface of the defective workpiece are normalized to the world coordinate system to obtain the three-dimensional coordinates of each workpiece scanning point in three-dimensional space.

[0028] Optionally, the neighbor point set consists of several workpiece scanning points whose spatial distance from the corresponding workpiece scanning point is within a preset distance. The preset distance is set by the system based on the size that can reflect the smallest local details of the defective workpiece.

[0029] Specifically, the relationship building module obtains the adjacent point dataset in three-dimensional space for each workpiece scanning point in the smooth point cloud dataset based on the scanning perspective information of the vision sensor. The dataset includes: The relationship building module projects the workpiece scanning points belonging to the same geometric structure surface of the defective workpiece in the smooth point cloud dataset onto the corresponding two-dimensional view plane based on the scanning perspective information of the vision sensor to obtain several two-dimensional mapping points of the workpiece. The relationship building module establishes a corresponding planar topology for each workpiece 2D mapping point based on the distance vector between all workpiece 2D mapping points on the 2D view plane to obtain the connection relationship between each workpiece 2D mapping point and its adjacent 2D mapping points. Then, the planar topology of each workpiece 2D mapping point is mapped from the 2D view plane to the 3D space to obtain the adjacency point dataset of each workpiece scan point based on the connection relationship between each workpiece 2D mapping point and its adjacent 2D mapping points.

[0030] Optionally, the adjacency point dataset includes all workpiece scan points that are directly connected to the corresponding workpiece scan points in each direction.

[0031] S4. The defect analysis module reconstructs the model based on the adjacent point dataset of each workpiece scanning point, the three-dimensional coordinates and depth values ​​of each workpiece scanning point to obtain the three-dimensional workpiece model of the defective workpiece detected by the target, and compares the three-dimensional workpiece model with the standard workpiece model to obtain the location information, depth and area of ​​the damaged area of ​​the defective workpiece.

[0032] Specifically, the defect analysis module compares the 3D workpiece model with the standard workpiece model to obtain the location information, depth, and area of ​​the damaged area of ​​the defective workpiece, including: The defect analysis module extracts the surface structure feature vectors of each geometric surface of the 3D workpiece model and the standard structure feature vectors of each geometric surface of the standard workpiece model. Based on the distance between each surface structure feature vector and the corresponding standard structure feature vector, the module obtains the location information of the damaged area of ​​the 3D workpiece model. Then, based on the location information of the damaged area of ​​the 3D workpiece model, the module obtains the damage depth and damage area of ​​the damaged area.

[0033] S5. The rating assessment module evaluates the degree of damage to the missing workpiece based on the depth and area of ​​the damage to obtain the corresponding damage rating of the missing workpiece. The damage rating includes minor damage, moderate damage, and severe damage.

[0034] S6. The decision execution module executes the corresponding workpiece repair strategy based on the damage level of the defective workpiece.

[0035] Specifically, the decision execution module executes corresponding workpiece repair strategies based on the damage level of the defective workpiece, including: When the decision execution module determines that the damage level of the missing workpiece is minor, it generates a workpiece repair instruction based on the workpiece type, the location information of the damaged area, and the robot identifier, and sends it to the corresponding repair robot. When the decision execution module determines that the damage level of the missing workpiece is moderate, it generates a manual maintenance instruction based on the workpiece type, the location information of the damaged area, and the equipment identifier, and sends it to the corresponding management terminal. When the decision execution module determines that the damage level of the defective workpiece is minor, it generates a workpiece replacement instruction based on the workpiece type and robot identifier, and sends it to the corresponding repair robot.

[0036] This invention rapidly scans and measures each surface of a workpiece, then compares the resulting three-dimensional workpiece model with a standard workpiece model to identify the location, depth, and area of ​​any defects. This allows for timely detection of surface defects. Based on the defect area and depth, the invention assesses the degree of damage and employs appropriate repair strategies. This enables the repair robot to perform workpiece repair more intelligently, reducing the workload of operators, lowering labor costs, and improving the efficiency of detecting and repairing defective workpieces.

[0037] See Figure 2 In one embodiment, the workpiece repair system for performing the method of the present invention includes a management terminal, a repair robot, and a smart manufacturing cloud platform. The smart manufacturing cloud platform has communication connections with both the management terminal and the repair robot. The management terminal is a device used by an operator that has computing, storage, and communication functions, including smartphones, desktop computers, and laptops.

