A Machine Vision-Based Intelligent Detection Method and System for Surface Defects in Titanium Alloy Forgings

Abstract

This invention discloses a machine vision-based intelligent detection method and system for surface defects in titanium alloy forgings, relating to the field of forging inspection. It includes a preprocessing module for picking up and cleaning the forging; a camera module for acquiring and storing surface image data of the forging; and a coordination module for picking up the forging again to achieve spatial posture coordination. This invention first cleans the forging to eliminate surface interference, then captures surface details through multi-directional image acquisition, and combines posture adjustment to place the forging in a specific state. It utilizes the flow image of the cleaning fluid to analyze surface characteristics, and combines the image data with the flow image to calculate the error offset index, achieving accurate detection of surface defects in titanium alloy forgings. It improves defect recognition accuracy through two-dimensional information fusion and is suitable for forging inspection with consistent posture in continuous production.

Description

Technical Field

[0001] This invention relates to the field of forging inspection technology, specifically to a machine vision-based intelligent detection method and system for surface defects in titanium alloy forgings. Background Technology

[0002] Titanium alloy forgings are components made of titanium alloy through forging. They are characterized by high strength, low density, and excellent corrosion resistance, and can withstand high temperatures and complex stresses. They are widely used in aerospace, shipbuilding, medical, and other fields, and are key structural components of high-end equipment, combining excellent mechanical properties with lightweight advantages.

[0003] Patent application No. 202411713518.X discloses a method for detecting surface defects in rolled forgings, comprising: presetting an observation window for each pixel in an image of the forging surface; obtaining the noise performance of a pixel by measuring the uniformity of grayscale distribution within the observation window; acquiring the grayscale level and the corresponding number of pixels within the observation window; using the reciprocal of the variance of the number of pixels corresponding to the grayscale level as the pixel's noise index; using the negative of the ratio of the pixel's noise index to the difference between the extreme values ​​of grayscale levels in the observation window as the exponent of an exponential function with base e to obtain the pixel's noise performance; dividing the pixels in the observation window into a high-grayscale set and a low-grayscale set using the Otsu thresholding method; using the ratio of the number of pixels in the set to the shortest path length as the set's stagnation index; and using the negative of the minimum stagnation index of the high-grayscale set and the low-grayscale set as the exponent of an exponential function with base e to obtain the observed noise performance. The concentration of the window is considered; the noise level of a pixel is obtained by multiplying the noise performance of the pixel by the concentration of the observation window of that pixel; the size of the pixel window is preset, and the weight of the pixel is obtained by the proportion of the noise level of the pixel in the overall noise level of the pixels in the window; the mean of the pixel is obtained by multiplying the gray value of the pixel in the window with the weight; in the guided filtering algorithm, the mean of the pixel is used for filtering and denoising to obtain a denoised forging surface image; the denoised forging surface image is processed by threshold segmentation to obtain the defect detection result of the forging surface image. This application aims to solve the problem that "existing technologies can improve the efficiency of forging defect detection by training convolutional neural network models, but the acquisition of forging images is easily affected by photon noise, resulting in noise points in the acquired forging images. If defect identification is directly based on the undenoised forging images, the accuracy of the defect identification results is low".

[0004] However, when inspecting forgings, the existing technology often uses images of the forgings or measured parameters as the data for inspection. If we want to improve the inspection accuracy, we can only improve the data acquisition accuracy and algorithm optimization. We have not introduced new parameters for the inspection and acceptance of forging surface defects.

