Online defect detection method and system for calcium silicate veneer

Abstract

The present invention provides an online defect detection method and system for calcium silicate decorative panels, which relates to the technical field of online defect detection, including: transmitting the target calcium silicate decorative panel to the target detection area; when the first detection component completes positioning, turning on the supplementary light device combination, performing image acquisition to obtain a decorative panel image set; performing surface defect detection through the surface defect recognition model embedded in the surface defect detection component to obtain surface defect feature information; transmitting to the second detection component, performing ultrasonic detection, and obtaining an ultrasonic flaw detection curve; performing waveform feature analysis to obtain internal defect feature information; and determining the online defect detection result based on the surface defect feature information and the internal defect feature information. The present invention solves the technical problem that the traditional defect detection method usually uses a single detection device, is prone to missed detection or false detection, and cannot fully and accurately identify surface and internal defects, resulting in poor detection effect and low detection efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of online defect detection, and in particular to an online defect detection method and system for calcium silicate decorative panels. Background Art

[0002] During the production process of calcium silicate veneers, various defects may occur, such as scratches, cracks, bubbles, contamination, as well as internal cracks, voids, inclusions, etc. These defects not only affect the appearance of the veneer, but also reduce its structural performance and service life. Traditional manual inspection or single inspection methods, such as only using visual inspection or ultrasonic inspection, are prone to missed inspections or false inspections, and cannot fully and accurately identify surface and internal defects; moreover, most inspection methods only focus on defects in a single aspect and lack a comprehensive assessment of the overall quality, which affects product quality control and production decisions. Summary of the invention

[0003] The present application provides an online defect detection method and system for calcium silicate decorative panels, aiming to solve the technical problems that traditional defect detection methods usually use a single detection device, are prone to missed detection or false detection, and cannot comprehensively and accurately identify surface and internal defects, resulting in poor detection results and low detection efficiency.

[0004] In a first aspect disclosed in the present application, a method for online defect detection of calcium silicate decorative panels is provided, the method comprising: conveying a target calcium silicate decorative panel to a target detection area by means of a conveying device, wherein the target detection area sequentially deploys a first detection component and a second detection component along a conveying direction, wherein the first detection component comprises an image acquisition device combination and a fill light device combination arranged according to a predetermined arrangement scheme, and the second detection component comprises an ultrasonic detector; when the target calcium silicate decorative panel is positioned in the first detection component, the fill light device combination is turned on, and an image of the target calcium silicate decorative panel is captured by means of the image acquisition device combination to obtain a decorative panel image set, Among them, the panel image set includes multiple images at different angles; the panel image set is subjected to surface defect detection by a surface defect recognition model embedded in a surface defect detection component to obtain surface defect feature information; when the detection is completed, the transmission device is started to transmit the target calcium silicate panel to a second detection component, and the target calcium silicate panel is subjected to ultrasonic detection by the ultrasonic detector to obtain an ultrasonic flaw detection curve; the ultrasonic flaw detection curve is subjected to waveform feature analysis to obtain internal defect feature information according to the analysis result; based on the surface defect feature information and the internal defect feature information, the online defect detection result of the target calcium silicate panel is determined.

[0005] According to a second aspect disclosed in the present application, an online defect detection system for calcium silicate decorative panels is provided, and the system is used for the above-mentioned online defect detection method for calcium silicate decorative panels, and the system includes: a conveying module, and the conveying module is used to convey the target calcium silicate decorative panel to a target detection area through a conveying device, wherein the target detection area sequentially deploys a first detection component and a second detection component along the conveying direction, wherein the first detection component includes an image acquisition device combination and a fill light device combination arranged according to a predetermined arrangement scheme, and the second detection component includes an ultrasonic detector; an image acquisition module, and the image acquisition module is used to turn on the fill light device combination when the target calcium silicate decorative panel is positioned in the first detection component, and to perform image acquisition of the target calcium silicate decorative panel through the image acquisition device combination to obtain a decorative panel image set, wherein the decorative panel image set The combination includes multiple images at different angles; a defect detection module, which is used to perform surface defect detection on the panel image set through a surface defect recognition model embedded in a surface defect detection component to obtain surface defect feature information; an ultrasonic detection module, which is used to start the transmission device to transmit the target calcium silicate panel to a second detection component after the detection is completed, and perform ultrasonic detection of the target calcium silicate panel through the ultrasonic detector to obtain an ultrasonic flaw detection curve; a waveform feature analysis module, which is used to perform waveform feature analysis on the ultrasonic flaw detection curve and obtain internal defect feature information according to the analysis result; a detection result acquisition module, which is used to determine the online defect detection result of the target calcium silicate panel based on the surface defect feature information and the internal defect feature information.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] Combining image acquisition equipment and ultrasonic detectors, through multi-angle image acquisition and ultrasonic testing, comprehensive detection of surface and internal defects of calcium silicate decorative panels is achieved, and the accuracy of defect detection is improved; the calcium silicate decorative panels are sequentially transmitted to different detection components by conveying equipment, and the automated process design makes the detection process efficient and continuous, which meets the needs of mass production. Compared with manual inspection, automated inspection greatly reduces time and labor costs; through surface defect recognition models and ultrasonic feature analysis, the surface and internal defect feature information of the decorative panels can be fully obtained, and these two types of information can be comprehensively analyzed to provide comprehensive online defect detection results, thereby improving the reliability of quality control. In summary, the online defect detection method for calcium silicate decorative panels reduces the influence of human factors through automated and intelligent detection methods, improves the stability and consistency of the detection results, and improves the quality control level of calcium silicate decorative panels.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of an online defect detection method for calcium silicate decorative panels provided in an embodiment of the present application.

