Nondestructive testing method and system for carbon fiber composite panels based on machine vision

Abstract

The present invention provides a non-destructive testing method and system for carbon fiber composite plates based on machine vision, which relates to the technical field of material quality testing, wherein the method comprises: starting quality testing on carbon fiber composite plates in the process of transportation through machine vision testing equipment; when quality testing interference occurs on the carbon fiber composite plates, continuing quality testing on the carbon fiber composite plates through machine vision testing equipment so as to achieve an optimal balance between quality testing accuracy and the strength of overcoming quality testing interference. When quality testing interference occurs on the carbon fiber composite plates, the present invention continues quality testing on the carbon fiber composite plates through machine vision testing equipment on the premise of achieving an optimal balance between quality testing accuracy and the strength of overcoming quality testing interference, thereby avoiding the waste of quality testing accuracy control resources and quality testing interference overcoming resources, and improving the efficiency and applicability of quality testing.

Description

Technical Field

[0001] The present invention relates to the technical field of material quality detection, and in particular to a non-destructive detection method and system for carbon fiber composite plates based on machine vision. Background Art

[0002] Currently, to ensure the performance of the final carbon fiber composite sheet product, material quality testing is required. In this context, machine vision technology, as a non-contact, efficient, and automated detection method, has gradually become one of the important technologies for carbon fiber composite sheet material quality testing.

[0003] However, to improve the efficiency of carbon fiber composite sheet production lines, material quality inspection is typically completed during the sheet's conveying process. However, this process can introduce quality inspection interference (such as vibration, shaking, bouncing, tilting, and ambient light fluctuations). Traditional machine vision technology, when faced with these quality inspection interferences, typically pursues the highest possible quality inspection accuracy or exerts the utmost effort to overcome these interferences. This fails to achieve an optimal balance between the two, resulting in significant waste of resources for quality inspection accuracy control and interference mitigation, severely limiting quality inspection efficiency and applicability. Summary of the Invention

[0004] One of the purposes of the present invention is to provide a non-destructive testing method for carbon fiber composite plates based on machine vision to solve the problems in the background technology.

[0005] The embodiment of the present invention provides a nondestructive testing method for carbon fiber composite plates based on machine vision, comprising:

[0006] The quality of carbon fiber composite panels during transportation is inspected using machine vision inspection equipment.

[0007] When quality inspection interference occurs in the carbon fiber composite plate, the carbon fiber composite plate is continuously inspected by machine vision inspection equipment to achieve an optimal balance between quality inspection accuracy and the strength to overcome the quality inspection interference.

[0008] Optionally, the method of continuing to perform quality inspection on the carbon fiber composite plate by using machine vision inspection equipment to achieve an optimal balance between quality inspection accuracy and the strength of overcoming quality inspection interference includes:

[0009] Obtain the inspected and uninspected areas of the carbon fiber composite sheet by the machine vision inspection equipment at the last moment before the quality inspection of the carbon fiber composite sheet is disturbed;

[0010] Based on the quality inspection results of the inspected areas, predict the quality of the uninspected areas;

[0011] Based on the predicted quality conditions, plan the optimal quality inspection accuracy requirements for uninspected areas;

[0012] Obtain the interference caused by quality detection interference to undetected areas;

[0013] Planning the optimal interference overcoming strength requirement when machine vision inspection equipment performs quality inspection on uninspected areas, so that interference is overcome to the point where the quality inspection accuracy meets the optimal quality inspection accuracy requirement;

[0014] Based on the requirements of optimal interference overcoming strength, corresponding interference overcoming constraints are implemented in the process of continuing quality inspection of uninspected areas through machine vision inspection equipment.

[0015] Optionally, the predicting of the quality of the undetected area based on the quality detection result of the detected area includes:

[0016] The pre-trained quality prediction model is used to predict the quality of undetected areas based on the quality detection results of the detected areas.

[0017] Optionally, planning the optimal quality detection accuracy requirement for the undetected area based on the predicted quality condition includes:

[0018] Divide the undetected area into multiple grid areas;

[0019] Determining the quality sub-situation of each grid area from the predicted quality situation;

[0020] Based on the quality inspection accuracy requirement library, determine the quality inspection accuracy requirement of each grid area according to the quality sub-situation of each grid area;

[0021] The grid areas and their respective quality detection accuracy requirements are fused as close to the optimization target as possible to obtain multiple fused grid areas and their respective fused quality detection accuracy requirements; the fused quality detection accuracy requirement is the highest requirement among the quality detection accuracy requirements of the multiple grid areas fused into the same fused grid area;

[0022] Based on the fusion quality detection accuracy requirements of each fusion grid area, determine the optimal quality detection accuracy requirements of the undetected area;

[0023] The optimization objectives include:

[0024] The size of each fused grid area is less than the size of the single maximum inspection area of ​​the machine vision inspection device by a difference that does not exceed the difference threshold;

[0025] The quality detection accuracy requirements of multiple grid areas fused into the same fused grid area must not differ by more than the gap threshold.