[0005] To address this, a machine vision-based intelligent detection method and system for surface defects in titanium alloy forgings are proposed. Summary of the Invention

[0006] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a machine vision-based intelligent detection method and system for surface defects of titanium alloy forgings, which can effectively solve the problems of the existing technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions;

[0008] This invention discloses an intelligent detection system for surface defects in titanium alloy forgings based on machine vision, comprising:

[0009] The system comprises the following modules: a preprocessing module for picking up and cleaning the forging; a camera module for acquiring and storing surface image data of the forging; a coordination module for picking up the forging again to achieve spatial attitude coordination; a testing module for guiding the cleaning fluid through the top surface of the forging to the cleaning component and acquiring images of the cleaning fluid flow as it passes over the forging surface; an inspection module for receiving the surface image data of the forging stored in the camera module and the images of the cleaning fluid flow acquired by the testing module, and combining the surface image data of the forging with the images of the cleaning fluid flow to inspect the forging error offset index; and a judgment module for setting a pass / fail threshold, receiving the forging error offset index from the inspection module, and determining whether the forging is qualified based on a comparison between the forging error offset index and the pass / fail threshold.

[0010] Furthermore, the forging is produced by forging production equipment and transmitted and output by the output conveyor belt of the forging production equipment. The preprocessing module picks up the forging, cleans the forging, puts the forging back into the output conveyor belt of the forging production equipment, and triggers the camera module to run.

[0011] When the forging production equipment continuously outputs forgings from its output end, the output posture and position of each forging relative to the surface of the output end conveyor belt are consistent.

[0012] The pretreatment module is integrated with a robotic arm module and a cleaning module. When the pretreatment module is running, the robotic arm module picks up the forging when the forging is transported to the designated position by the conveyor belt at the output end of the forging production equipment, and transfers the forging to the cleaning module. In the cleaning module, the surface dust is cleaned and the oil film is removed. Then, the forging is placed back on the surface of the conveyor belt along the original transfer path.

[0013] The cleaning module consists of a container, cleaning fluid, and an ultrasonic vibration component. The cleaning fluid is stored in the container, and the ultrasonic component is integrated on the surface of the container. When the ultrasonic component is running, it conducts through the container to cause the cleaning fluid to vibrate. When the forging is transferred to the cleaning fluid by the robotic arm module, it is cleaned to remove dust and oil film from the surface of the forging. The output conveyor belt, robotic arm module, and cleaning module of the forging production equipment operate in a dust-free environment.

[0014] Furthermore, during the forging output process, the camera module acquires surface image data of the visible surface of the forging in the positive direction, which includes: top, left, right, front, and back.

[0015] When the camera module stores the image data of the forging surface, it simultaneously marks the acquisition direction of each image data of the forging surface.

[0016] Before storing the forging surface image data, the camera module simultaneously performs contour recognition on the forging surface image data. Based on the contour recognition results, the forging surface image data is segmented, and the forging area image in the forging surface image data is used as the storage target to perform marking and storage operations.

[0017] Furthermore, when the coordination module picks up and coordinates the spatial posture of the forging, it uses the robotic arm module in the preprocessing module for picking and coordination. The coordination operation follows the following:

[0018] Identify the endpoint connection lines of each image data acquisition surface on the forging, obtain the longest endpoint connection line from each image data acquisition surface, determine whether each endpoint connection line is completely on the forging surface pointed to by its corresponding image data acquisition surface, retain the endpoint connection lines with the determination result of yes, and select the longest endpoint connection line as the indicator line. Use the surface of the forging surface where the indicator line is located as the most upward face to perform the first spatial attitude coordination of the forging.

[0019] The spatial attitude of the forging is coordinated again so that the indicator line on the surface of the forging coincides with the center line of the conveyor belt at the output end of the forging production equipment from a top view, thus completing the secondary spatial attitude coordination of the forging.

[0020] The forging is tilted to a preset angle to complete the third spatial posture coordination of the forging;

[0021] Among them, after the forging completes the third spatial attitude coordination, the forging still meets the constraint conditions after the second spatial attitude coordination, and the angle formed by the top surface of the forging and the surface of the conveyor belt is consistent with the preset tilt angle.

[0022] Furthermore, the test module is integrated with a device for replenishing the cleaning fluid in the container of the cleaning module and a high-definition industrial camera. Each time the test module is run, the cleaning fluid replenishment device outputs replenishment fluid according to a preset amount. When the replenishment fluid flows over the surface of the forging, it first contacts the highest end of the indicator line on the top surface of the forging and flows under gravity. During the flow, the high-definition industrial camera captures images of the cleaning fluid flow.