[0010] Figure 2 A schematic diagram of the structure of an online defect detection system for calcium silicate panels provided in an embodiment of the present application.

[0011] Explanation of the reference numerals: transmission module 10 , image acquisition module 20 , defect detection module 30 , ultrasonic detection module 40 , waveform feature analysis module 50 , detection result acquisition module 60 . DETAILED DESCRIPTION

[0012] The embodiments of the present application provide an online defect detection method for calcium silicate panels, thereby solving the technical problem that traditional defect detection methods usually use a single detection device, are prone to missed detections or false detections, and cannot comprehensively and accurately identify surface and internal defects, resulting in poor detection effects and low detection efficiency.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] like Figure 1 As shown, the embodiment of the present application provides an online defect detection method for calcium silicate decorative panels, the method comprising:

[0015] The target calcium silicate panel is transported to a target detection area by a transport device, wherein the target detection area sequentially deploys a first detection component and a second detection component along a transport direction, wherein the first detection component includes an image acquisition device combination and a fill light device combination arranged according to a predetermined arrangement scheme, and the second detection component includes an ultrasonic detector.

[0016] The calcium silicate decorative plate is moved to the target detection area by using a conveyor belt or other conveying equipment, and the conveying equipment is used to accurately convey the decorative plate into the detection area. The target detection area is sequentially provided with two detection components along the conveying direction, namely, a first detection component and a second detection component.

[0017] For the first detection component, according to a predetermined layout plan, multiple cameras are arranged in the first detection component. These cameras cover multiple angles of the panel to ensure that the image of the panel surface is fully acquired, and are equipped with appropriate fill lighting equipment, such as LED lights, to provide uniform and sufficient lighting conditions to eliminate possible shadows and uneven light problems on the panel surface, thereby improving the quality of image acquisition.

[0018] For the second detection component, an ultrasonic detector is arranged in the second detection component for performing ultrasonic detection on the trim. The ultrasonic detector can generate and receive ultrasonic waves and record the reflected signals when the ultrasonic waves propagate inside the trim, thereby detecting internal defects.

[0019] Through this target inspection area, the calcium silicate veneer can be fully inspected for defects, including surface defects and internal defects, to ensure that the quality of the veneer meets the requirements.

[0020] Furthermore, the method for obtaining the predetermined layout plan includes:

[0021] Obtain a defect detection task, including defect detection accuracy; collect and obtain the cross-sectional shape and cross-sectional dimensions of the target calcium silicate decorative panel; select a fill light device combination based on the cross-sectional shape and cross-sectional dimensions, select the image acquisition device combination in combination with the defect detection accuracy, and determine the device layout position coordinates; add the device layout position coordinates to the predetermined layout plan.

[0022] A defect detection task is received, which includes specific requirements and objectives of the detection, including the type of defects to be detected and the accuracy requirements of the detection, for example, the detection accuracy of the defect size, etc.

[0023] Use measuring equipment, such as a laser rangefinder, caliper, etc., to measure the cross-sectional shape and cross-sectional dimensions of the target calcium silicate veneer, such as a rectangular, trapezoidal, etc., and the cross-sectional dimensions such as width, height, thickness, etc.