[0038] The intelligent manufacturing cloud platform includes a data separation module, a relationship building module, a defect analysis module, a level assessment module, and a decision execution module.

[0039] The data separation module is used to perform depth-level division of all workpiece scanning points based on the depth information of each workpiece scanning point in the point cloud dataset and the standard structural information of the defective workpiece detected by the target, so as to divide the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset.

[0040] The relation construction module is used to construct a neighboring surface centered on each workpiece scanning point in the three-dimensional space based on the three-dimensional coordinates of each workpiece scanning point in the foreground layer point cloud dataset. It also obtains the deviation curvature of each discrete workpiece scanning point based on the curvature of all neighboring surfaces. Discrete workpiece scanning points with deviation curvature exceeding the surface curvature threshold are treated as noise points and removed from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. Then, the module obtains the adjacent point dataset of each workpiece scanning point in the three-dimensional space based on the scanning perspective information of the vision sensor.

[0041] The defect analysis module is used to reconstruct the model based on the adjacent point dataset of each workpiece scanning point, the three-dimensional coordinates and depth values ​​of each workpiece scanning point to obtain the three-dimensional workpiece model of the defective workpiece detected by the target, and compare the three-dimensional workpiece model with the standard workpiece model to obtain the location information, damage depth and damage area of ​​the damaged area of ​​the defective workpiece.

[0042] The rating module is used to assess the degree of damage to the missing workpiece based on the depth and area of ​​the damage to obtain the corresponding damage rating of the missing workpiece. The damage rating includes minor damage, moderate damage and severe damage.

[0043] The decision execution module is used to execute the corresponding workpiece repair strategy based on the damage level of the defective workpiece.

[0044] The foregoing description illustrates and describes a preferred embodiment of the present invention. However, as previously stated, it should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the inventive concept described herein through the foregoing teachings or techniques or knowledge in related fields. Any modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A workpiece repair method based on data analysis, characterized in that: The following steps are involved: Step S1. The repair robot scans the damaged workpiece in the target area using a vision sensor to obtain a point cloud dataset, and sends the point cloud dataset and the deflection angle information of the vision sensor to the intelligent manufacturing cloud platform. Step S2. In the intelligent manufacturing cloud platform, the data separation module performs depth-level division on all workpiece scanning points to divide the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset. Step S3. In the intelligent manufacturing cloud platform, the relationship building module removes discrete workpiece scanning points whose deviation curvature exceeds the surface curvature threshold as noise points from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. Then, based on the scanning perspective information of the vision sensor, it obtains the adjacent point dataset of each workpiece scanning point in the smooth point cloud dataset in three-dimensional space. Step S4. In the intelligent manufacturing cloud platform, the defect analysis module reconstructs the model based on the adjacent point dataset of each workpiece scanning point, the three-dimensional coordinates and depth values ​​of each workpiece scanning point to obtain the three-dimensional workpiece model of the target detected defective workpiece, and compares the three-dimensional workpiece model with the standard workpiece model to obtain the location information, damage depth and damage area of ​​the damaged area of ​​the defective workpiece. Step S5. In the intelligent manufacturing cloud platform, the level assessment module assesses the degree of damage of the missing workpiece based on the depth and area of ​​the damage to obtain the corresponding damage level of the missing workpiece. The damage level includes minor damage, moderate damage and severe damage. Step S6. The decision execution module executes the corresponding workpiece repair strategy according to the damage level of the defective workpiece.

2. The workpiece repair method based on data analysis according to claim 1, characterized in that: Step S1 includes: The repair robot responds to the workpiece inspection request sent by the management terminal, scans the defective workpiece in the target area through its peripheral vision sensor, collects the point cloud dataset of several geometric surfaces of the defective workpiece, and sends the point cloud dataset and the deflection angle information of the vision sensor to the intelligent manufacturing cloud platform. The workpiece inspection request includes a robot identifier, a target inspection area, and workpiece list information; the workpiece list information includes the workpiece number, workpiece name, workpiece model, standard structure information, and standard workpiece model for each workpiece to be inspected.