[0023] The cleaning fluid flows over the top surface of the forging and falls into a container, where it merges with the cleaning fluid placed in the container for subsequent cleaning of the forging.

[0024] During the operation of the coordination module, the cleaning solution in the container is output through the preset electronically controlled valve of the container, and the amount of cleaning solution is equal to the preset amount of replenishment solution.

[0025] Furthermore, the first set of forging surface image data stored by the camera module and the cleaning fluid flow image collected by the test module are pointed at the forging. The system user performs manual inspection. When the inspection result is qualified, the forging surface image data and cleaning fluid flow image corresponding to the forging are used as reference images and images. The forging surface image data and cleaning fluid flow images subsequently acquired by the system are compared with the reference images and images through the inspection module to obtain the forging error offset index.

[0026] During the manual inspection of forgings by system users, if a forging is found to be unqualified, the system refreshes and re-collects surface image data and cleaning fluid flow images of the forgings transmitted by the conveyor belt at the output end of the forging production equipment. The system users then perform manual inspection on the forgings again until the manual inspection result is qualified. The surface image data and cleaning fluid flow images of the forgings corresponding to the inspection result are used as reference images, and the system enters the automated operation state to continuously determine whether the forgings are qualified.

[0027] Furthermore, the system is in an automated operation state, and when the judgment module determines that two consecutive forgings are unqualified, the system ends the automated operation state, and the user on the system end can manually inspect the forging production equipment by stopping the operation.

[0028] The robotic arm module picks up the forgings that are deemed unqualified on the conveyor belt at the output end of the forging production equipment in real time, so as to complete the sorting with qualified forgings.

[0029] Furthermore, the comparison and verification logic in the verification module is expressed as follows:

[0030] ;

[0031] In the formula: Forging error offset index; The total number of viewing angles for acquiring image data of forgings; Let be the similarity between the image of the forging surface pointed to by the i-th acquisition viewpoint and the reference image; Configure the weights for the i-th acquisition viewpoint; The degree of flow trajectory deviation between the reference image and the current inspection forging corresponding to the cleaning fluid flow image;

[0032] Among them, the configuration weights are all positive numbers and follow the rules. The configuration weight values ​​follow the rule that the larger the area of ​​the forging surface that the acquisition viewpoint points to, the higher the value of the configuration weight.

[0033] Furthermore, the preprocessing module interacts with the camera module and the coordination module via a wireless network, the coordination module interacts with the test module via a wireless network, the test module interacts with the inspection module via a wireless network, and the inspection module interacts with the judgment module via a wireless network.

[0034] On the other hand, intelligent detection methods for surface defects in titanium alloy forgings based on machine vision include:

[0035] Pick up the forgings, preprocess them, and acquire surface image data after preprocessing. Coordinate the spatial orientation of the forgings so that the cleaning fluid flows over the top surface of the forgings. Acquire images of the cleaning fluid flow during the process. Compare the surface image data of the forgings with the cleaning fluid flow images with reference surface image data of the forgings and cleaning fluid flow images to check the forging error deviation index. Set a pass / fail threshold and compare it with the forging error deviation index to determine whether the forgings are qualified. Based on the qualification results, sort the forgings to distinguish between qualified and unqualified forgings.

[0036] Compared with the known prior art, the technical solution provided by this invention has the following beneficial effects:

[0037] This invention provides a machine vision-based intelligent detection method and system for surface defects in titanium alloy forgings. During execution, the method and system first clean the forgings to eliminate surface interference, then capture surface details through multi-directional image acquisition, and adjust the forging's posture to a specific state. Surface characteristics are analyzed using the flow image of the cleaning fluid, and the image data and flow image are combined to calculate the error offset index, achieving accurate detection of surface defects in titanium alloy forgings. It improves defect recognition accuracy through dual-dimensional information fusion, adapts to the inspection of forgings with consistent postures in continuous production, automates the entire process from preprocessing to judgment, reduces manual intervention, and can automatically calibrate using reference standards. When continuous non-conformities occur, timely shutdown and maintenance are performed, ensuring both inspection efficiency and improved reliability of quality control. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0039] Figure 1 This is a schematic diagram of a machine vision-based intelligent detection system for surface defects in titanium alloy forgings.