[0024] According to the shape and size of the section, select the appropriate combination and number of fill-in lighting equipment to ensure uniform lighting and no shadows, and use a combination of LED lights at different angles to cover the entire section. According to the defect detection accuracy requirements, select image acquisition equipment with corresponding resolution and shooting speed, such as industrial cameras, and design the layout of image acquisition equipment to ensure that all areas of the trim are covered, considering multi-angle and multi-position layouts to obtain comprehensive image data.

[0025] According to the size and shape of the panel, calculate the optimal layout coordinates of the fill light equipment and image acquisition equipment to ensure that these devices can fully cover the inspection area and do not interfere with each other. Record the calculated equipment layout coordinates in the predetermined layout plan to guide the installation of the fill light equipment and image acquisition equipment to achieve the best inspection effect.

[0026] When the target calcium silicate plaque is positioned in the first detection component, the fill light device combination is turned on, and the image of the target calcium silicate plaque is captured by the image capture device combination to obtain a plaque image set, wherein the plaque image set includes multiple images at different angles.

[0027] When the target calcium silicate plaque is transferred to the first detection component, a sensor or a positioning device is used to confirm that the plaque has accurately reached the predetermined position, thereby determining that the plaque is at the best shooting angle to avoid image deviation or distortion.

[0028] Start the combination of fill-in-light devices arranged in the first detection component. These fill-in-light devices, such as LED lights or other uniform light sources, provide sufficient and uniform lighting to eliminate shadows and uneven light on the surface of the panel, thereby improving image quality.

[0029] The image acquisition device combination is started. According to the predetermined layout plan, these devices are arranged in different positions to photograph the surface of the veneer at multiple angles. The multi-angle shooting ensures that every area of ​​the veneer is fully covered without missing any possible defects. Through these image acquisition devices, multiple images of the target calcium silicate veneer are captured. These images include images of different angles and areas of the veneer, forming a complete veneer image set.

[0030] The surface defect recognition model embedded in the surface defect detection component is used to perform surface defect detection on the set of decorative panel images to obtain surface defect feature information.

[0031] Start the surface defect detection component, which has an embedded surface defect recognition model that can recognize a variety of surface defects through training. Input the set of veneer images obtained from the first detection component into the surface defect detection component. The model extracts features from each input image and automatically extracts low-level to high-level features in the image through deep learning models such as convolutional neural networks. These features include information such as edges, textures, and shapes. Based on the extracted features, the model analyzes and classifies the image and identifies defective areas. Specifically, the model determines whether there are defects in the image and the type of defects by comparing the detected features with a pre-defined defect feature library, and locates the specific location of the defects on the veneer. Summarize the defect detection results in all images to generate complete surface defect feature information, which includes detailed data such as the type, location, size, and severity of the defect.

[0032] Furthermore, the surface defect recognition model is constructed, and the method includes:

[0033] A surface defect image set of calcium silicate veneer is obtained by data mining technology; feature analysis is performed on the surface defect image set to obtain a surface defect feature set; model training is performed on the surface defect feature set based on a convolutional neural network to obtain the surface defect recognition model.

[0034] The surface defect images of calcium silicate veneer accumulated in the past inspection process are collected and stored in the database. The images are annotated to identify the defect type, location and range in each image.

[0035] Use image processing algorithms such as edge detection, texture analysis, morphological operations, etc. to extract low-level features from images, including color histograms, texture features, shape features, etc. Use pre-trained convolutional neural networks to extract high-level features from images, which can automatically learn complex feature representations in images.

[0036] The extracted features are classified and the features of different types of defects are summarized, for example, the features of scratches, cracks, bubbles and other types of defects. Clustering algorithms such as K-means and DBSCAN are used to perform cluster analysis on the features, and similar features are clustered into groups to form a surface defect feature set.

[0037] The surface defect image collection is divided into a training set and a validation set, for example, 80% of the data is used for training and 20% of the data is used for validation. The training set is used to train the convolutional neural network model. During the process, hyperparameters such as learning rate and batch size are continuously adjusted to optimize model performance. The back propagation algorithm is used to update the model weights. During the training process, the validation set is used to monitor the performance of the model to prevent overfitting, and the generalization ability of the model is improved by adjusting the model structure and training strategy. Finally, when the model converges, a high-precision surface defect recognition model is obtained, providing technical support for defect detection of calcium silicate veneers.

[0038] Furthermore, the method of performing feature analysis on the surface defect image set to obtain a surface defect feature set includes:

[0039] Obtain a preset labeling scheme; perform labeling classification on the surface defect image set according to the preset labeling scheme to obtain a surface defect label set; perform cluster analysis on the surface defect image set according to the surface defect label set to obtain multiple surface defect image clusters; perform defect feature extraction on the surface defect images in the multiple surface defect image clusters to obtain multiple defect feature extraction results; obtain the surface defect feature set based on the multiple surface defect image clusters and the multiple defect feature extraction results.

