A method for detecting regions of interest in continuous casting leaks based on spatial features

Abstract

This invention discloses a method for detecting the region of interest (ROI) in continuous casting leaks based on spatial features, belonging to the field of iron and steel metallurgy. Its features include: visualization of the temperature rate of the continuous casting mold, extraction of connected region features, determination of hot and cold regions, and detection of the ROI in continuous casting leaks. This invention combines the spatial features of leaks in the continuous casting process with the ROI detection method, helping operators observe the ROI characteristics of leaks and providing a method for rapid and accurate identification of leaks in continuous casting.

Description

Technical Field

[0001] This invention belongs to the field of continuous casting technology in iron and steel metallurgy, and is a method for detecting regions of interest in continuous casting leakage based on spatial characteristics. Background Technology

[0002] As a major catastrophic accident in continuous casting production, steel leakage not only disrupts the normal continuous casting process and causes production disorder, but also reduces the production efficiency of the continuous casting machine and damages the equipment, resulting in huge economic losses for enterprises. In order to reduce the interference of steel leakage accidents on continuous casting production, the main influencing factors of steel leakage have received widespread attention from metallurgists, and steel leakage prediction methods are constantly being optimized and researched.

[0003] In actual casting processes, due to the complex and invisible temperature field within the continuous casting mold, thermocouples are typically used to measure the temperature signal within the mold. This allows for the capture of typical temperature patterns in bonded cracks, the establishment of logical judgment models for cracks, or the development of intelligent models such as BP neural networks and clustering to predict continuous casting cracks. However, recent production practices have revealed some shortcomings, primarily manifested in frequent missed detections and widespread false alarms. The main reason is that the formation and development of continuous casting cracks exhibit spatial and temporal propagation characteristics, making it difficult for one-dimensional temperature time-series trends to capture the spatial morphology and dynamic evolution of bonded cracks. Therefore, there is an urgent need to explore new, more comprehensive, accurate, and efficient methods for detecting and predicting cracks.

[0004] Patent document CN109954854 discloses a method for monitoring steel leakage in a continuous casting machine mold. The method for determining the region of interest (ROI) of the cast billet includes: acquiring a grayscale histogram of a preprocessed cast billet image; calculating the average of two peaks in the grayscale histogram and using this average as a grayscale threshold; determining the pixel value of each pixel in the cast billet image relative to the grayscale threshold, identifying pixels with values ​​greater than the threshold as the cast billet region; and determining the minimum bounding rectangle of the cast billet region, which is then considered the ROI. However, when the temperature fluctuation within the mold is significant, numerous abnormal small regions may appear in the corresponding cast billet image, leading to a deviation in the selection of the ROI.

[0005] Patent document CN113579190 discloses a method for predicting steel leakage in slab continuous casting based on regional characteristics. This method automatically shields abnormally fluctuating thermocouples based on their state, eliminating abnormal temperature values. It judges temperature rise and then temperature drop using logical rules, finally identifying the bonding process based on the quantitative characteristics and propagation properties of regional thermocouple temperature rises, making the judgment entirely based on the bonding development sequence. However, this method still relies on a one-dimensional thermocouple temperature signal to establish a logical judgment model for predicting steel leakage, failing to intuitively and effectively represent the two-dimensional spatial characteristics of steel leakage, which may lead to false alarms. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an intuitive and rapid method for detecting regions of interest (ROIs) in continuous casting based on spatial features. This method combines the spatial features of steel leakage during continuous casting with ROI detection methods. By analyzing the spatial features of the thermal image of the crystallizer temperature rate, the ROI is determined, thereby helping operators to more intuitively capture the features of the ROI when steel leakage occurs, and facilitating rapid and accurate identification of steel leakage in continuous casting.

[0007] To achieve the above objectives, the technical solution adopted by this invention is: a method for detecting regions of interest in continuous casting leakage based on spatial features, which mainly includes the following steps:

[0008] 1) Visualization of temperature rate in continuous casting crystallizer

[0009] ①Acquire the temperature signal of the crystallizer thermocouple in real time and remove the thermocouples with faults;

[0010] ② By using nonlinear interpolation, the temperature at non-thermocouple measuring points is obtained, and a thermal image of the crystallizer temperature rate is drawn to visualize the crystallizer temperature rate.

[0011] 2) Connected region feature extraction

[0012] ① The temperature rate thermal image of the crystallizer is preprocessed using a threshold segmentation algorithm;

[0013] ② Extract the maximum temperature rate G of the connected region max ;

[0014] ③ Obtain the area S of the connected region;

[0015] ④ Calculate the x-coordinate and y-coordinate of the centroid of the connected region according to formulas (1) and (2);

[0016]

[0017]

[0018] In the formula, X and Y are the x and y coordinates of the centroid of the connected region; S is the area of ​​the connected region in the temperature rate thermal image; M 10 M is the sum of all x-coordinates within a connected region; 01 It is the sum of all ordinates within a connected region;

[0019] 3) Determination of hot and cold zones

[0020] ①The maximum temperature rate G max A connected region greater than G1 is identified as a hot region;

[0021] ②The maximum temperature rate G max Connected regions smaller than G2 are classified as cold regions.

[0022] 4) Detection of regions of interest in continuous casting leaks

[0023] ① In the temperature rate thermal image, obtain the largest thermal region. If the area of ​​this thermal region is greater than S... min Then, this hot region is designated as the region of interest, and the x-coordinate of the centroid of the region of interest is recorded. hi and the vertical axis Y hi Otherwise, return to step 1) and proceed to the next step.

[0024] ② According to formula (3), calculate the x-coordinate of the centroid of all cold regions. c x-coordinate of the centroid of the region of interest hi Do the difference, if X d Less than X lim And the centroid of the cold region is Y. c Less than Y lim If the cold region is identified as the region of interest, then return to step 1) to make a judgment at the next time step.

[0025] X d =|X c -X hi | (3)

[0026] In the formula, X d X is the absolute value of the difference between the x-coordinate of the barycenter of the cold region and the x-coordinate of the barycenter of the region of interest; c X is the x-coordinate of the centroid of the cold region. hi The x-coordinate of the centroid of the region of interest.

[0027] Specifically, the temperature rate threshold G1 for the hot region is set to 0.8℃ / s; the temperature rate threshold G2 for the cold region is set to -0.3℃ / s; the minimum area of ​​the hot region is set to 1973 pixels; and the difference X between the centroid abscissa of the cold region and the centroid abscissa of the hot region of interest is set to... limSet to 19; the ordinate of the centroid of the cold region is Y. lim Set to 38.

[0028] The beneficial effects of this invention's spatial feature-based method for detecting regions of interest (ROIs) in continuous casting leaks are as follows: It combines the spatial characteristics of the hot and cold regions in the continuous casting process with ROI detection methods. By determining the spatial characteristics of the crystallizer temperature rate thermal image, the ROI is identified, thus helping operators more intuitively capture the characteristics of the ROI when a leak occurs, facilitating rapid and accurate identification of continuous casting leaks. This method is applicable to the detection of ROIs in continuous casting of slabs, round billets, and irregularly shaped billets. Attached Figure Description

[0029] Figure 1 Flowchart for detecting regions of interest in continuous casting leaks;

[0030] Figure 2 It is a thermal image of the crystallizer temperature rate;

[0031] Figure 3 It is a temperature rate thermal image after threshold segmentation;

[0032] Figure 4 The hot and cold regions of interest are those in the continuous casting leak.

[0033] In the diagram: 1. