Real-time detection device and detection method for vacuum tank lining

Abstract

The invention belongs to the technical field of metallurgy, and particularly relates to a real-time detection device and method for a vacuum tank lining. A real-time detection device for a vacuum tank lining comprises an infrared imaging mechanism, an image collector, a data processor and a signal output device which are sequentially connected, the infrared imaging mechanism comprises a plurality of infrared detectors arranged in the circumferential direction of a vacuum tank, the infrared detectors are evenly distributed and connected with the image collector, and the image collector is connected with the data processor. The infrared detector is used for acquiring a thermal radiation image of the surface of the lining of the vacuum tank, the image collector is used for collecting image data generated by the infrared detector, and the data processor is used for analyzing and processing the image collected by the image collector, judging the state of the lining and giving feedback. And the signal output device outputs a processing result of the data processor. The method can accurately judge the use states of abrasion, aging, cracks and the like of the vacuum tank lining, so as to determine whether the lining needs to be replaced or not.

Description

Technical Field

[0001] The invention belongs to the technical field of metallurgy, and in particular relates to a real-time detection device and a detection method for a vacuum tank lining. Background Art

[0002] Vacuum tanks are widely used in the metallurgical industry to hold molten steel at around 1600°C. In a high-temperature vacuum environment, the integrity of the lining material plays a decisive role in whether the lining can hold the molten steel. As a consumable material, as the use time increases, the lining will gradually wear, age, or even crack, affecting the sealing and safety of the equipment, and requires timely repair and replacement.

[0003] Existing detection methods are mostly manual inspection or destructive detection, which have the problems of low efficiency, poor accuracy and long detection cycle. Therefore, there is an urgent need for a device and method that can detect the state of the vacuum tank lining in real time and non-contact. Summary of the invention

[0004] In view of the above problems, the purpose of the present invention is to provide a real-time detection device and method for the vacuum tank lining, which can accurately judge the wear, aging, cracks and other usage conditions of the vacuum tank lining, thereby determining whether the lining needs to be replaced.

[0005] The technical solution of the present invention is: a real-time detection device for the lining of a vacuum groove, comprising an infrared imaging mechanism, an image collector, a data processor and a signal output device connected in sequence, the infrared imaging mechanism comprising a plurality of infrared detectors arranged around the vacuum groove, the plurality of infrared detectors being evenly distributed, the infrared detectors being connected to the image collector, the infrared detectors being used to obtain a thermal radiation image of the surface of the lining of the vacuum groove, the image collector being used to collect image data generated by the infrared detector, the data processor being used to analyze and process the image collected by the image collector, determine the state of the lining and give feedback, and the signal output device outputs the processing result of the data processor.

[0006] There are four infrared detectors, including a first infrared detector, a second infrared detector, a third infrared detector and a fourth infrared detector. The probes of the first infrared detector, the second infrared detector, the third infrared detector and the fourth infrared detector are arranged at 90 degrees in pairs.

[0007] A real-time detection method for a vacuum tank lining, using the real-time detection device for a vacuum tank lining as described above, comprises the following steps: S1: According to the design requirements, multiple infrared detectors are installed on the outside of the vacuum tank to ensure that the surface area of ​​the vacuum tank lining can be fully covered; S2: During the operation of the vacuum tank, the infrared detector continuously scans the outside of the vacuum tank to generate thermal imaging images. The image collector collects the image data generated by the infrared detector. The data processor analyzes and processes the images collected by the image collector. The specific process is as follows: S21: preprocessing the acquired infrared image, including noise filtering, image enhancement and temperature calibration; S22: Analyze the infrared image through image processing algorithms to detect wear, cracks and aging areas of the lining. The image processing algorithms used include edge detection, pattern recognition, and thermal distribution anomaly detection; S23: Generate a lining status report based on the processing result of step S22, and mark the damaged or abnormal area; S3: The signal output device outputs the processing result of the data processor 7, and outputs a warning signal in advance according to the processing result to remind the user to replace the lining in time.

[0008] In step S2, the specific process of noise filtering is: Step 1: Read the infrared image and convert it into a grayscale image; Step 2: Select a Gaussian filter to smooth the image; Step 3: For areas with high noise, further median filtering or bilateral filtering is performed.

[0009] In step S2, the specific process of image enhancement is as follows: Step 1: Convert the image to grayscale; Step 2: Use histogram equalization or CLAHE to enhance the grayscale image; Step 3: Use the enhanced image for subsequent processing.

[0010] In step S2, the specific process of temperature calibration is as follows: Step 1: Obtain the temperature calibration data of the infrared detector and establish the corresponding relationship between pixel value and temperature value; Step 2: Convert the pixel value of the infrared image into a temperature value image; Step 3: Use the temperature value image for subsequent analysis and detection.