[0040] Identify common surface defect types such as scratches, cracks, bubbles, pits, contamination, etc. and assign a label to each defect type to obtain preset labeling schemes.

[0041] Label the surface defect images, mark the defects in each image according to the preset labeling scheme, and assign corresponding labels. Record the labeling information of each image, including image number, defect type, defect location and range and other detailed data. Summarize all the labeled data to form a complete surface defect label set to provide basic data for subsequent analysis.

[0042] According to the surface defect label set, the image set is clustered and analyzed to identify image clusters with similar labels. Based on the clustering results, multiple surface defect image clusters are obtained. Each image cluster contains images with similar defect labels, reflecting surface defects of different types and patterns.

[0043] Select appropriate feature extraction methods, including traditional image processing techniques such as edge detection and morphological analysis, and define the defect features to be extracted, such as shape features, texture features, color features, etc.

[0044] The images in each image cluster are preprocessed, including noise removal, contrast enhancement, size standardization, etc. The selected feature extraction method is used to calculate the features of each image to generate a defect feature description. For example, the edge detection algorithm is used to extract the contour and shape features of the defect, and the texture analysis algorithm is used to extract the surface texture features of the defect. The extracted features are combined into a feature vector. Each image corresponds to a feature vector containing all defined features.

[0045] The feature data extracted from each image cluster are aggregated to form a comprehensive defect feature dataset, which contains the feature vectors of all images in all image clusters. According to the preset labeling scheme, the feature data are classified and labeled to ensure that the defect type corresponding to each feature vector is clear. This feature set provides basic data for subsequent machine learning model training, ensuring that the model can efficiently and accurately identify and classify various surface defects.

[0046] When the detection is completed, the conveying device is started to convey the target calcium silicate decorative plate to the second detection component, and the target calcium silicate decorative plate is ultrasonically detected by the ultrasonic detector to obtain an ultrasonic flaw detection curve.

[0047] After confirming that the test is completed, start the transmission device to transfer the target calcium silicate decorative plate from the first detection component to the second detection component, and start the ultrasonic detector to perform ultrasonic testing on the decorative plate. The ultrasonic detector includes a transmitting probe and a receiving probe, which are used to transmit ultrasonic waves and receive reflected wave signals. The ultrasonic detector transmits ultrasonic waves to the calcium silicate decorative plate. The ultrasonic waves propagate inside the decorative plate and are reflected when encountering material interfaces or defects. The receiving probe captures the reflected ultrasonic signals, which contain information about the internal structure of the decorative plate. The received ultrasonic signals are converted into electrical signals to form ultrasonic flaw detection curves. The ultrasonic flaw detection curves show the reflection intensity and time difference of ultrasonic waves when propagating inside the decorative plate, which can be used to analyze internal defects.

[0048] Perform waveform feature analysis on the ultrasonic flaw detection curve, and obtain internal defect feature information according to the analysis result; determine the online defect detection result of the target calcium silicate decorative plate based on the surface defect feature information and the internal defect feature information.

[0049] Use the pre-built defect feature waveform analyzer to analyze the ultrasonic flaw detection curve, extract the waveform features reflecting the internal defects, compare the extracted waveform features with the known defect feature library, and classify the possible defect types, such as cracks, voids, inclusions, etc. By analyzing the waveform features, the specific location and size of the defects inside the veneer can be determined. For example, the propagation speed of ultrasound in different materials is different, and the reflection time can help locate the location of the defect. The waveform feature information obtained by the analysis is summarized to form internal defect feature information, including detailed data such as defect type, location, size, and severity.

[0050] The surface defect feature information and internal defect feature information are integrated to form a complete veneer defect feature data set. This information provides a comprehensive defect situation on the surface and interior of the veneer, which can be used to evaluate the severity and impact of the defects of the veneer. For example, for serious defects, repair or scrapping measures can be taken; for minor defects, the veneer can continue to be used or undergo secondary processing.

[0051] Furthermore, the method of performing waveform characteristic analysis on the ultrasonic flaw detection curve and obtaining internal defect characteristic information according to the analysis result includes:

[0052] According to the ultrasonic flaw detection data records of calcium silicate veneer, a set of sample ultrasonic flaw detection curves is obtained; the defect characteristic waveforms of the sample ultrasonic flaw detection curve set are marked to obtain a set of sample defect characteristic waveforms; based on a neural network, a defect characteristic waveform analyzer is constructed by using the sample ultrasonic flaw detection curve set as input and the sample defect characteristic waveform set as output; based on the defect characteristic waveform analyzer, a waveform characteristic analysis is performed on the ultrasonic flaw detection curve to obtain a waveform characteristic analysis result; and the internal defect characteristic information is obtained according to the waveform characteristic analysis result.