[0040] Figure 2 This is a flowchart illustrating a machine vision-based intelligent detection method for surface defects in titanium alloy forgings. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0042] The present invention will be further described below with reference to embodiments.

[0043] Example 1:

[0044] This embodiment describes a machine vision-based intelligent detection system for surface defects in titanium alloy forgings, such as... Figure 1 As shown, it includes:

[0045] The pre-processing module is used to pick up forgings and clean them.

[0046] The forgings are produced by the forging production equipment and transported out by the output conveyor belt of the forging production equipment. The pre-processing module picks up the forgings, cleans them, puts them back into the output conveyor belt of the forging production equipment, and triggers the camera module to run.

[0047] When forgings are continuously output from the output end of the forging production equipment, the output posture of each forging and its position relative to the surface of the output conveyor belt are consistent.

[0048] The pretreatment module is integrated with the robotic arm module and the cleaning module. When the pretreatment module is running, the robotic arm module picks up the forging when the forging is transported to the designated position by the conveyor belt at the output end of the forging production equipment, and transfers the forging to the cleaning module. In the cleaning module, the surface dust is cleaned and the oil film is removed. Then, the forging is placed back on the surface of the conveyor belt along the original transfer path.

[0049] The cleaning module consists of a container, cleaning fluid, and an ultrasonic vibration component. The cleaning fluid is stored in the container, and the ultrasonic component is integrated on the surface of the container. When the ultrasonic component is running, it conducts through the container to cause the cleaning fluid to vibrate. When the forging is transferred to the cleaning fluid by the robotic arm module, it is cleaned to remove dust and oil film from the surface of the forging. The output conveyor belt, robotic arm module, and cleaning module of the forging production equipment operate in a dust-free environment.

[0050] The camera module is used to acquire and store image data of the forging surface.

[0051] During the forging output process, the camera module acquires surface image data of the visible surface of the forging in the positive direction. The visible surface of the forging in the positive direction includes: top, left, right, front, and back.

[0052] When the camera module stores the image data of the forging surface, it simultaneously marks the acquisition direction of each image data of the forging surface;

[0053] Before storing the forging surface image data, the camera module simultaneously performs contour recognition on the forging surface image data, and segments the forging surface image data based on the contour recognition results. The forging area image in the forging surface image data is used as the storage target, and the marking and storage operations are performed.

[0054] The coordination module is used to pick up the forging again and put the forging in a suspended state to complete the spatial attitude coordination.

[0055] When the coordination module picks up and coordinates the spatial posture of the forging, it uses the robotic arm module in the preprocessing module for picking and coordination. The coordination operation follows the following:

[0056] Identify the endpoint connection lines of each image data acquisition surface on the forging, obtain the longest endpoint connection line from each image data acquisition surface, determine whether each endpoint connection line is completely on the forging surface pointed to by its corresponding image data acquisition surface, retain the endpoint connection lines with the determination result of yes, and select the longest endpoint connection line as the indicator line. Use the surface of the forging surface where the indicator line is located as the most upward face to perform the first spatial attitude coordination of the forging.

[0057] The spatial attitude of the forging is coordinated again so that the indicator line on the surface of the forging coincides with the center line of the conveyor belt at the output end of the forging production equipment from a top view, thus completing the secondary spatial attitude coordination of the forging.

[0058] The forging is tilted to a preset angle to complete the third spatial posture coordination of the forging;

[0059] Among them, after the forging completes the third spatial attitude coordination, the forging still meets the constraint conditions after the second spatial attitude coordination, and the angle formed by the top surface of the forging and the surface of the conveyor belt is consistent with the preset tilt angle.

