Aluminum alloy die-casting production line filter screen detection method and system

Abstract

The invention discloses a filter screen detection method and system for an aluminum alloy die-casting production line, and the method comprises the steps: obtaining a filter screen placement image on the aluminum alloy die-casting production line, carrying out the segmentation and preprocessing of the filter screen placement image, and obtaining a to-be-detected filter screen placement image; the to-be-detected filter screen placement image is input into a filter screen detection model for detection, a filter screen detection result is output, the filter screen detection model is a YOLOv11 filter screen detection model, and the YOLOv11 filter screen detection model is obtained through training of a training data set with label data. According to the invention, the placement condition of the filter screen is monitored in real time, the cost and error of manual inspection are reduced, powerful support is provided for the quality control of the aluminum alloy die-casting production line, and the stable operation of the production line and the reliability of the product quality are guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of computer vision technology, and in particular, to a method and system for detecting filters in an aluminum alloy die-casting production line. Background Art

[0002] In an aluminum alloy die-casting production line, the filter screen is a core component to ensure die-casting quality and stable operation of the equipment. Whether it is correctly installed plays a decisive role. Once the filter screen is omitted, impurities in the aluminum alloy liquid will enter the die-casting mold, which will not only cause defects such as sand holes and air holes on the product surface, seriously affecting the product quality, but also may cause mold wear and equipment failures, greatly reducing production efficiency and increasing production costs.

[0003] Currently, the detection of the filter screen in the aluminum alloy die-casting production line mainly relies on manual inspection tours. This method consumes a large amount of manpower and time, and the efficiency is extremely low. Moreover, due to the limited attention of people, it is very easy to miss inspections. Some factories have tried to use traditional sensor detection technologies. However, this technology is easily affected by complex electromagnetic interference and mechanical vibrations in the production environment, resulting in inaccurate detection results and unable to meet the urgent needs of the aluminum alloy die-casting production line for efficient and accurate detection.

[0004] With the rapid development of artificial intelligence technology, machine vision detection technology has been increasingly widely used in the field of industrial production. Among them, object detection algorithms based on deep learning have shown powerful advantages. However, existing general object detection algorithms have problems such as inaccurate recognition of filter screen features and poor adaptability of detection models when facing specific scenarios such as filter screens in aluminum alloy die-casting production lines. Therefore, it is imperative to develop a detection method and system specifically for the correct placement of filter screens in aluminum alloy die-casting production lines. Summary of the Invention

[0005] To solve the above technical problems existing in the prior art, the present invention proposes a method and system for detecting filters in an aluminum alloy die-casting production line, which accurately detects whether the filter screen is correctly placed through image processing and deep learning technologies, thereby improving the automation level of the production line and product quality.

[0006] On the one hand, to achieve the above object, the present invention provides a method for detecting filters in an aluminum alloy die-casting production line, including:

[0007] Obtain an image of the filter screen placement on the aluminum alloy die-casting production line, and perform segmentation and preprocessing operations on the image of the filter screen placement to obtain an image of the filter screen placement to be detected;

[0008] Place the image of the filter screen to be detected into the filter screen detection model for detection, and output the filter screen detection result. Among them, the filter screen detection model is the YOLOv11 filter screen detection model, and the YOLOv11 filter screen detection model is trained by a training data set with labeled data.

[0009] Preferably, obtaining the image of the filter screen placement on the aluminum alloy die-casting production line includes:

[0010] Obtain the image of the filter screen placement on the aluminum alloy die-casting production line through an industrial camera. The image of the filter screen placement contains several filter screen placement points, and each filter screen placement point is located in a fixed area of the image.

[0011] Preferably, splitting the image of the filter screen placement includes:

[0012] Determine each filter screen placement area according to the preset coordinate information or the position information of the detected filter screen placement points, and split the image of the filter screen placement into several small images according to the coordinate range of each filter screen placement point. Each small image contains a single filter screen placement point and its surrounding area, and the pixels of each small image are the same.

[0013] Preferably, performing preprocessing operations on the image of the filter screen placement includes:

[0014] Convert the segmented image from color to grayscale, perform edge enhancement processing on the grayscale image using the Sobel operator, and then perform noise suppression processing using the Gaussian blur or median filtering method to obtain the processed image of the filter screen placement.

[0015] Preferably, obtaining the image of the filter screen placement to be detected includes:

[0016] Perform circular contour detection on the processed image of the filter screen placement, apply the Hough circle transform method to identify the circular structure in the image, and extract the region of interest corresponding to the filter screen based on the detected circular contour;

[0017] Perform contrast-limited adaptive histogram equalization processing on each region of interest, enhance the image contrast and amplify the noise, and use the Gaussian blur method to smooth the edges of the enhanced image to obtain the image of the filter screen placement to be detected.