[0053] Collect historical ultrasonic flaw detection data records, which are stored in the database, including flaw detection curves under different batches and conditions. Filter out high-quality flaw detection data, remove records with excessive noise or abnormalities, ensure the quality of sample data, format data from different sources, unify them into the same format and unit, and form a sample ultrasonic flaw detection curve collection.

[0054] Manually annotate the ultrasonic flaw detection curve, identify and mark the defect characteristic waveform, including the starting point, end point, and waveform characteristics (such as amplitude, frequency, etc.) of the defect. Classify the marked defect characteristic waveforms, mark each defect type, such as cracks, pores, inclusions, etc., and record the characteristics of each defect characteristic waveform, including reflection intensity, waveform shape, frequency characteristics, etc. Arrange the marked defect characteristic waveforms into a set to form a sample defect characteristic waveform set, each sample contains complete waveform characteristics and corresponding defect type annotations.

[0055] Construct a neural network model, divide the sample ultrasonic flaw detection curve set into a training set and a test set, for example, 80% for training and 20% for testing, use the sample ultrasonic flaw detection curve set as input, and the sample defect feature waveform set as output, train the neural network model, use the back propagation algorithm to update the model weights, minimize the loss function, and continuously adjust the learning rate, batch size and other hyperparameters during the training process to optimize the model performance. Use the test set to evaluate the performance of the model, calculate indicators such as accuracy, verify the generalization ability of the model, optimize the model according to the evaluation results, and finally obtain a defect feature waveform analyzer with an accuracy rate that meets the requirements, which is used to analyze the newly collected ultrasonic flaw detection data in real time.

[0056] The ultrasonic flaw detection curve data collected in real time from the production line is input into the defect characteristic waveform analyzer. The analyzer performs characteristic analysis on the input ultrasonic flaw detection curve based on the pre-trained neural network model. Specifically, the analyzer extracts key waveform features from the ultrasonic flaw detection curve, including the amplitude, frequency, time delay, waveform shape, etc. of the reflected signal. Based on the extracted waveform features, the analyzer identifies the possible defect type and locates the specific position of the defect in the material. The analyzer outputs the waveform characteristic analysis results, including detailed information of each detected defect, such as defect type, location, size, and characteristic waveform.

[0057] The internal defect feature information is extracted from the waveform feature analysis results, including the defect type, location, size and other related features. The defect feature information of each detection point is summarized to form a complete internal defect feature data set.

[0058] Furthermore, the method for obtaining the internal defect feature information according to the waveform feature analysis result includes:

[0059] The waveform feature analysis result includes a defect-free feature and a defect feature type; when the waveform feature analysis result is a defect-free feature, the internal defect feature information is directly output as empty; otherwise, the defect feature type is obtained, and the defect position and defect amplitude of the defect feature type are obtained; based on the defect feature type, the defect position, and the defect amplitude, the internal defect feature information is obtained.

[0060] The waveform feature analysis results include defect-free features and defect feature types. The defect-free features indicate that the flaw detection curve does not show any abnormal waveform and no obvious defects are detected inside the material. The defect feature types indicate the defect types detected in the flaw detection curve, such as cracks, pores, inclusions, etc.

[0061] If the waveform feature analysis result shows no defect feature, the internal defect feature information is empty, that is, the result of recording the inspection point in the database is no defect.

[0062] Otherwise, the defect feature type is extracted from the waveform feature analysis results, the specific type of the defect is clarified, and the specific location of the defect is determined, which can be calculated from the time difference and transmission path of the ultrasonic reflection. At the same time, the amplitude of the defect reflection signal is measured, and the amplitude is usually related to the severity of the defect.

[0063] Integrate information such as defect feature type, defect location and defect amplitude to form complete internal defect feature information.

[0064] Furthermore, the method further comprises:

[0065] The recognition effect of the surface defect recognition model is verified to obtain the defect recognition accuracy of the model; if the defect recognition accuracy does not reach the preset defect recognition accuracy, the defect recognition deviation is obtained based on the difference between the defect recognition accuracy and the preset defect recognition accuracy; based on the model optimization algorithm and the defect recognition deviation, the surface defect recognition model is iteratively updated to obtain a surface defect recognition optimization model.