[0011] In step S2, the specific process of the edge detection algorithm is: Step 1: Preprocessing: Use Gaussian filtering to smooth the image, grayscale the infrared image, and perform noise reduction to eliminate the influence of noise; Step 2: Gradient calculation: Use the edge detection algorithm Sobel operator to derive the image and obtain the gradient magnitude and direction of each pixel; Step 3: Non-maximum suppression: suppress non-maximum values ​​along the gradient direction to ensure that the edges appear as thin lines; Step 4: Dual threshold detection: set high and low thresholds to classify pixels into strong edge, weak edge, and non-edge; Step 5: Edge tracking: Get a complete edge image by connecting strong edges and weak edges; Step 6: Edge merging and filtering: Merge and filter the noise edges and output the final edge map.

[0012] In step S2, the specific process of the pattern recognition algorithm is as follows: Step 1: Data collection: Collect infrared image data containing normal and damaged areas and label them; Step 2: Data preprocessing: standardize and enhance the image, including rotation, cropping, and brightness adjustment; Step 3: Model training: Use the labeled data set to train the CNN model. The model structure includes convolutional layer, pooling layer, and fully connected layer. Step 4: Damage area detection: Input the detection image into the trained model to predict the damage type and location; Step 5: Output results: Mark the damaged area on the original image.

[0013] In step S2, the specific process of the heat distribution anomaly detection algorithm is as follows: Step 1: Temperature mapping: Convert the infrared image into a temperature distribution map so that the temperature of each area can be observed; Step 2: Threshold segmentation: Based on the set temperature threshold, the temperature abnormality area is segmented, and the upper and lower thresholds are set to identify areas with too high or too low temperatures; Step 3: Regional analysis: Perform connected region analysis on the segmented abnormal regions to determine their area and shape and filter out small area noise; Step 4: Mark abnormal areas: Mark abnormal areas on the image for further detection and analysis.

[0014] The technical effects of the present invention are: 1. The present invention detects infrared radiation on the surface of an object through an infrared detector to generate a corresponding thermal imaging image. Since the thermal distribution of the vacuum tank lining will be abnormal when it is worn, aged or cracked, infrared imaging can intuitively present these abnormal areas. The temperature distribution in the normal area is relatively uniform, while the damaged area will experience temperature fluctuations or abnormal increase / decrease, and the detection result is accurate; 2. The present invention processes the infrared image of the vacuum tank through three algorithms: edge detection, pattern recognition, and thermal distribution anomaly detection, to generate a temperature distribution map of the lining surface, and then accurately determine the location of the vacuum tank lining where wear, aging or cracks occur.

[0015] The following is a further description with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a structural schematic diagram of a real-time detection device for a vacuum tank liner according to an embodiment of the present invention.

[0017] Figure numerals: 1-first infrared detector, 2-vacuum tank, 3-second infrared detector, 4-third infrared detector, 5-fourth infrared detector, 6-image collector, 7-data processor, 8-signal outputter. DETAILED DESCRIPTION Example 1

[0018] like Figure 1 As shown, a real-time detection device for the lining of a vacuum tank comprises an infrared imaging mechanism, an image collector 6, a data processor 7 and a signal output device 8 which are connected in sequence, wherein the infrared imaging mechanism comprises a plurality of infrared detectors arranged in the circumference of the vacuum tank 2, the plurality of infrared detectors are evenly distributed, the infrared detectors are connected to the image collector 6, the infrared detectors are used to obtain a thermal radiation image of the surface of the lining of the vacuum tank, the image collector 6 is used to collect image data generated by the infrared detectors, the data processor 7 is used to analyze and process the image collected by the image collector 6, determine the state of the lining and give feedback, and the signal output device 8 outputs the processing result of the data processor 7.

[0019] There are four infrared detectors, including a first infrared detector 1, a second infrared detector 3, a third infrared detector 4 and a fourth infrared detector 5. The probes of the first infrared detector 1, the second infrared detector 3, the third infrared detector 4 and the fourth infrared detector 5 are arranged at 90 degrees in pairs.