[0026] Optionally, when the planned machine vision inspection equipment performs quality inspection on the uninspected area, the interference situation is overcome to the optimal interference overcoming strength requirement so that the quality inspection accuracy meets the optimal quality inspection accuracy requirement, including:

[0027] Determine the interference sub-situations of each fusion grid area in the optimal quality detection accuracy requirement from the interference situation;

[0028] Quantify the severity of the interference sub-situation in each fusion grid area, the predicted degree of aggravation trend, and the number of potential interference factors to comprehensively reflect the priority value of each fusion grid area that needs to be detected first;

[0029] Plan the inspection time period for each fused grid area when the machine vision inspection equipment performs quality inspection on each fused grid area in descending order of priority value;

[0030] Based on the interference overcoming strength requirement database, determine the interference overcoming strength requirement of each fusion grid area according to the predicted interference change of each interference sub-situation in the detection period and the respective fusion quality detection accuracy requirements;

[0031] Based on the respective interference overcoming strength requirements and respective priority values ​​of the fused grid areas, the optimal interference overcoming strength requirement is determined so that the interference situation is overcome to enable the quality detection accuracy to meet the optimal quality detection accuracy requirement.

[0032] Optionally, the quantified priority value of each fused grid area that requires priority detection, which is comprehensively reflected by the severity of the interference sub-situation of each fused grid area, the predicted degree of aggravation trend, and the number of potential interference factors, includes:

[0033] Based on the weights set in advance according to the degree of influence of the severity, the degree of aggravation trend and the number of potential interference factors on the priority value, the severity, the degree of aggravation trend and the number of potential interference factors of each fused grid area are weightedly calculated respectively to obtain the priority value of each fused grid area.

[0034] Optionally, the quality detection interference includes at least: vibration, shaking, bouncing, tilting of the carbon fiber composite plate and ambient light fluctuation.

[0035] Optional non-destructive testing methods for carbon fiber composite plates based on machine vision also include:

[0036] Based on the results of quality inspection of carbon fiber composite panels, a quality inspection early warning report is generated and output.

[0037] Optional non-destructive testing methods for carbon fiber composite plates based on machine vision also include:

[0038] The results of quality inspection of carbon fiber composite panels are uploaded to the cloud for storage.

[0039] The embodiment of the present invention provides a nondestructive testing system for carbon fiber composite plates based on machine vision, comprising:

[0040] The first quality inspection module is used to start quality inspection of the carbon fiber composite plate in the process of transportation through machine vision inspection equipment;

[0041] The second quality inspection module is used to continue to perform quality inspection on the carbon fiber composite plate through machine vision inspection equipment when quality inspection interference occurs to the carbon fiber composite plate, so as to achieve an optimal balance between quality inspection accuracy and interference overcoming strength of quality inspection interference.

[0042] The present invention has achieved the following beneficial effects:

[0043] The present invention uses machine vision inspection equipment to start quality inspection on carbon fiber composite plates in the transportation process. When quality inspection interference occurs in the carbon fiber composite plates, the machine vision inspection equipment is used to continue quality inspection on the carbon fiber composite plates under the premise of achieving an optimal balance between quality inspection accuracy and the strength of overcoming the interference of quality inspection interference, thereby avoiding the waste of quality inspection accuracy control resources and quality inspection interference overcoming resources, and improving the efficiency and applicability of quality inspection.

[0044] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0045] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0047] Figure 1 Flowchart of a nondestructive testing method for carbon fiber composite plates based on machine vision in an embodiment of the present invention;

[0048] Figure 2 is another flow chart of a nondestructive testing method for carbon fiber composite plates based on machine vision in an embodiment of the present invention;

[0049] Figure 3 This is another flow chart of the nondestructive testing method of carbon fiber composite plate based on machine vision in an embodiment of the present invention;

[0050] Figure 4 is another flow chart of a nondestructive testing method for carbon fiber composite plates based on machine vision in an embodiment of the present invention;

[0051] Figure 5 Schematic diagram of a nondestructive testing system for carbon fiber composite plates based on machine vision in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0053] Example 1:

[0054] The embodiment of the present invention provides a nondestructive testing method for carbon fiber composite plates based on machine vision, such as Figure 1 As shown, including:

[0055] S1. Use machine vision inspection equipment to conduct quality inspection on the carbon fiber composite panels during transportation;

[0056] In S1, the machine vision inspection equipment includes at least: a manipulator with a machine vision camera at its end; the machine vision inspection equipment collects quality information (such as a surface image of the carbon fiber composite sheet) during the conveying process (such as the process of conveying it to the next process), analyzes the quality information for quality issues, and implements quality inspection; because the machine vision inspection equipment does not come into contact with the sheet during the inspection process, it does not cause damage to the sheet, thereby implementing non-destructive inspection;

[0057] S2. When quality inspection interference occurs on the carbon fiber composite sheet, the machine vision inspection equipment is used to continue to conduct quality inspection on the carbon fiber composite sheet so as to achieve an optimal balance between quality inspection accuracy and the strength to overcome the quality inspection interference;

[0058] In S2, quality inspection interference includes at least: vibration, shaking, bouncing, tilting of carbon fiber composite plates and fluctuations in ambient light; quality inspection accuracy refers to the accuracy of machine vision inspection equipment in quality inspection of carbon fiber composite plates; the strength of overcoming quality inspection interference refers to the strength of machine vision inspection equipment in overcoming quality inspection interference through measures; once quality inspection interference occurs, when continuing to perform quality inspection on carbon fiber composite plates through machine vision inspection equipment, it is ensured that the quality inspection accuracy and the strength of overcoming quality inspection interference reach an optimal balance.

[0059] The present invention uses machine vision inspection equipment to start quality inspection on carbon fiber composite plates in the transportation process. When quality inspection interference occurs in the carbon fiber composite plates, the machine vision inspection equipment is used to continue quality inspection on the carbon fiber composite plates under the premise of achieving an optimal balance between quality inspection accuracy and the strength of overcoming the interference of quality inspection interference, thereby avoiding the waste of quality inspection accuracy control resources and quality inspection interference overcoming resources, and improving the efficiency and applicability of quality inspection.

[0060] Example 2:

[0061] In the embodiment of the present invention, Figure 2 As shown, in S2, the carbon fiber composite plate is continuously subjected to quality inspection by machine vision inspection equipment so as to achieve an optimal balance between quality inspection accuracy and the strength of overcoming quality inspection interference, including:

[0062] S21. Obtain the inspected area and the uninspected area of ​​the carbon fiber composite sheet in the inspection area of ​​the machine vision inspection device at the last moment before the quality inspection interference occurs in the carbon fiber composite sheet;

[0063] In S21, the machine vision inspection equipment will perform regional inspection on the carbon fiber composite plate. At the last moment before the carbon fiber composite plate encounters quality inspection interference, there will be some areas in the inspection area that have been inspected, i.e., inspected areas, and there will be some areas that have not been inspected, i.e., uninspected areas.

[0064] S22. Predicting the quality of undetected areas based on the quality detection results of the detected areas. S22 specifically includes: using a pre-trained quality prediction model to predict the quality of undetected areas based on the quality detection results of the detected areas.

[0065] In S22, the machine vision inspection equipment has inspected the inspected area and will generate quality inspection results. The quality problems of the carbon fiber plate are positionally correlated (such as bubbles, scratches and cracks have positional continuity). Therefore, the quality situation of the uninspected area can be predicted based on the quality inspection results of the inspected area. The quality situation refers to the possible quality problems in the uninspected area. When pre-training the quality situation prediction model, a large amount of quality information of carbon fiber composite plates marked with quality problems (such as plate surface images marked with bubble, scratch and crack location areas) is used for machine learning training, and the model trained to convergence is used as the quality situation prediction model; so that the quality situation prediction model can predict the quality situation of the uninspected area based on the quality inspection results of the inspected area; specifically, first, it is necessary to collect and organize the labeled image data containing the quality problems of the carbon fiber composite plate and use it as a data set; the data set should include images with bubbles, scratches, Images with defective areas such as cracks can be annotated using polygonal or rectangular boxes to ensure sufficient diversity in the dataset, including different types of quality issues, different shooting angles, lighting conditions, and composite surface conditions. The dataset is divided into training, validation, and test sets in a ratio of 7:1.5:1.5 (i.e., 70% for training, 15% for validation, and 15% for testing) to ensure that the model can be effectively evaluated and tuned during training. The existing public deep learning framework PyTorch is selected for machine learning training, and the cross-entropy loss function is selected as the loss function. Machine learning training is performed on the training set until the loss function converges. The performance of the model in training is evaluated on the validation set, and hyperparameters such as the learning rate and batch size are adjusted according to actual needs. After training to convergence, the final model is saved and tested on the test set. The final model that passes the test is used as the quality prediction model.

[0066] S23. Based on the predicted quality conditions, plan the optimal quality inspection accuracy requirements for the uninspected areas;

[0067] In S23, for the predicted quality condition of the undetected area, it is possible to plan how to control the accuracy of quality detection of the undetected area so that the quality condition can be detected and verified, and then the optimal quality detection accuracy requirement of the undetected area can be planned based on the quality condition; Figure 3 As shown, S23 specifically includes:

[0068] S231, dividing the undetected area into a plurality of grid areas;

[0069] In S231, the technician may pre-set a division rule based on actual needs and divide the undetected area into multiple grid areas according to the set division plan. For example, the total area of ​​the undetected area is 100 square meters, which can be divided into 10 grid areas of 10 square meters each.

[0070] S232, determining the quality sub-conditions of each grid area from the predicted quality conditions;

[0071] In S232, the quality sub-situation is the quality problem situation related to each grid area in the predicted quality situation;

[0072] S233. Based on the quality inspection accuracy requirement database, determine the quality inspection accuracy requirement for each grid area according to the quality sub-situation of each grid area;

[0073] In S233, the quality inspection accuracy requirement library pre-sets quality inspection accuracy requirements corresponding to different quality sub-situations. When the machine vision inspection equipment performs quality inspection at the quality inspection accuracy required by the quality inspection accuracy requirement, it can verify whether the corresponding quality sub-situation exists in the inspection area; the quality inspection accuracy requirements of each grid area are directly queried and determined based on the library; the quality inspection accuracy requirements in the quality inspection accuracy requirement library can be set in advance by technical personnel based on actual needs and combined with the product characteristics of the carbon fiber composite plate. For example, for quality sub-situations that require detection of minor defects (such as cracks, bubbles, surface micro-blemishes, etc.), the accuracy is required to be at the micron level or higher, that is, the quality inspection accuracy requirement is set to machine vision inspection accuracy reaching the micron level or higher; the quality inspection accuracy requirements in the quality inspection accuracy requirement library can also be set using the quality requirements in standards such as GB / T21490-2008 "Carbon Fiber Sheets for Structural Reinforcement and Repair" and GB / T44542-2024 "Determination of Ash Content and Impurity Content of Carbon Fiber and Its Precursor";

[0074] S234. Fusing the grid areas and their respective quality detection accuracy requirements as close to the optimization target as possible to obtain multiple fused grid areas and their respective fused quality detection accuracy requirements; wherein the fused quality detection accuracy requirement is the highest requirement among the quality detection accuracy requirements of the multiple grid areas fused into the same fused grid area;

[0075] In S234, fusion refers to fusing all grid areas into multiple fused grid areas, and fusing the quality inspection accuracy requirements of the multiple grid areas fused into the same fused grid area into a single fused quality inspection accuracy requirement. For the fusion process, constraints closest to the optimization goal are imposed so that the obtained fused grid area and its respective fused quality inspection accuracy requirements are optimized, which is used to determine the optimal quality inspection accuracy requirement for the uninspected area. For the fusion of quality inspection accuracy requirements, the highest requirement among the quality inspection accuracy requirements of the multiple grid areas fused into the same fused grid area is taken as the fused quality inspection accuracy requirement of the corresponding fused grid area (when the machine vision inspection equipment performs quality inspection on the fused grid area according to the highest requirement, the quality inspection accuracy requirements of all grid areas can be met). Specifically, for example, the uninspected area is 100 square meters, which is divided into 10 grid areas of 10 square meters each. The multiple grid areas fused into the same fused grid area and their respective quality inspection accuracy requirements are grid area A and the inspection accuracy requirement of 0.5 mm, and grid area B and the inspection accuracy requirement of 1 mm, respectively. If the fused quality inspection accuracy requirement takes the highest requirement, the inspection accuracy requirement of 0.5 mm is taken as the fused quality inspection accuracy requirement.

[0076] S235. Determine the optimal quality detection accuracy requirement for the undetected area based on the fusion quality detection accuracy requirement of each fusion grid area;

[0077] In S235, the fusion quality detection accuracy requirements of each fusion grid area are taken as the optimal quality detection accuracy requirements of the undetected area;

[0078] The optimization objectives include:

[0079] Objective 1: The size of each fused grid area must not exceed the difference threshold value for the size difference below the single maximum inspection area size of the machine vision inspection equipment;

[0080] In goal one, the single maximum detection area size of the machine vision inspection equipment refers to the size of the area that it can cover for quality inspection at a single time, which is determined by its own configuration; the difference threshold is a pre-set threshold representing a smaller size difference; the smaller the difference between the size of each fused grid area and the size of the single maximum detection area, the higher the utilization rate of the fused grid area for quality inspection by the machine vision inspection equipment, and the difference between the size of each fused grid area and the size of the single maximum detection area does not exceed the difference threshold as the optimization goal, and the fusion closest to the optimization goal is performed to maximize the utilization rate of the single detection of the machine vision inspection equipment; specifically, for example: the single maximum detection area size of the machine vision inspection equipment is 20 square meters, and the difference threshold is 5 square meters. If the size of the fused grid area is 18 square meters, the difference is 2 square meters, which meets the threshold limit, and then goal one is achieved;

[0081] Objective 2: The quality inspection accuracy requirements of multiple grid areas fused into the same fusion grid area must not exceed the gap threshold between the two requirements.

[0082] In the second goal, the requirement gap refers to the absolute value of the difference between the required accuracies of the two quality detection accuracy requirements; the gap threshold is a pre-set threshold representing a smaller gap threshold; the smaller the requirement gap between the quality detection accuracy requirements of the multiple grid areas integrated into the same fusion grid area, the higher the requirement is when the highest requirement is taken as the fusion quality detection accuracy requirement, the more it can avoid the situation where the area that does not actually need too high accuracy requirements for detection is detected with high accuracy, that is, the more it can avoid the waste of accuracy setting; the requirement gap between the quality detection accuracy requirements of the multiple grid areas integrated into the same fusion grid area does not exceed the gap threshold as the optimization goal, and the fusion closest to the optimization goal is performed; specifically, for example: setting the gap threshold to 0.3mm, the requirement gaps between the quality detection accuracy requirements of the multiple grid areas integrated into the same fusion grid area are 0.2mm, 0.1mm, and 0.02mm respectively, then the second goal is achieved;

[0083] S24. Obtaining interference caused by quality detection interference to undetected areas;

[0084] In S24, the interference condition refers to the specific interference condition of the undetected area in the quality detection interference;

[0085] S25. Planning the optimal interference overcoming strength requirement for overcoming interference when the machine vision inspection equipment performs quality inspection on the uninspected area so that the quality inspection accuracy meets the optimal quality inspection accuracy requirement;

[0086] In S25, the strength of overcoming the quality detection interference determines the quality detection accuracy. The greater the strength of overcoming the quality detection interference, the more the quality detection accuracy is improved. The optimal balance between the two is that the interference is overcome to the extent that the quality detection accuracy just meets the optimal quality detection accuracy requirement. Therefore, after determining the interference and the optimal quality detection accuracy requirement, the optimal interference overcoming strength requirement is planned. Figure 4 As shown, S25 specifically includes:

[0087] S251, determining the interference sub-situations of each fusion grid area in the optimal quality detection accuracy requirement from the interference situation;

[0088] In S251, the interference sub-case is the interference case caused by the fusion grid area in the interference case;

[0089] S252: Quantify the priority value of each fused grid area that requires priority detection, which is comprehensively reflected by the severity of the interference sub-situation in each fused grid area, the predicted degree of aggravation trend, and the number of potential interference factors;

[0090] In S252, the severity of the interference sub-situation refers to the severity of the interference caused to the fusion grid area by the interference sub-situation, which can be obtained by evaluating the severity of the interference sub-situation according to a pre-set severity evaluation system (including the severity of different interference sub-situations); the degree of aggravation trend refers to the degree of the trend of the interference sub-situation showing an aggravating trend. When predicting it, a model that is pre-trained on machine learning based on a large number of development trends of different historical interference sub-situations can be used as training samples to make predictions based on the interference sub-situations; specifically, when training the model, first collect the development trends of various interference sub-situations in historical data, such as time series, frequency, intensity, influencing factors, etc., use these data to create training samples, and use the existing publicly available deep neural network LSTM (LSTM is suitable for processing trend data and can effectively capture long-term dependencies in time series, thereby improving the accuracy of trend prediction) to perform machine learning training based on the training samples. Cross-validation and hyperparameter tuning are used in the training process to optimize model performance and avoid overfitting, and the model is obtained at the end of training; potential interference factors refer to potential interference factors that may exist in the fusion grid area reflected by the interference sub-situation, which can be set in advance by technical personnel according to the potential interference factors that may exist in the fusion grid area reflected by different interference sub-situations. For example: the interference sub-situation is the bouncing action of the plate, but during its bouncing process, surface light reflection or refraction may occur, affecting the normal operation of the machine vision camera on the machine vision inspection equipment, then the potential interference factor is the surface light reflection or refraction phenomenon; the greater the severity, the degree of aggravation trend and the number of potential interference factors, the more priority detection is needed to avoid the difficulty of subsequent quality inspection of the corresponding fusion grid area, and the priority value of each fusion grid area that needs to be inspected first is quantified by these three factors; specifically, when quantifying these three factors, these three factors are weighted to obtain the corresponding priority value;

[0091] S253, planning the inspection time period for each fused grid area when the machine vision inspection equipment performs quality inspection on each fused grid area in descending order of priority value;

[0092] In S253, based on the inspection speed of the machine vision inspection device (the inspection speed is expressed in units of size / second), the size of each fused grid area, etc., the inspection time of each fused grid area can be planned. Then, in combination with the current time, the inspection time period of each fused grid area is planned when the machine vision inspection device performs quality inspection on each fused grid area in descending order of priority value. The inspection time period is a certain time period in the future. Specifically, for example, if the inspection time of fused grid area C is 5 seconds and the inspection time of fused grid area D is 10 seconds, and the priority value of fused grid area C is greater than that of fused grid area D, then the inspection time period of each fused grid area is planned to be within the next 5 seconds for fused grid area C and within the next 5 to 15 seconds for fused grid area D.

[0093] S254. Based on the interference overcoming strength requirement database, determine the interference overcoming strength requirement for each fusion grid area according to the predicted interference change of the interference sub-situation of each fusion grid area during the detection period and the respective fusion quality detection accuracy requirements;

[0094] In S254, the prediction of interference change refers to the situation in which the predicted interference sub-situation will change during the detection period. When predicting it, a model that is pre-trained based on a large number of historical interference sub-situations as training samples can be used for machine learning training to make predictions based on the interference change situation. When training the model, it is first necessary to collect the interference change situations of different historical interference sub-situations, covering factors such as interference type, intensity, and equipment status. Through data cleaning and annotation, a training sample set is constructed. The neural network LSTM suitable for time series prediction is developed using the PyTorch framework and trained in the Anaconda environment. The training sample set is divided into training set, validation set, and test set in a ratio of 7:1:2. The cross entropy or mean square error loss function is used and the optimizer Adam is used for training. The interference overcoming strength requirement library pre-sets the interference overcoming strength requirements corresponding to different groups of predicted interference changes and fusion quality detection accuracy requirements. Under the interference overcoming strength required by the interference overcoming strength requirements, corresponding interference overcoming measures (such as vibration, shaking, bouncing, etc.) are taken. During the tilting action, motion compensation is performed on the captured surface image. The greater the amount of compensation, the greater the interference overcoming force. The motion compensation of the captured image of the moving object here belongs to the scope of existing technology and will not be elaborated on. Overcoming the corresponding predicted interference changes can ensure that the corresponding fusion quality detection accuracy requirements are met. The library can be directly checked to determine the interference overcoming force requirements for each fusion grid area. The interference overcoming force requirements in the interference overcoming force requirement library can be set by technical personnel in advance through experiments for the interference overcoming force of the interference overcoming measures required for fusion grid areas with different predicted interference changes and fusion quality detection accuracy requirements. Specifically, for example: assuming that in an actual production environment, there is a fusion grid area E, whose predicted interference change is that vibration and light reflection have a gradually increasing trend, and the fusion quality detection accuracy requirement is 1mm, then the required interference overcoming measure is to use stronger motion compensation for the machine vision camera to offset the vibration, and at the same time use the camera image compensation algorithm (camera motion compensation and image compensation both belong to the scope of existing technology and will not be elaborated on) to adapt to light changes, then the interference overcoming force requirement is set to a higher force of 10.

[0095] S255. Based on the respective interference overcoming strength requirements and respective priority values ​​of the fused grid regions, determine the optimal interference overcoming strength requirement for overcoming the interference to enable the quality detection accuracy to meet the optimal quality detection accuracy requirement;

[0096] In S255, the interference overcoming strength requirements and the priority values ​​of the respective fused grid areas are used as the optimal interference overcoming strength requirements. In this way, when the machine vision inspection device continues to perform quality inspection on the uninspected area, corresponding interference overcoming constraints are applied, so that the machine vision inspection device can perform quality inspection on the fused grid areas in descending order of priority values. Each time a fused grid area is detected, the corresponding interference change situation is overcome according to the corresponding interference overcoming strength requirements.

[0097] S26. Based on the optimal interference overcoming strength requirement, during the process of continuing to perform quality inspection on the uninspected area using the machine vision inspection equipment, corresponding interference overcoming constraints are performed;

[0098] In S26, after the planning is completed, in the process of continuing to perform quality inspection on the uninspected area through the machine vision inspection equipment, based on the optimal interference overcoming strength requirement, corresponding interference overcoming constraints are performed, so that the quality inspection accuracy meets the optimal quality inspection accuracy requirement, and the quality inspection accuracy and the strength of overcoming quality inspection interference reach an optimal balance; performing interference overcoming constraints means controlling the process of continuing to perform quality inspection on the uninspected area through the machine vision inspection equipment to be strictly executed in accordance with the optimal interference overcoming strength requirement.

[0099] The embodiment of the present invention predicts the quality of the uninspected area based on the quality inspection results of the inspected area, and based on its targeted planning of the optimal quality inspection accuracy requirements for the uninspected area, finally plans that when the machine vision inspection equipment performs quality inspection on the uninspected area, the interference situation is overcome to the point that the quality inspection accuracy meets the optimal interference overcoming strength requirements of the optimal quality inspection accuracy requirements, so that in the process of continuing to perform quality inspection on the uninspected area through the machine vision inspection equipment, when the optimal interference overcoming strength requirements are used to perform corresponding interference overcoming constraints, the quality inspection accuracy of the machine vision inspection equipment just meets the optimal quality inspection accuracy requirements, and no excessive or insufficient interference overcoming is performed, so that the quality inspection accuracy and the strength of overcoming quality inspection interference reach an optimal balance, which greatly avoids the waste of quality inspection accuracy control resources and quality inspection interference overcoming resources, improves the rationality of the configuration of the two resources, and improves the quality inspection efficiency.

[0100] A quality inspection accuracy requirement library is introduced to quickly determine the quality inspection accuracy requirements of each grid area. Two optimization goals are set, and each grid area and its respective quality inspection accuracy requirements are fused as close to the optimization goal to obtain multiple fused grid areas and their respective fused quality inspection accuracy requirements. Finally, the fused quality inspection accuracy requirements of each fused grid area are used as the optimal quality inspection accuracy requirements for the uninspected area. This greatly improves the accuracy, comprehensiveness and efficiency of the optimal quality inspection accuracy requirement planning, and improves the accuracy of the planning when the optimal quality inspection accuracy requirement is used as the optimal interference overcoming strength requirement planning, further enabling the quality inspection accuracy and the interference overcoming strength of quality inspection interference to accurately achieve the optimal balance.

[0101] Determine the interference sub-situation of each fused grid area, quantify the severity of the interference sub-situation of each fused grid area, the predicted degree of aggravation trend and the priority value of each fused grid area that needs to be detected first, and use the interference overcoming strength requirement of each fused grid area and its priority value as the optimal interference overcoming strength requirement. This not only ensures that the area with greater difficulty in subsequent quality inspection will be inspected first, but also realizes that the interference overcoming strength requirement can overcome the predicted interference changes of the fused grid area during the inspection period, greatly improving the suitability of the optimal interference overcoming strength requirement planning, and further improving the system's ability to achieve an optimal balance between quality inspection accuracy and interference overcoming strength of quality inspection interference when quality inspection interference occurs.