3. The workpiece repair method based on data analysis according to claim 2, characterized in that: The robot identifier is used to uniquely identify the robot; the deflection angle information is used to indicate the change in angle of the vision sensor during the scanning of various geometric surfaces of the damaged workpiece. The standard structural information is used to characterize the initial structural information of the damaged workpiece, that is, the shape structure in the undamaged state; the standard workpiece model is the three-dimensional structural model of the damaged workpiece in the undamaged state. The point cloud dataset includes several workpiece scanning points on each geometric surface of the damaged workpiece. The visual sensor includes a camera and a laser, the laser being used to emit structured light of a specific pattern onto the surface of an object; the specific patterns include: point structured light pattern, line structured light pattern, multi-line structured light pattern, and grid structured light pattern.

4. The workpiece repair method based on data analysis according to claim 1, characterized in that: Step S2 includes: The data separation module analyzes the standard structural information of the damaged workpiece to obtain the spatial size range of the damaged workpiece, and divides all workpiece scanning points in the point cloud data into depth regions within the corresponding depth range based on the spatial size range and the depth information of each workpiece scanning point in the point cloud data; the spatial size range is the spatial volume occupied by the corresponding damaged workpiece in the actual scene; the depth information is used to characterize the relative distance between the object surface corresponding to each workpiece scanning point in the actual scene and the laser of the vision sensor. The data separation module performs depth statistics on the depth values ​​of the workpiece scanning points in each depth region based on the depth information of the workpiece scanning points, and takes the depth region containing the most workpiece scanning points as the main depth region. The depth value is used to characterize the relative distance between the corresponding workpiece scanning point and the vision sensor. The data analysis module obtains the depth values ​​of all workpiece scanning points in the neighboring depth regions whose depth range is smaller than that of the main depth region to obtain the minimum depth value of the neighboring depth region. Then, based on the maximum depth value of the main depth region and the minimum depth value of the neighboring depth region, the point cloud data is divided into a background layer point cloud dataset and a foreground layer point cloud dataset. The neighboring depth region is the previous depth region and the next depth region that are connected to the depth range of the corresponding main depth region.

5. The workpiece repair method based on data analysis according to claim 1, characterized in that: Step S3 includes: In the intelligent manufacturing cloud platform, the relationship building module constructs a neighboring surface centered on each workpiece scanning point based on the three-dimensional coordinates of each workpiece scanning point in the three-dimensional space in the foreground layer point cloud dataset. The relationship construction module obtains the deviation curvature of each discrete workpiece scanning point based on the curvature of all neighboring surfaces. Discrete workpiece scanning points with deviation curvature exceeding the surface curvature threshold are treated as noise points and removed from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. The surface curvature threshold is a value preset by the system to determine whether a workpiece scanning point falls on a neighboring surface. The discrete workpiece scanning point is the workpiece scanning point that does not fall on a neighboring surface. The relationship building module obtains the adjacent point dataset in three-dimensional space for each workpiece scanning point in the smooth point cloud dataset based on the scanning perspective information of the vision sensor; the scanning perspective information is the angle information when the vision sensor scans the corresponding geometric surface in the defective workpiece.

6. The workpiece repair method based on data analysis according to claim 5, characterized in that: The relationship construction module constructs a neighboring surface centered on each workpiece scanning point based on the three-dimensional coordinates of each workpiece scanning point in the three-dimensional space of the foreground layer point cloud dataset, including: The relationship building module extracts several workpiece scanning points whose spatial distance from the corresponding workpiece scanning point is within a preset distance based on the three-dimensional coordinates of each workpiece scanning point to obtain the neighboring point set of the target workpiece scanning point; the neighboring point set consists of several workpiece scanning points whose spatial distance from the corresponding workpiece scanning point is within a preset distance, and the preset distance is preset by the system based on the size that can reflect the smallest local details of the defective workpiece; The relation building module constructs a neighboring surface centered on the workpiece scanning point of the target analysis, based on the three-dimensional coordinates of each workpiece scanning point in the neighboring point set and the normal vector between each workpiece scanning point.