[0066] A validation dataset is selected from data that has not been used for training, ensuring that the dataset contains various defect types and normal samples. Each sample in the validation dataset is accurately labeled as a benchmark for verifying the performance of the model. The validation dataset is input into the surface defect recognition model to obtain the model's prediction results. By comparing the model's prediction results with the actual labeling results, the model's defect recognition accuracy is calculated.

[0067] According to the actual application needs and quality control requirements, a preset defect recognition accuracy standard is set. If the actual recognition accuracy of the model is lower than the preset standard, the difference between the two is calculated as the defect recognition deviation.

[0068] Select appropriate model optimization algorithms, such as learning rate adjustment, data enhancement, etc. Based on the defect recognition deviation, gradually adjust the model parameters and retrain. After each update, use the validation set to evaluate the model performance and gradually improve the recognition accuracy. Further, by increasing the diversity of training data, such as rotation, scaling, cropping, etc., the generalization ability of the model is improved. Use the adjusted parameters and optimization algorithm to retrain the model. The process requires multiple iterations to gradually optimize the model performance. Use the validation data set to verify the optimized model again and calculate the new recognition accuracy. If the recognition accuracy of the optimized model meets or exceeds the preset standard, the model is determined to be the final surface defect recognition optimization model. By optimizing the model, the reliability of defect detection can be improved.

[0069] In summary, the online defect detection method for calcium silicate decorative panels provided in the embodiments of the present application has the following technical effects:

[0070] Combining image acquisition equipment and ultrasonic detectors, through multi-angle image acquisition and ultrasonic testing, comprehensive detection of surface and internal defects of calcium silicate decorative panels is achieved, and the accuracy of defect detection is improved; the calcium silicate decorative panels are sequentially transmitted to different detection components by conveying equipment, and the automated process design makes the detection process efficient and continuous, which meets the needs of mass production. Compared with manual inspection, automated inspection greatly reduces time and labor costs; through surface defect recognition models and ultrasonic feature analysis, the surface and internal defect feature information of the decorative panels can be fully obtained, and these two types of information can be comprehensively analyzed to provide comprehensive online defect detection results, thereby improving the reliability of quality control. In summary, the online defect detection method for calcium silicate decorative panels reduces the influence of human factors through automated and intelligent detection methods, improves the stability and consistency of the detection results, and improves the quality control level of calcium silicate decorative panels.

[0071] Based on the same inventive concept as the online defect detection method for calcium silicate decorative panels in the aforementioned embodiment, Figure 2 As shown, an embodiment of the present application provides an online defect detection system for calcium silicate decorative panels, the system comprising:

[0072] A transmission module 10, wherein the transmission module 10 is used to transmit the target calcium silicate decorative plate to the target detection area through a transmission device, wherein the target detection area sequentially deploys a first detection component and a second detection component along the transmission direction, wherein the first detection component includes an image acquisition device combination and a fill light device combination arranged according to a predetermined arrangement scheme, and the second detection component includes an ultrasonic detector; an image acquisition module 20, wherein the image acquisition module 20 is used to turn on the fill light device combination when the target calcium silicate decorative plate is positioned in the first detection component, and to perform image acquisition of the target calcium silicate decorative plate through the image acquisition device combination to obtain a decorative plate image set, wherein the decorative plate image set includes multiple images at different angles; a defect detection module 30, wherein the defect detection module 30 is used to detect the target calcium silicate decorative plate through a transmission device. A surface defect recognition model embedded in a surface defect detection component performs surface defect detection on the panel image set to obtain surface defect feature information; an ultrasonic detection module 40, wherein the ultrasonic detection module 40 is used to start the conveying device to convey the target calcium silicate panel to the second detection component after the detection is completed, and to perform ultrasonic detection of the target calcium silicate panel through the ultrasonic detector to obtain an ultrasonic flaw detection curve; a waveform feature analysis module 50, wherein the waveform feature analysis module 50 is used to perform waveform feature analysis on the ultrasonic flaw detection curve and obtain internal defect feature information based on the analysis result; a detection result acquisition module 60, wherein the detection result acquisition module 60 is used to determine the online defect detection result of the target calcium silicate panel based on the surface defect feature information and the internal defect feature information.

[0073] Furthermore, the system further includes a module for obtaining a predetermined layout plan to perform the following operation steps:

[0074] Obtain a defect detection task, including defect detection accuracy; collect and obtain the cross-sectional shape and cross-sectional dimensions of the target calcium silicate decorative panel; select a fill light device combination based on the cross-sectional shape and cross-sectional dimensions, select the image acquisition device combination in combination with the defect detection accuracy, and determine the device layout position coordinates; add the device layout position coordinates to the predetermined layout plan.