[0102] Example 3:

[0103] In the embodiment of the present invention, the quantified severity of the interference sub-situation of each fused grid area, the predicted degree of aggravation trend, and the number of potential interference factors comprehensively reflect the priority value of each fused grid area that needs to be detected first, including:

[0104] Based on the weights set in advance according to the degree of influence of the severity, the degree of aggravation trend and the number of potential interference factors on the priority value, the severity, the degree of aggravation trend and the number of potential interference factors of each fused grid area are weightedly calculated respectively to obtain the priority value of each fused grid area.

[0105] In the embodiment of the present invention, the formula for weighted calculation of the severity, aggravation trend, and number of potential interference factors of a single fused grid area is: ,in, is the priority value, For severity, To increase the trend, is the number of potential interference factors, is the severity weight, To increase the weight of the trend degree, is the weight of the number of potential interference factors. Specifically, for example, suppose there is a fused grid area F, whose severity, aggravation trend, and number of potential interference factors are 8, 5, and 3, respectively. The weight of the severity is set to 0.4, the weight of the aggravation trend is set to 0.3, and the weight of the number of potential interference factors is set to 0.3. Then, the priority value of fused grid area F = (8 × 0.4) + (5 × 0.3) + (3 × 0.3) = 3.2 + 1.5 + 0.9 = 5.6.

[0106] Example 4:

[0107] In the embodiment of the present invention, the quality detection interference includes at least: vibration, shaking, bouncing, tilting of the carbon fiber composite plate and ambient light fluctuation.

[0108] In an embodiment of the present invention, quality detection interference can be interference situations such as vibration, shaking, bouncing, tilting of the carbon fiber composite plate and fluctuations in ambient light. Specifically, vibration interference may come from the operation of the equipment or the external environment, affecting the accuracy of the sensor, resulting in image blur or position offset. Shaking and bouncing interference may be caused by the movement of the equipment, causing visual errors and morphological distortion. In addition, the tilting movement of the carbon fiber composite plate may also cause changes in the viewing angle, affecting the detection results. Fluctuations in ambient light may cause overexposure or underexposure of the image.

[0109] Example 5:

[0110] In an embodiment of the present invention, the nondestructive testing method for carbon fiber composite plates based on machine vision further includes:

[0111] Based on the results of quality inspection of carbon fiber composite panels, a quality inspection early warning report is generated and output.

[0112] In an embodiment of the present invention, the results of a quality inspection of a carbon fiber composite sheet material include quality issues present in the sheet material. Based on these issues, a quality inspection warning report can be generated and output according to a preset report template. The output can be reviewed by relevant management personnel. Specifically, the preset report template includes a report title, basic information (such as the inspection date and the equipment used), inspection objectives and methods, inspection results (such as specific data on cracks, bubbles, and other issues), problem analysis and assessment, and treatment recommendations. The system automatically fills in most of this information and can display images and data tables of the issues, helping management personnel quickly understand the quality issues.

[0113] Example 6:

[0114] In an embodiment of the present invention, a nondestructive testing method for carbon fiber composite plates based on machine vision is characterized by further comprising:

[0115] The results of quality inspection of carbon fiber composite panels are uploaded to the cloud for storage.

[0116] In an embodiment of the present invention, the results of quality inspections of carbon fiber composite panels can also be uploaded to the cloud for storage, allowing for easy retrieval and review at any time. Specifically, the quality inspection results are uploaded to the cloud for storage via a wireless network. This allows users to access the cloud at any time through terminal devices (such as computers, mobile phones, etc.) to retrieve and review historical data, ensuring convenient access and long-term storage of information. For example, in actual applications, when managers need to view the inspection results of a batch of panels, they do not need to go on-site to search. Instead, they simply enter the relevant batch number through the cloud system to quickly retrieve and review the historical inspection data for that batch, greatly improving management and maintenance efficiency.

[0117] Example 7:

[0118] The embodiment of the present invention provides a nondestructive testing system for carbon fiber composite plates based on machine vision, such as Figure 5 As shown, including:

[0119] The first quality inspection module 1 is used to start quality inspection of the carbon fiber composite plate in the process of transportation through machine vision inspection equipment;

[0120] The second quality inspection module 2 is used to continue to perform quality inspection on the carbon fiber composite plate through machine vision inspection equipment when quality inspection interference occurs to the carbon fiber composite plate, so as to achieve an optimal balance between quality inspection accuracy and interference overcoming strength of quality inspection interference.