[0018] Preferably, training the filter screen detection model by a training data set with labeled data includes:

[0019] Extract data features through the YOLOv11 model and retain the generated optimal parameters to achieve accurate detection of the filter screen state. After the model training is completed, it can automatically detect the state of the filter screen on the aluminum alloy die-casting production line and output the detection result;

[0020] Among them, the training data set is a filter screen image on an aluminum alloy die-casting production line collected by an industrial lens, and includes examples of whether the filter screen of the die-casting model is correctly placed.

[0021] Preferably, the method further includes determining the filter screen detection result, including:

[0022] If all the filter screens are correctly identified, the detection passes and a confirmation signal is sent;

[0023] If any one of the filter screens is not correctly identified, an alarm is triggered to remind, the die-casting machine stops pouring and waits for the worker to re-place the filter screen, and feedback information is provided through the controller to indicate the specific position of the filter screen.

[0024] On the other hand, to achieve the above object, the present invention also provides a filter screen detection system for an aluminum alloy die-casting production line, including:

[0025] An image acquisition module: used to obtain the filter screen placement image on the aluminum alloy die-casting production line, and perform segmentation and preprocessing operations on the filter screen placement image to obtain the filter screen placement image to be detected;

[0026] A filter screen detection model construction module, used to construct a filter screen detection model, wherein the filter screen detection model is trained by a training data set with labeled data;

[0027] A filter screen detection module, used to input the filter screen placement image to be detected into a pre-trained YOLOv11 model and set parameters for prediction, and output a detection result.

[0028] A computer device includes a memory, a processor, and a computer program stored on the memory, and the processor executes the computer program to implement the steps of the filter screen detection method for the aluminum alloy die-casting production line.

[0029] A computer-readable storage medium stores a computer program / instructions, and when the computer program / instructions are executed by a processor, the steps of the filter screen detection method for the aluminum alloy die-casting production line are implemented.

[0030] Compared with the prior art, the present invention has the following advantages and technical effects:

[0031] (1) During the filter screen detection process, the present invention significantly improves the image quality through image preprocessing techniques (including grayscale conversion, enhancing the edges of the grayscale image with the Sobel operator, noise suppression, circular contour detection, region of interest extraction, contrast enhancement, and edge blurring), and enhances the ability to extract filter screen features; combined with the trained YOLOv11 filter screen detection model, it can efficiently and accurately judge whether the filter screen is correctly placed;

[0032] (2) The system of the present invention processes by dividing the image into four regions respectively, ensuring that each filter placement point is independently detected, thereby improving the comprehensiveness and accuracy of detection. If a filter is detected to be missing, the system will immediately trigger an alarm to remind the worker to place it again, effectively avoiding production quality problems caused by missing filters. In addition, the system realizes real-time monitoring of the filter placement situation, reduces the cost and error of manual inspection, provides strong support for the quality control of the aluminum alloy die-casting production line, and ensures the stable operation of the production line and the reliability of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments and descriptions thereof of this application are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0034] Figure 1 is a flowchart of a method for detecting filters on an aluminum alloy die-casting production line according to an embodiment of the present invention;

[0035] Figure 2 is a result diagram after image preprocessing and filter contour positioning and processing according to an embodiment of the present invention;

[0036] Figure 3 is a detection result diagram presented by the controller after the original image of an embodiment of the present invention is fed into the program. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The following will refer to the drawings and combine with the embodiments to detail this application.

[0038] It should be noted that the steps shown in the flowchart of the drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0039] Such as Figures 1 - 3 , the present invention proposes a method for detecting filters on an aluminum alloy die-casting production line, including:

[0040] Obtain the filter placement image on the aluminum alloy die-casting production line, and perform segmentation and preprocessing operations on the filter placement image to obtain the filter placement image to be detected;

[0041] Place the image of the filter screen to be detected into the filter screen detection model for detection, and output the filter screen detection result. Among them, the filter screen detection model is the YOLOv11 filter screen detection model, and the YOLOv11 filter screen detection model is trained by a training data set with labeled data. The training data set is historical scenario data including the normal installation and missing placement of filter screens.