[0060] The test module is used to guide the cleaning fluid through the top surface of the forging to the cleaning part, and to capture images of the cleaning fluid flow as it flows over the surface of the forging.

[0061] The test module is integrated with a device for replenishing the cleaning fluid in the container of the cleaning module and a high-definition industrial camera. Each time the test module runs, the cleaning fluid replenishment device outputs replenishment fluid according to the preset amount. When the replenishment fluid flows over the surface of the forging, it first contacts the highest end of the indicator line on the top surface of the forging and flows under gravity. During the flow, the high-definition industrial camera captures images of the cleaning fluid flow.

[0062] The cleaning fluid flows over the top surface of the forging and falls into a container, where it merges with the cleaning fluid placed in the container for subsequent cleaning of the forging.

[0063] During the operation of the coordination module, the cleaning solution in the container is output through the pre-set electronically controlled valve of the container, and the amount of cleaning solution is equal to the pre-set amount of replenishing solution.

[0064] The inspection module is used to receive the forging surface image data stored in the camera module and the cleaning fluid flow image collected by the test module, and to inspect the forging error deviation index by combining the forging surface image data and the cleaning fluid flow image.

[0065] The first set of forging surface image data stored by the camera module and the cleaning fluid flow image acquired by the test module are pointed at the forging. The system user performs manual inspection. When the inspection result is qualified, the forging surface image data and cleaning fluid flow image corresponding to the forging are used as reference images and images. The forging surface image data and cleaning fluid flow image acquired by the system subsequently are compared with the reference images and images through the inspection module to obtain the forging error offset index.

[0066] During the manual inspection of forgings by system users, if a forging is found to be unqualified, the system refreshes and re-collects surface image data and cleaning fluid flow images of the forgings transmitted by the conveyor belt at the output end of the forging production equipment. The system users then perform manual inspection on the forgings again until the manual inspection result is qualified. The surface image data and cleaning fluid flow images of the forgings corresponding to the inspection result are used as reference images, and the system enters the automated operation state to continuously determine whether the forgings are qualified.

[0067] When the system is in an automated operation state, and the judgment module determines that two consecutive forgings are unqualified, the system ends the automated operation state and the user on the system side can manually inspect the forging production equipment by stopping the operation.

[0068] The robotic arm module picks up the forgings that are deemed unqualified on the conveyor belt at the output end of the forging production equipment in real time, so as to complete the sorting with qualified forgings.

[0069] The comparison and verification logic in the verification module is expressed as follows:

[0070] ;

[0071] In the formula: Forging error offset index; This represents the total number of viewing angles for acquiring image data of forgings; Let be the similarity between the image of the forging surface pointed to by the i-th acquisition viewpoint and the reference image; Configure the weights for the i-th acquisition viewpoint; The degree of flow trajectory deviation between the reference image and the current inspection forging corresponding to the cleaning fluid flow image;

[0072] Among them, the configuration weights are all positive numbers and follow the rules. The configuration weight values ​​follow the rule that the larger the area of ​​the forging surface pointed to by the acquisition viewpoint, the higher the value of the configuration weight.

[0073] It should be noted that for the similarity calculation between the forging surface image data and the reference image, any one or more similarity calculation methods in the existing technology can be used for comprehensive similarity evaluation. During the evaluation process, image texture features and structural similarity should be considered, and the similarity calculation results of each group of images should be between 0 and 1.