[0121] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A nondestructive testing method for carbon fiber composite plates based on machine vision, characterized in that: include: The quality of carbon fiber composite panels during transportation is inspected using machine vision inspection equipment. When quality inspection interference occurs on the carbon fiber composite sheet, the machine vision inspection equipment is used to continue to conduct quality inspection on the carbon fiber composite sheet so as to achieve the optimal balance between quality inspection accuracy and the strength to overcome the quality inspection interference. The carbon fiber composite plate is continuously subjected to quality inspection by machine vision inspection equipment to achieve an optimal balance between quality inspection accuracy and the strength of overcoming quality inspection interference, including: Obtain the inspected and uninspected areas of the carbon fiber composite sheet by the machine vision inspection equipment at the last moment before the quality inspection of the carbon fiber composite sheet is disturbed; Based on the quality inspection results of the inspected areas, predict the quality of the uninspected areas; Based on the predicted quality conditions, plan the optimal quality inspection accuracy requirements for uninspected areas; Obtain the interference caused by quality detection interference to undetected areas; Planning the optimal interference overcoming strength requirement when machine vision inspection equipment performs quality inspection on uninspected areas, so that interference is overcome to the point where the quality inspection accuracy meets the optimal quality inspection accuracy requirement; Based on the requirements of optimal interference overcoming strength, corresponding interference overcoming constraints are applied during the process of continuing quality inspection of uninspected areas using machine vision inspection equipment; The optimal quality inspection accuracy requirements for uninspected areas are planned based on the predicted quality conditions, including: Divide the undetected area into multiple grid areas; Determining the quality sub-situation of each grid area from the predicted quality situation; Based on the quality inspection accuracy requirement library, determine the quality inspection accuracy requirement of each grid area according to the quality sub-situation of each grid area; The grid areas and their respective quality detection accuracy requirements are fused as close to the optimization target as possible to obtain multiple fused grid areas and their respective fused quality detection accuracy requirements; the fused quality detection accuracy requirement is the highest requirement among the quality detection accuracy requirements of the multiple grid areas fused into the same fused grid area; Based on the fusion quality detection accuracy requirements of each fusion grid area, determine the optimal quality detection accuracy requirements of the undetected area; The optimization objectives include: The size of each fused grid area is less than the size of the single maximum inspection area of ​​the machine vision inspection device by a difference that does not exceed the difference threshold; The quality detection accuracy requirements of multiple grid areas fused into the same fused grid area must not exceed the gap threshold. When the planned machine vision inspection equipment performs quality inspection on the uninspected area, the interference situation is overcome to the optimal interference overcoming strength requirement so that the quality inspection accuracy meets the optimal quality inspection accuracy requirement, including: Determine the interference sub-situations of each fusion grid area in the optimal quality detection accuracy requirement from the interference situation; Quantify the severity of the interference sub-situation in each fusion grid area, the predicted aggravation trend, and the number of potential interference factors to comprehensively reflect the priority value of each fusion grid area that needs to be detected first; Plan the inspection time period for each fused grid area when the machine vision inspection equipment performs quality inspection on each fused grid area in descending order of priority value; Based on the interference overcoming strength requirement database, determine the interference overcoming strength requirement of each fusion grid area according to the predicted interference change of each interference sub-situation in the detection period and the respective fusion quality detection accuracy requirements; Based on the respective interference overcoming strength requirements and respective priority values ​​of the fused grid areas, the optimal interference overcoming strength requirement is determined so that the interference situation is overcome to the point where the quality detection accuracy meets the optimal quality detection accuracy requirement.

2. The nondestructive testing method for carbon fiber composite plates based on machine vision according to claim 1, characterized in that: The method of predicting the quality of undetected areas based on the quality detection results of the detected areas includes: The pre-trained quality prediction model is used to predict the quality of undetected areas based on the quality detection results of the detected areas.

3. The nondestructive testing method for carbon fiber composite plates based on machine vision according to claim 1, characterized in that: The priority value of each fusion grid area that needs to be detected first, which is comprehensively reflected by the severity of the interference sub-situation of each fusion grid area, the predicted degree of aggravation trend, and the number of potential interference factors, includes: Based on the weights set in advance according to the degree of influence of the severity, the degree of aggravation trend and the number of potential interference factors on the priority value, the severity, the degree of aggravation trend and the number of potential interference factors of each fused grid area are weightedly calculated respectively to obtain the priority value of each fused grid area.

4. The nondestructive testing method for carbon fiber composite plates based on machine vision according to claim 1, characterized in that: The quality detection interference includes at least: vibration, shaking, bouncing, tilting of the carbon fiber composite plate and ambient light fluctuation.

5. The nondestructive testing method for carbon fiber composite plates based on machine vision according to claim 1, characterized in that: Also includes: Based on the results of quality inspection of carbon fiber composite panels, a quality inspection early warning report is generated and output.

6. The nondestructive testing method for carbon fiber composite plates based on machine vision according to claim 1, characterized in that: Also includes: The results of quality inspection of carbon fiber composite panels are uploaded to the cloud for storage.

7. A nondestructive testing system for carbon fiber composite plates based on machine vision, the system applying the nondestructive testing method for carbon fiber composite plates based on machine vision according to any one of claims 1 to 6, characterized in that: include: The first quality inspection module is used to start quality inspection of the carbon fiber composite plate during transportation by using machine vision inspection equipment; The second quality inspection module is used to continue to perform quality inspection on the carbon fiber composite plate through machine vision inspection equipment when quality inspection interference occurs to the carbon fiber composite plate, so as to achieve an optimal balance between quality inspection accuracy and interference overcoming strength of quality inspection interference.

Images