[0042] During the filter screen detection process, in this embodiment, the image quality is significantly improved through image preprocessing techniques (including grayscale conversion, enhancing the edges of the grayscale image with the Sobel operator, noise suppression, circular contour detection, region of interest extraction, contrast enhancement, and edge blurring), and the ability to extract filter screen features is enhanced. Combined with the trained YOLOv11 filter screen detection model, it can efficiently and accurately determine whether the filter screen is correctly placed. By dividing the image into four regions for separate processing, it is ensured that each filter screen placement point is independently detected, thereby improving the comprehensiveness and accuracy of the detection. If a missing placement of the filter screen is detected, the system will immediately trigger an alarm to remind the worker to place it again, effectively avoiding production quality problems caused by the missing placement of the filter screen.

[0043] Furthermore, obtaining the filter screen placement image on the aluminum alloy die-casting production line includes:

[0044] Obtain the filter screen placement image on the aluminum alloy die-casting production line through an industrial camera. The filter screen placement image contains several filter screen placement points, and each filter screen placement point is located in a fixed area of the image.

[0045] Specifically, obtain the filter screen placement image on the production line through an industrial camera. The original image contains four filter screen placement points.

[0046] Furthermore, dividing the filter screen placement image includes:

[0047] Determine each filter screen placement area according to the preset coordinate information or the position information of the detected filter screen placement points. Divide the filter screen placement image into several small images according to the coordinate range of each filter screen placement point. Each small image contains a single filter screen placement point and its surrounding area, and the pixels of each small image are the same.

[0048] Specifically, in this embodiment, the original image is divided into four sub-images, and the size of each sub-image is the same. The division method is that the obtained original image contains four filter screen placement points. Determine each filter screen placement area according to the preset coordinate information or the position information of the detected filter screen placement points. Accurately divide the original image into four small images according to the coordinate range of each filter screen placement point. Each small image accurately contains a single filter screen placement point and its surrounding area, ensuring that subsequent steps can analyze each filter screen independently.

[0049] Using the method of horizontal and vertical bisection, first, calculate the width W = Xmax - Xmin and height H = Ymax - Ymin of the filter screen placement image. Then, divide the image into two parts horizontally, each part with a width of Divide the image into two parts vertically, each part with a height of By horizontal and vertical bisection, four regions can be obtained, corresponding to four small images respectively.

[0050] Furthermore, perform preprocessing operations on the filter screen placement image, including:

[0051] Convert the segmented image from color to grayscale, enhance the edges of the grayscale image using the Sobel operator, and perform noise suppression processing using Gaussian blur or median filtering methods to obtain the processed filter screen placement image.

[0052] Specifically, perform grayscale conversion, edge enhancement of the grayscale image using the Sobel operator, and noise suppression processing on each small image. First, convert each small image from color to grayscale. Then use the Sobel operator to enhance the edges of the grayscale image. To remove the existing noise, use Gaussian blur or median filtering methods for noise suppression processing. These processes aim to improve the image quality and thus enhance the accuracy of subsequent image analysis.

[0053] Among them, the grayscale conversion process is:

[0054] Convert each sub - image from an RGB color image to a grayscale image:

[0055] Gray = 0.299×R + 0.587×0.114×B;

[0056] The edge enhancement of the grayscale image using the Sobel operator is:

[0057] Specifically, the Sobel operator consists of two 3x3 matrices, which are used for edge detection in the horizontal direction (G x ) and vertical direction (G y ) respectively. Then, the gradient magnitude of each pixel point can be calculated by the following formula:

[0058]

[0059] The noise suppression processing is:

[0060] Use Gaussian filtering to suppress noise in the grayscale image. The size of the Gaussian kernel is 9×9, and the standard deviation σ = 2. The noise in the processed image is significantly reduced.

[0061] Furthermore, obtain the filter screen placement image to be detected, including:

[0062] Perform circular contour detection on the processed filter screen placement image, apply the Hough circle transform method to identify the circular structure in the image, and extract the region of interest (ROI) corresponding to the filter screen based on the detected circular contour;

[0063] Perform Contrast Limited Adaptive Histogram Equalization (CLAHE) processing on each of the said regions of interest, enhance the contrast of the image, and use the Gaussian blur method to smooth the edges of the image after the enhancement processing to obtain the filter screen placement image to be detected.

[0064] In this embodiment, the parameter settings for the CLAHE processing are: the contrast limit threshold is 40, and the grid size is 8×8. This effectively enhances the contrast between the filter screen and the background, making the filter screen area more obvious.

[0065] Specifically, the circular contour detection is as follows:

[0066] Use the Hough Circle Transform to detect the circular contour of the filter screen.