[0074] The reference image and the corresponding cleaning fluid flow image of the current forging under inspection are preferably captured using an optical flow algorithm to determine the flow trajectory. Based on the captured flow trajectory estimation... :

[0075]

[0076] In the formula: The total number of reference points on the flow trajectory of the cleaning fluid in the reference image and the current inspection forging; The coordinates of the j-th reference point on the corresponding flow trajectory of the reference image; The coordinates of the j-th reference point on the flow trajectory corresponding to the current inspection forging's corresponding cleaning fluid flow image; Let be the angle between the direction vectors of the j-th segment on the two flow trajectories; The direction vector of the j-th segment on the flow trajectory corresponding to the reference image. The direction vector of the j-th segment on the flow trajectory of the cleaning fluid corresponding to the current inspection forging. The modulus length; The direction vector of the j-th segment on the flow trajectory corresponding to the reference image. With the direction vector of the (j+1)th segment The included angle between them, and the direction vector of the j-th segment on the flow trajectory of the cleaning fluid corresponding to the current inspection forging. With the direction vector of the (j+1)th segment The angle between them;

[0077] also, In calculation, it can be: and Configure weights separately, with both weights being positive numbers and summed to 2. Configure the weights in the specified locations. The weight is always greater than the configuration and The weights, and configured in The weight values ​​follow the rule that the larger the top surface area of ​​the forging after attitude coordination, the larger its configuration weight value.

[0078] The judgment module is used to set the pass / fail judgment threshold, receive the forging error offset index from the inspection module, and determine whether the forging is pass / fail based on the comparison between the forging error offset index and the pass / fail judgment threshold.

[0079] The preprocessing module interacts with the camera module and the coordination module via a wireless network. The coordination module interacts with the test module via a wireless network. The test module interacts with the inspection module via a wireless network. The inspection module interacts with the judgment module via a wireless network.

[0080] In this embodiment, the preprocessing module picks up the forging and cleans it. The camera module then collects and stores the surface image data of the forging. The coordination module picks up the forging again, suspending it in mid-air to coordinate its spatial posture. The test module further guides the cleaning fluid through the top surface of the forging to the cleaned component, and captures the flow image of the cleaning fluid as it flows across the surface of the forging. The inspection module receives the surface image data of the forging stored in the camera module and the flow image of the cleaning fluid collected by the test module. The forging error offset index is checked by combining the surface image data of the forging and the flow image of the cleaning fluid. Finally, the judgment module sets the pass / fail threshold, receives the forging error offset index from the inspection module, and determines whether the forging is qualified by comparing the forging error offset index with the pass / fail threshold.

[0081] In the above embodiments, the system, through cleaning treatment, multi-view image acquisition and cleaning fluid flow analysis, combined with precise posture adjustment to obtain reference images, can efficiently detect surface defects of titanium alloy forgings, realize batch automated inspection, improve inspection accuracy and efficiency, reduce manual intervention, ensure output quality stability by sorting out defective products in real time, and ensure inspection reliability by operating in a dust-free environment.

[0082] Example 2:

[0083] At the implementation level, based on Example 1, this example refers to... Figure 2 The intelligent detection system for surface defects of titanium alloy forgings based on machine vision in Example 1 will be further described in detail below:

[0084] A machine vision-based intelligent detection method for surface defects in titanium alloy forgings includes:

[0085] Pick up the forging, preprocess the forging, and after the preprocessing is completed, acquire the surface image data of the forging;

[0086] Coordinate the spatial orientation of the forging so that the cleaning fluid flows over the top surface of the forging. During the process of the cleaning fluid flowing over the top surface of the forging, capture images of the cleaning fluid flow.

[0087] By comparing the forging surface image data and cleaning fluid flow image with the reference forging surface image data and cleaning fluid flow image, the forging error offset index is verified.

[0088] Set a pass / fail threshold and compare it with the forging error deviation index to determine whether the forging is qualified.

[0089] Based on the results of the forging's qualification assessment, the forgings are sorted to distinguish between qualified and unqualified forgings.

[0090] In summary, the methods and systems described in the above embodiments, during execution, first clean the forgings to eliminate surface interference, then capture surface details through multi-directional image acquisition, and combine posture adjustment to place the forgings in a specific state. They then utilize the flow image analysis of the cleaning fluid to analyze surface characteristics, and combine the image data with the flow image to calculate the error offset index, thereby achieving accurate detection of surface defects in titanium alloy forgings. This improves defect recognition accuracy through dual-dimensional information fusion, adapts to the inspection of forgings with consistent postures in continuous production, automates the entire process from preprocessing to judgment, reduces manual intervention, and can be automatically calibrated using reference standards. When continuous non-conformities occur, timely shutdown and maintenance are performed, ensuring both inspection efficiency and improving the reliability of quality control.