[0075] Furthermore, the system also includes a surface defect recognition model building module to perform the following operation steps:

[0076] A surface defect image set of calcium silicate veneer is obtained by data mining technology; feature analysis is performed on the surface defect image set to obtain a surface defect feature set; model training is performed on the surface defect feature set based on a convolutional neural network to obtain the surface defect recognition model.

[0077] Furthermore, the system further includes a surface defect feature set acquisition module to perform the following operation steps:

[0078] Obtain a preset labeling scheme; perform labeling classification on the surface defect image set according to the preset labeling scheme to obtain a surface defect label set; perform cluster analysis on the surface defect image set according to the surface defect label set to obtain multiple surface defect image clusters; perform defect feature extraction on the surface defect images in the multiple surface defect image clusters to obtain multiple defect feature extraction results; obtain the surface defect feature set based on the multiple surface defect image clusters and the multiple defect feature extraction results.

[0079] Furthermore, the system further includes an internal defect characteristic information acquisition module to perform the following operation steps:

[0080] According to the ultrasonic flaw detection data records of calcium silicate veneer, a set of sample ultrasonic flaw detection curves is obtained; the defect characteristic waveforms of the sample ultrasonic flaw detection curve set are marked to obtain a set of sample defect characteristic waveforms; based on a neural network, a defect characteristic waveform analyzer is constructed by using the sample ultrasonic flaw detection curve set as input and the sample defect characteristic waveform set as output; based on the defect characteristic waveform analyzer, a waveform characteristic analysis is performed on the ultrasonic flaw detection curve to obtain a waveform characteristic analysis result; and the internal defect characteristic information is obtained according to the waveform characteristic analysis result.

[0081] Furthermore, the system further includes an auxiliary feature information determination module to perform the following operation steps:

[0082] The waveform feature analysis result includes a defect-free feature and a defect feature type; when the waveform feature analysis result is a defect-free feature, the internal defect feature information is directly output as empty; otherwise, the defect feature type is obtained, and the defect position and defect amplitude of the defect feature type are obtained; based on the defect feature type, the defect position, and the defect amplitude, the internal defect feature information is obtained.

[0083] Furthermore, the system also includes a model optimization module to perform the following steps:

[0084] The recognition effect of the surface defect recognition model is verified to obtain the defect recognition accuracy of the model; if the defect recognition accuracy does not reach the preset defect recognition accuracy, the defect recognition deviation is obtained based on the difference between the defect recognition accuracy and the preset defect recognition accuracy; based on the model optimization algorithm and the defect recognition deviation, the surface defect recognition model is iteratively updated to obtain a surface defect recognition optimization model.

[0085] Through the aforementioned detailed description of the online defect detection method for calcium silicate decorative panels in this specification, those skilled in the art can clearly understand the online defect detection system for calcium silicate decorative panels in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0086] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An online defect detection method for calcium silicate decorative panels, characterized in that: The method comprises: The target calcium silicate decorative plate is transported to the target detection area by a transport device, wherein the target detection area sequentially deploys a first detection component and a second detection component along the transport direction, wherein the first detection component includes an image acquisition device combination and a fill light device combination arranged according to a predetermined arrangement scheme, and the second detection component includes an ultrasonic detector; When the target calcium silicate decorative plate is positioned in the first detection component, the fill light device combination is turned on, and the image of the target calcium silicate decorative plate is captured by the image acquisition device combination to obtain a decorative plate image set, wherein the decorative plate image set includes multiple images at different angles; Performing surface defect detection on the set of decorative panel images by using a surface defect recognition model embedded in the surface defect detection component to obtain surface defect feature information; When the detection is completed, the conveying device is started to convey the target calcium silicate decorative plate to the second detection component, and the target calcium silicate decorative plate is ultrasonically detected by the ultrasonic detector to obtain an ultrasonic flaw detection curve; Performing waveform characteristic analysis on the ultrasonic flaw detection curve, and obtaining internal defect characteristic information according to the analysis result; Determining an online defect detection result of the target calcium silicate decorative plate based on the surface defect feature information and the internal defect feature information; The method of performing waveform feature analysis on the ultrasonic flaw detection curve and obtaining internal defect feature information according to the analysis result includes: According to the ultrasonic flaw detection data records of calcium silicate decorative panels, a set of sample ultrasonic flaw detection curves is obtained; Marking the defect characteristic waveforms of the sample ultrasonic flaw detection curve set to obtain a sample defect characteristic waveform set; Based on a neural network, the sample ultrasonic flaw detection curve set is used as input, and the sample defect characteristic waveform set is used as output to construct a defect characteristic waveform analyzer; Based on the defect characteristic waveform analyzer, waveform characteristic analysis is performed on the ultrasonic flaw detection curve to obtain a waveform characteristic analysis result; Obtaining the internal defect characteristic information according to the waveform characteristic analysis result; The surface defect recognition model is constructed, and the method includes: The surface defect image collection of calcium silicate veneer is obtained by data mining technology; Performing feature analysis on the surface defect image set to obtain a surface defect feature set; Based on a convolutional neural network, model training is performed on the surface defect feature set to obtain the surface defect recognition model; The method of performing feature analysis on the surface defect image set to obtain a surface defect feature set includes: Get preset labeling scheme; Classify the surface defect image set by labeling according to the preset labeling scheme to obtain a surface defect label set; Performing cluster analysis on the surface defect image set according to the surface defect label set to obtain a plurality of surface defect image clusters; Extracting defect features from the surface defect images in the plurality of surface defect image clusters to obtain a plurality of defect feature extraction results; The surface defect feature set is obtained based on the multiple surface defect image clusters and the multiple defect feature extraction results.

2. The method for online defect detection of calcium silicate decorative panels according to claim 1, characterized in that: The method for obtaining the predetermined layout plan includes: Obtain defect detection tasks, including defect detection accuracy; Acquiring the cross-sectional shape and cross-sectional dimensions of the target calcium silicate decorative plate; Selecting a combination of supplementary light equipment according to the cross-sectional shape and cross-sectional size, selecting a combination of image acquisition equipment in combination with the defect detection accuracy, and determining the equipment layout position coordinates; The device layout location coordinates are added to the predetermined layout plan.

3. The method for online defect detection of calcium silicate decorative panels according to claim 1, characterized in that: The method for obtaining the internal defect feature information according to the waveform feature analysis result includes: The waveform feature analysis results include defect-free features and defect feature types; When the waveform feature analysis result is no defect feature, directly outputting the internal defect feature information as empty; Otherwise, obtaining the defect feature type, and obtaining the defect position and defect amplitude of the defect feature type; The internal defect feature information is obtained based on the defect feature type, the defect position, and the defect amplitude.

4. The method for online defect detection of calcium silicate decorative panels according to claim 1, characterized in that: The method also includes: Verifying the recognition effect of the surface defect recognition model to obtain the defect recognition accuracy of the model; If the defect recognition accuracy does not reach the preset defect recognition accuracy, obtaining a defect recognition deviation based on a difference between the defect recognition accuracy and the preset defect recognition accuracy; Based on the model optimization algorithm and the defect recognition deviation, the surface defect recognition model is iteratively updated to obtain a surface defect recognition optimization model.

5. An online defect detection system for calcium silicate decorative panels, characterized in that: For implementing the online defect detection method for calcium silicate decorative panels according to any one of claims 1 to 4, the system comprises: A transmission module, wherein the transmission module is used to transmit the target calcium silicate decorative plate to the target detection area through a transmission device, wherein the target detection area sequentially deploys a first detection component and a second detection component along the transmission direction, wherein the first detection component includes an image acquisition device combination and a fill light device combination arranged according to a predetermined arrangement scheme, and the second detection component includes an ultrasonic detector; An image acquisition module, wherein when the target calcium silicate decorative plate is positioned in the first detection component, the image acquisition module is used to start the fill light device combination, and to acquire an image of the target calcium silicate decorative plate through the image acquisition device combination to obtain a decorative plate image set, wherein the decorative plate image set includes multiple images at different angles; A defect detection module, the defect detection module is used to perform surface defect detection on the set of decorative plate images through a surface defect recognition model embedded in the surface defect detection component to obtain surface defect feature information; An ultrasonic testing module, wherein after the testing is completed, the ultrasonic testing module is used to start the conveying device to convey the target calcium silicate decorative plate to the second testing component, and to perform ultrasonic testing on the target calcium silicate decorative plate through the ultrasonic detector to obtain an ultrasonic flaw detection curve; A waveform feature analysis module, which is used to perform waveform feature analysis on the ultrasonic flaw detection curve and obtain internal defect feature information according to the analysis result; A detection result acquisition module is used to determine the online defect detection result of the target calcium silicate decorative panel based on the surface defect feature information and the internal defect feature information.

Images