[0067] In this embodiment, the parameter settings are: the minimum circle radius: 95 pixels; the maximum circle radius: 130 pixels; the minimum distance between the centers of the circles: 50 pixels. The detected circular contour is used to determine the center position and radius of the filter screen.

[0068] The extraction of the region of interest (ROI) includes:

[0069] According to the detected circular contour, extract the region of interest (ROI) of the filter screen. The size of the ROI region is 290×250 pixels, with the center of the filter screen as the center.

[0070] Use CLAHE histogram equalization to enhance the contrast between the filter screen and the background. After enhancement, the gray-scale distribution of the image is more uniform, and the filter screen area is more obvious.

[0071] Furthermore, input the filter screen placement image to be detected into the filter screen detection model for detection, and output the filter screen detection result, including:

[0072] Specifically, the training data set is specifically a set of filter screen images collected on an aluminum alloy die-casting production line using an industrial lens, totaling 7750 images. These images contain examples of whether the filter screen in the die-casting model is correctly placed. Among them, 4000 images are labeled as "YES" (normal), and 3750 images are labeled as "NO" (abnormal). The data is manually annotated and used to train the filter screen detection model to identify the state of the filter screen.

[0073] Data Division: The collected images are divided into a training set, a validation set, and a test set to facilitate model training, tuning, and performance evaluation. The specific division ratio is 5425 images for the training set, 1162 images for the validation set, and 1162 images for the test set.

[0074] Model Training: Using the above - divided dataset, training is carried out through the YOLOv11 model. During the training process, the model has extracted the features of these data and retained the best - generated parameters to achieve accurate detection of the filter screen status. After the model training is completed, it can automatically detect the status of the filter screen on the aluminum alloy die - casting production line and output the detection results.

[0075] Model Input:

[0076] The processed ROI region (290×250 pixels) is input into the trained YOLOv11 filter screen detection model. The model input size is 224×224, so the ROI region is resized by bilinear interpolation.

[0077] Prediction Results:

[0078] The local feature map is input into the main part of the model, and the prediction result of whether the filter screen is missing is output. The prediction result is a binary classification (NO: filter screen not installed, YES: filter screen installed).

[0079] Furthermore, the method further includes determining the filter screen detection results, specifically:

[0080] If all filter screens are correctly identified, the detection passes, and a confirmation signal is sent;

[0081] If any one filter screen is not correctly identified, an alarm is triggered to remind, the die - casting machine stops pouring and waits for the worker to re - place the filter screen, and feedback information is provided through the controller to indicate the specific position of the filter screen.

[0082] Specifically, the prediction results of the four sub - graphs are comprehensively determined. If all four sub - graphs are predicted as YES (filter screen installed), it is determined to be qualified. If any one sub - graph is predicted as NO (filter screen not installed), it is determined to be unqualified.

[0083] If it is determined to be unqualified, an alarm is triggered to remind the worker to re - place the filter screen. The alarm reminds in a dual way of sound prompt and light flashing. The die - casting machine stops pouring and waits for the worker to re - place the filter screen for re - detection. Feedback information is also provided through the controller to indicate which specific position of the filter screen needs to be adjusted or re - placed.

[0084] This embodiment also provides a filter screen detection system for an aluminum alloy die - casting production line, including:

[0085] Image acquisition module: It is used to acquire the image of the filter screen placement on the aluminum alloy die-casting production line, and perform segmentation and preprocessing operations on the image of the filter screen placement to obtain the image of the filter screen placement to be detected;

[0086] Filter screen detection model construction module: It is used to construct a filter screen detection model, where the filter screen detection model is trained by a training data set with labeled data;

[0087] Filter screen detection module: It is used to input the image of the filter screen placement to be detected into the pre-trained YOLOv11 model and the set parameters for prediction, and output the detection result.

[0088] Through steps such as grayscale conversion, noise suppression, circular contour detection, and contrast enhancement, the present invention significantly improves the accuracy of filter screen detection. Using the pre-trained YOLOv11 model further improves the detection performance of the model; image segmentation and ROI extraction reduce the computational amount of the model and improve the detection efficiency; Gaussian blur and contrast enhancement reduce background interference and lower the training cost of the model.

[0089] The entire detection process of the present invention can be completed within 1 second, meeting the real-time detection requirements of the production line; the alarm mechanism timely reminds workers to avoid production accidents caused by missing the placement of the filter screen; through steps such as image segmentation, preprocessing, contour positioning, model analysis, and result determination, the automatic detection of the filter screen on the aluminum alloy die-casting production line is realized, with the advantages of high precision, low cost, and real-time performance.