[0091] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A machine vision-based intelligent detection system for surface defects in titanium alloy forgings, characterized in that, include: The pre-processing module is used to pick up forgings and clean them. The camera module is used to acquire and store image data of the forging surface. The coordination module is used to pick up the forging again and put the forging in a suspended state to complete the spatial attitude coordination. The test module is used to guide the cleaning fluid through the top surface of the forging to the cleaning part, and to capture images of the cleaning fluid flow as it flows over the surface of the forging. The inspection module is used to receive the forging surface image data stored in the camera module and the cleaning fluid flow image collected by the test module, and to inspect the forging error deviation index by combining the forging surface image data and the cleaning fluid flow image. The comparison and verification logic in the verification module is expressed as follows: In the formula: Forging error offset index; This represents the total number of viewing angles for acquiring image data of forgings; Let be the similarity between the image of the forging surface pointed to by the i-th acquisition viewpoint and the reference image; Configure the weights for the i-th acquisition viewpoint; The degree of deviation of the flow trajectory between the reference image and the current inspection forging corresponding to the cleaning fluid flow image; Among them, the configuration weights are all positive numbers and follow the rules. The configuration weight values ​​follow the rule that the larger the area of ​​the forging surface pointed to by the acquisition viewpoint, the higher the value of the configuration weight. The judgment module is used to set the pass / fail judgment threshold, receive the forging error offset index from the inspection module, and determine whether the forging is qualified based on the comparison between the forging error offset index and the pass / fail judgment threshold.

2. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 1, characterized in that, The forging is produced by the forging production equipment and transmitted and output by the output conveyor belt of the forging production equipment. The preprocessing module picks up the forging, cleans the forging, puts the forging back into the output conveyor belt of the forging production equipment, and triggers the camera module to run. When the forging production equipment continuously outputs forgings from its output end, the output posture and position of each forging relative to the surface of the output end conveyor belt are consistent. The pretreatment module is integrated with a robotic arm module and a cleaning module. When the pretreatment module is running, the robotic arm module picks up the forging when the forging is transported to the designated position by the conveyor belt at the output end of the forging production equipment, and transfers the forging to the cleaning module. In the cleaning module, the surface dust is cleaned and the oil film is removed. Then, the forging is placed back on the surface of the conveyor belt along the transfer path. The cleaning module consists of a container, cleaning fluid, and an ultrasonic vibration component. The cleaning fluid is stored in the container, and the ultrasonic vibration component is integrated on the surface of the container. When the ultrasonic vibration component is running, it conducts through the container to cause the cleaning fluid to vibrate. When the forging is transferred to the cleaning fluid by the robotic arm module, it is cleaned to remove dust and oil film from the surface of the forging. The output conveyor belt, robotic arm module, and cleaning module of the forging production equipment operate in a dust-free environment.

3. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 1, characterized in that, During the forging output process, the camera module acquires surface image data of the visible surface of the forging in the positive direction, which includes: top, left, right, front, and back. When the camera module stores the image data of the forging surface, it simultaneously marks the acquisition direction of each image data of the forging surface. Before storing the forging surface image data, the camera module simultaneously performs contour recognition on the forging surface image data. Based on the contour recognition results, the forging surface image data is segmented, and the forging area image in the forging surface image data is used as the storage target to perform marking and storage operations.

4. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 1, characterized in that, When the coordination module picks up and coordinates the spatial posture of the forging, it uses the robotic arm module in the preprocessing module for picking and coordination. The coordination operation follows the following: Identify the endpoint connection lines of each image data acquisition surface on the forging, obtain the longest endpoint connection line from each image data acquisition surface, determine whether each endpoint connection line is completely on the forging surface pointed to by its corresponding image data acquisition surface, retain the endpoint connection lines with the determination result of yes, and select the longest endpoint connection line as the indicator line, take the forging surface where the indicator line is located as the upward surface, and perform the first spatial attitude coordination of the forging. The spatial attitude of the forging is coordinated again so that the indicator line on the surface of the forging coincides with the center line of the conveyor belt at the output end of the forging production equipment from a top view, thus completing the secondary spatial attitude coordination of the forging. The forging is tilted to a preset angle to complete the third spatial posture coordination of the forging; Among them, after the forging completes the third spatial attitude coordination, the forging still meets the constraint conditions after the second spatial attitude coordination, and the angle formed by the top surface of the forging and the surface of the conveyor belt is consistent with the preset tilt angle.

5. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 1, characterized in that, The test module is integrated with a device for holding cleaning fluid in a container of the replenishing cleaning module and a high-definition industrial camera. Each time the test module is run, the cleaning fluid replenishing device outputs replenishing fluid according to a preset amount. When the replenishing fluid flows over the surface of the forging, it first contacts the highest end of the indicator line on the top surface of the forging and flows under gravity. During the flow, the high-definition industrial camera captures images of the cleaning fluid flow. The cleaning fluid flows over the top surface of the forging and falls into a container, where it merges with the cleaning fluid placed in the container for subsequent cleaning of the forging. During the operation of the coordination module, the cleaning solution in the container is output through the preset electronically controlled valve of the container, and the amount of cleaning solution is equal to the preset amount of replenishment solution.

6. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 1, characterized in that, The camera module stores the first set of forging surface image data and the first cleaning fluid flow image collected by the test module, pointing to the forging. The system user performs manual inspection. When the inspection result is qualified, the forging surface image data and cleaning fluid flow image corresponding to the forging are used as reference images and images. The forging surface image data and cleaning fluid flow images subsequently acquired by the system are compared and inspected with the reference images and images through the inspection module to obtain the forging error offset index. During the manual inspection of forgings by system users, if a forging is found to be unqualified, the system refreshes and re-collects surface image data and cleaning fluid flow images of the forgings transmitted by the conveyor belt at the output end of the forging production equipment. The system users then perform manual inspection on the forgings again until the manual inspection result is qualified. The surface image data and cleaning fluid flow images of the forgings corresponding to the inspection result are used as reference images, and the system enters the automated operation state to continuously determine whether the forgings are qualified.

7. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 6, characterized in that, The system is in an automated operation state. When the judgment module determines that two consecutive forgings are unqualified, the system ends the automated operation state and the user on the system end can manually inspect the forging production equipment by stopping the operation. The robotic arm module picks up the forgings that are deemed unqualified on the conveyor belt at the output end of the forging production equipment in real time, so as to complete the sorting with qualified forgings.

8. The intelligent detection system for surface defects of titanium alloy forgings based on machine vision according to claim 1, characterized in that, The preprocessing module interacts with the camera module and the coordination module via a wireless network. The coordination module interacts with the test module via a wireless network. The test module interacts with the inspection module via a wireless network. The inspection module interacts with the judgment module via a wireless network.

9. A machine vision-based intelligent detection method for surface defects in titanium alloy forgings, wherein the method is an implementation method of any one of the machine vision-based intelligent detection systems for surface defects in titanium alloy forgings as described in claims 1-8, characterized in that, include: Pick up the forging, preprocess the forging, and after the preprocessing is completed, acquire the surface image data of the forging; Coordinate the spatial orientation of the forging so that the cleaning fluid flows over the top surface of the forging. During the process of the cleaning fluid flowing over the top surface of the forging, capture images of the cleaning fluid flow. By comparing the forging surface image data and cleaning fluid flow image with the reference forging surface image data and cleaning fluid flow image, the forging error offset index is verified. Set a pass / fail threshold and compare it with the forging error deviation index to determine whether the forging is qualified. Based on the results of the forging's qualification assessment, the forgings are sorted to distinguish between qualified and unqualified forgings.

Images