7. The workpiece repair method based on data analysis according to claim 5, characterized in that: The relationship building module obtains the adjacent point dataset in three-dimensional space for each workpiece scanning point in the smooth point cloud dataset based on the scanning perspective information of the vision sensor. The dataset includes: The relationship building module projects the workpiece scanning points belonging to the same geometric structure surface of the defective workpiece in the smooth point cloud dataset onto the corresponding two-dimensional view plane based on the scanning perspective information of the vision sensor to obtain several two-dimensional mapping points of the workpiece. The relationship construction module establishes a corresponding planar topology for each workpiece 2D mapping point based on the distance vector between all workpiece 2D mapping points on the 2D view plane to obtain the connection relationship between each workpiece 2D mapping point and its adjacent 2D mapping points. Then, the planar topology of each workpiece 2D mapping point is mapped from the 2D view plane to the 3D space to obtain the adjacency point dataset of each workpiece scanning point based on the connection relationship between each workpiece 2D mapping point and its adjacent 2D mapping points. The adjacency point dataset contains all workpiece scanning points that are directly connected to the corresponding workpiece scanning point in each direction.

8. The workpiece repair method based on data analysis according to claim 1, characterized in that: The step of comparing the three-dimensional workpiece model with the standard workpiece model to obtain the location information, depth of damage, and area of ​​the damaged area of ​​the defective workpiece includes: The defect analysis module extracts the surface structure feature vectors of each geometric surface of the 3D workpiece model and the standard structure feature vectors of each geometric surface of the standard workpiece model. Based on the distance between each surface structure feature vector and the corresponding standard structure feature vector, the module obtains the location information of the damaged area of ​​the 3D workpiece model. Then, based on the location information of the damaged area of ​​the 3D workpiece model, the module obtains the damage depth and damage area of ​​the damaged area.

9. The workpiece repair method based on data analysis according to claim 1, characterized in that: Step S6 includes: When the decision execution module determines that the damage level of the missing workpiece is minor, it generates a workpiece repair instruction based on the workpiece type, the location information of the damaged area, and the robot identifier, and sends it to the corresponding repair robot. When the decision execution module determines that the damage level of the missing workpiece is moderate, it generates a manual maintenance instruction based on the workpiece type, the location information of the damaged area, and the equipment identifier, and sends it to the corresponding management terminal. When the decision execution module determines that the damage level of the defective workpiece is minor, it generates a workpiece replacement instruction based on the workpiece type and robot identifier, and sends it to the corresponding repair robot.

10. A workpiece repair system based on data analysis, wherein the workpiece is repaired using the method described in any one of claims 1 to 9, characterized in that: It includes a management terminal, a repair robot, and a smart manufacturing cloud platform; the smart manufacturing cloud platform has communication connections with both the management terminal and the repair robot. The intelligent manufacturing cloud platform includes a data separation module, a relationship building module, a defect analysis module, a level assessment module, and a decision execution module; The data separation module is used to perform depth-level division of all workpiece scanning points according to the depth information of each workpiece scanning point in the point cloud dataset and the standard structural information of the defective workpiece detected by the target, so as to divide the point cloud dataset into a background layer point cloud dataset and a foreground layer point cloud dataset. The relationship construction module is used to construct a neighboring surface centered on each workpiece scanning point based on the three-dimensional coordinates of each workpiece scanning point in the three-dimensional space in the foreground layer point cloud dataset, and to obtain the deviation curvature of each discrete workpiece scanning point based on the curvature of all neighboring surfaces. Discrete workpiece scanning points with deviation curvature exceeding the curvature threshold are treated as noise points and removed from the foreground layer point cloud dataset to obtain a smooth point cloud dataset. Then, the adjacent point dataset of each workpiece scanning point in the three-dimensional space is obtained based on the scanning perspective information of the visual sensor. The defect analysis module is used to reconstruct the model based on the adjacent point dataset of each workpiece scanning point, the three-dimensional coordinates and depth values ​​of each workpiece scanning point to obtain the three-dimensional workpiece model of the target detected defective workpiece, and compare the three-dimensional workpiece model with the standard workpiece model to obtain the location information, damage depth and damage area of ​​the damaged area of ​​the defective workpiece. The grade assessment module is used to assess the degree of damage to the missing workpiece based on the depth and area of ​​the damage to obtain the corresponding damage grade of the missing workpiece. The damage grades include minor damage, moderate damage, and severe damage. The decision execution module is used to execute corresponding workpiece repair strategies based on the damage level of the defective workpiece.

Images