[0090] This embodiment also provides a computer device, including a memory, a processor, and a computer program stored on the memory. The processor executes the computer program to implement the steps of the method for detecting the filter screen on the aluminum alloy die-casting production line.

[0091] This embodiment also provides a computer-readable storage medium, on which a computer program / instructions are stored. When the computer program / instructions are executed by a processor, the steps of the method for detecting the filter screen on the aluminum alloy die-casting production line are implemented.

[0092] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present application should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for detecting a filter screen in an aluminum alloy die-casting production line, characterized in that Including: Obtain the filter screen placement image on the aluminum alloy die-casting production line, and perform segmentation and preprocessing operations on the filter screen placement image to obtain the filter screen placement image to be detected; Input the filter screen placement image to be detected into the filter screen detection model for detection, and output the filter screen detection result. Among them, the filter screen detection model is the YOLOv11 filter screen detection model, and the YOLOv11 filter screen detection model is trained by a training data set with labeled data.

2. The method for detecting the filter screen of an aluminum alloy die-casting production line according to claim 1, wherein Obtaining the filter screen placement image on the aluminum alloy die-casting production line includes: Obtain the filter screen placement image on the aluminum alloy die-casting production line through an industrial camera. The filter screen placement image contains several filter screen placement points, and each filter screen placement point is located in a fixed area of the image.

3. The filter screen detection method for the aluminum alloy die-casting production line according to claim 1, characterized in that Segmenting the filter screen placement image includes: Determine each filter screen placement area according to the preset coordinate information or the position information of the detected filter screen placement points, and segment the filter screen placement image into several small images according to the coordinate range of each filter screen placement point. Each small image contains a single filter screen placement point and its surrounding area, and the pixels of each small image are the same.

4. The method for detecting the filter screen of the aluminum alloy die-casting production line according to claim 3, wherein, Performing preprocessing operations on the filter screen placement image includes: Convert the segmented image from color to grayscale, perform edge enhancement processing on the grayscale image using the Sobel operator, and then perform noise suppression processing using the Gaussian blur or median filtering method to obtain the processed filter screen placement image.

5. The filter screen detection method for an aluminum alloy die-casting production line according to claim 4, characterized in that, Obtaining the filter screen placement image to be detected includes: Perform circular contour detection on the processed filter screen placement image, apply the Hough circle transform method to identify the circular structure in the image, and extract the region of interest corresponding to the filter screen based on the detected circular contour; Perform contrast-limited adaptive histogram equalization processing on each region of interest, enhance the image contrast and amplify the noise, and use the Gaussian blur method to smooth the edges of the enhanced image to obtain the filter screen placement image to be detected.

6. The method for detecting the filter screen of an aluminum alloy die-casting production line according to claim 1, wherein, Training the filter screen detection model by a training data set with labeled data includes: Extract data features through the YOLOv11 model and retain the generated optimal parameters to achieve accurate detection of the filter screen state. After the model training is completed, it can automatically detect the state of the filter screen on the aluminum alloy die-casting production line and output the detection result; Among them, the training data set is the filter screen image on the aluminum alloy die-casting production line collected by an industrial lens, and contains examples of whether the filter screen of the die-casting model is correctly placed.

7. The method for detecting the filter screen of the aluminum alloy die-casting production line according to claim 1, wherein The method further includes determining the filter screen detection result, including: If all filter screens are correctly identified, the detection passes and a confirmation signal is sent; If any one of the filter screens is not correctly identified, an alarm is triggered to remind, the die-casting machine stops pouring and waits for the worker to re-place the filter screen, and feedback information is provided through the controller to indicate the specific position of the filter screen.

8. A filter screen detection system for an aluminum alloy die-casting production line, characterized in that, Including: Image acquisition module: used to obtain the filter screen placement image on the aluminum alloy die-casting production line, and perform segmentation and preprocessing operations on the filter screen placement image to obtain the filter screen placement image to be detected; A filter screen detection model construction module, which is used to construct a filter screen detection model, and the filter screen detection model is obtained by training with a training data set with labeled data; A filter screen detection module, which is used to input the image of the filter screen to be detected into the pre-trained YOLOv11 model and the set parameters for prediction, and output the detection result.

9. A computer device, comprising a memory, a processor, and a computer program stored on the memory, characterized in that, The processor executes the computer program to implement the steps of the filter screen detection method for the aluminum alloy die-casting production line according to any one of claims 1-7.

10. A computer-readable storage medium having computer programs / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, the steps of the filter screen detection method for the aluminum alloy die-casting production line according to any one of claims 1-7 are implemented.

Images