[0020] The present invention detects infrared radiation on the surface of an object through an infrared detector and generates a corresponding thermal image. Since the thermal distribution of the vacuum tank lining will be abnormal when it is worn, aged or cracked, infrared imaging can intuitively present these abnormal areas. The temperature distribution in the normal area is relatively uniform, while the damaged area will experience temperature fluctuations or abnormal increase / decrease, and the detection result is accurate. Example 2

[0021] A real-time detection method for a vacuum tank lining, using the real-time detection device for a vacuum tank lining as described above, comprises the following steps: S1: According to the design requirements, multiple infrared detectors are installed on the outside of the vacuum tank 2 to ensure that the surface area of ​​the lining of the vacuum tank 2 can be fully covered; S2: During the operation of the vacuum tank 2, the infrared detector continuously scans the outside of the vacuum tank 2 to generate a thermal imaging image. The image collector 6 collects the image data generated by the infrared detector. The data processor 7 analyzes and processes the image collected by the image collector 6. The specific process is as follows: S21: Preprocessing the acquired infrared images, including noise filtering, image enhancement and temperature calibration; through such a preprocessing process, the clarity and accuracy of the infrared images can be significantly improved, providing more reliable input for subsequent image recognition algorithms, and improving the detection system's recognition accuracy for lining wear, cracks, aging and other problems; S22: Analyze the infrared image through image processing algorithms to detect wear, cracks and aging areas of the lining. The image processing algorithms used include edge detection, pattern recognition, and thermal distribution anomaly detection; S23: Generate a lining status report based on the processing result of step S22, and mark the damaged or abnormal area; S3: The signal output device 8 outputs the processing result of the data processor 7, and outputs a warning signal in advance according to the processing result to remind the user to replace the lining in time.

[0022] The present invention processes the infrared image of the vacuum tank through three algorithms of edge detection, pattern recognition and thermal distribution anomaly detection to generate a temperature distribution map of the lining surface, thereby accurately determining the locations where the vacuum tank lining is worn, aged or cracked.

[0023] In step S2, the specific process of noise filtering is: Step 1: Read the infrared image and convert it into a grayscale image; Step 2: Select a Gaussian filter to smooth the image; Step 3: For areas with high noise, further median filtering or bilateral filtering is performed.

[0024] In step S2, the specific process of image enhancement is as follows: Step 1: Convert the image to grayscale; Step 2: Use histogram equalization or CLAHE to enhance the grayscale image; Step 3: Use the enhanced image for subsequent processing.

[0025] In step S2, the specific process of temperature calibration is as follows: Step 1: Obtain the temperature calibration data of the infrared detector and establish the corresponding relationship between pixel value and temperature value; Step 2: Convert the pixel value of the infrared image into a temperature value image; Step 3: Use the temperature value image for subsequent analysis and detection.

[0026] In step S2, the specific process of the edge detection algorithm is: Step 1: Preprocessing: Use Gaussian filtering to smooth the image, grayscale the infrared image, and perform noise reduction to eliminate the influence of noise; Step 2: Gradient calculation: Use the edge detection algorithm Sobel operator to derive the image and obtain the gradient magnitude and direction of each pixel; Step 3: Non-maximum suppression: suppress non-maximum values ​​along the gradient direction to ensure that the edges appear as thin lines; Step 4: Dual threshold detection: set high and low thresholds to classify pixels into strong edge, weak edge, and non-edge; Step 5: Edge tracking: Get a complete edge image by connecting strong edges and weak edges; Step 6: Edge merging and filtering: Merge and filter the noise edges and output the final edge map.

[0027] In step S2, the specific process of the pattern recognition algorithm is as follows: Step 1: Data collection: Collect infrared image data containing normal and damaged areas and label them; Step 2: Data preprocessing: standardize and enhance the image, including rotation, cropping, and brightness adjustment; Step 3: Model training: Use the labeled data set to train the CNN model. The model structure includes convolutional layer, pooling layer, and fully connected layer. Step 4: Damage area detection: Input the detection image into the trained model to predict the damage type and location; Step 5: Output results: Mark the damaged area on the original image.

[0028] In step S2, the specific process of the heat distribution anomaly detection algorithm is as follows: Step 1: Temperature mapping: Convert the infrared image into a temperature distribution map so that the temperature of each area can be observed; Step 2: Threshold segmentation: Based on the set temperature threshold, the temperature abnormality area is segmented, and the upper and lower thresholds are set to identify areas with too high or too low temperatures; Step 3: Regional analysis: Perform connected region analysis on the segmented abnormal regions to determine their area and shape and filter out small area noise; Step 4: Mark abnormal areas: Mark abnormal areas on the image for further detection and analysis.

[0029] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A real-time detection device for a vacuum tank lining, characterized in that: The invention comprises an infrared imaging mechanism, an image collector (6), a data processor (7) and a signal output device (8) which are connected in sequence. The infrared imaging mechanism comprises a plurality of infrared detectors arranged in the circumference of the vacuum tank (2). The plurality of infrared detectors are evenly distributed. The infrared detectors are connected to the image collector (6). The infrared detectors are used to obtain a thermal radiation image of the surface of the inner lining of the vacuum tank. The image collector (6) is used to collect image data generated by the infrared detectors. The data processor (7) is used to analyze and process the image collected by the image collector (6), determine the state of the inner lining and give feedback. The signal output device (8) outputs the processing result of the data processor (7).

2. A real-time detection device for a vacuum tank lining according to claim 1, characterized in that: There are four infrared detectors in total, including a first infrared detector (1), a second infrared detector (3), a third infrared detector (4) and a fourth infrared detector (5), and the probes of the first infrared detector (1), the second infrared detector (3), the third infrared detector (4) and the fourth infrared detector (5) are arranged at 90 degrees in pairs.

3. A real-time detection method for a vacuum tank lining, using a real-time detection device for a vacuum tank lining as claimed in claim 1, characterized in that: The following steps are involved: S1: According to the design requirements, multiple infrared detectors are installed on the outside of the vacuum tank (2) to ensure that the surface area of ​​the inner lining of the vacuum tank (2) can be fully covered; S2: During the operation of the vacuum tank (2), the infrared detector continuously scans the outside of the vacuum tank (2) to generate a thermal imaging image. The image collector (6) collects the image data generated by the infrared detector. The data processor (7) analyzes and processes the image collected by the image collector (6). The specific process is as follows: S21: preprocessing the acquired infrared image, including noise filtering, image enhancement and temperature calibration; S22: Analyze the infrared image through image processing algorithms to detect wear, cracks and aging areas of the lining. The image processing algorithms used include edge detection, pattern recognition, and thermal distribution anomaly detection; S23: Generate a lining status report based on the processing result of step S22, and mark the damaged or abnormal area; S3: The signal output device (8) outputs the processing result of the data processor (7), and outputs a warning signal in advance according to the processing result, reminding the user to replace the lining in time.

4. The real-time detection device for the vacuum tank lining according to claim 1, characterized in that: In step S2, the specific process of noise filtering is: Step 1: Read the infrared image and convert it into a grayscale image; Step 2: Select a Gaussian filter to smooth the image; Step 3: For areas with high noise, further median filtering or bilateral filtering is performed.

5. The real-time detection device for the vacuum tank lining according to claim 1, characterized in that: In step S2, the specific process of image enhancement is as follows: Step 1: Convert the image to grayscale; Step 2: Use histogram equalization or CLAHE to enhance the grayscale image; Step 3: Use the enhanced image for subsequent processing.

6. The real-time detection device for the vacuum tank lining according to claim 1, characterized in that: In step S2, the specific process of temperature calibration is as follows: Step 1: Obtain the temperature calibration data of the infrared detector and establish the corresponding relationship between pixel value and temperature value; Step 2: Convert the pixel value of the infrared image into a temperature value image; Step 3: Use the temperature value image for subsequent analysis and detection.

7. The real-time detection device for the vacuum tank lining according to claim 1, characterized in that: In step S2, the specific process of the edge detection algorithm is: Step 1: Preprocessing: Use Gaussian filtering to smooth the image, grayscale the infrared image, and perform noise reduction to eliminate the influence of noise; Step 2: Gradient calculation: Use the edge detection algorithm Sobel operator to derive the image and obtain the gradient magnitude and direction of each pixel; Step 3: Non-maximum suppression: suppress non-maximum values ​​along the gradient direction to ensure that the edges appear as thin lines; Step 4: Dual threshold detection: set high and low thresholds to classify pixels into strong edge, weak edge, and non-edge; Step 5: Edge tracking: Get a complete edge image by connecting strong edges and weak edges; Step 6: Edge merging and filtering: Merge and filter the noise edges and output the final edge map.

8. The real-time detection device for the vacuum tank lining according to claim 1, characterized in that: In step S2, the specific process of the pattern recognition algorithm is as follows: Step 1: Data collection: Collect infrared image data containing normal and damaged areas and label them; Step 2: Data preprocessing: standardize and enhance the image, including rotation, cropping, and brightness adjustment; Step 3: Model training: Use the labeled data set to train the CNN model. The model structure includes convolutional layer, pooling layer, and fully connected layer. Step 4: Damage area detection: Input the detection image into the trained model to predict the damage type and location; Step 5: Output results: Mark the damaged area on the original image.

9. The real-time detection device for the vacuum tank lining according to claim 1, characterized in that: In step S2, the specific process of the heat distribution anomaly detection algorithm is as follows: Step 1: Temperature mapping: Convert the infrared image into a temperature distribution map so that the temperature of each area can be observed; Step 2: Threshold segmentation: Based on the set temperature threshold, the temperature abnormality area is segmented, and the upper and lower thresholds are set to identify areas with too high or too low temperatures; Step 3: Regional analysis: Perform connected region analysis on the segmented abnormal regions to determine their area and shape and filter out small area noise; Step 4: Mark abnormal areas: Mark abnormal areas on the image for further detection and analysis.

Images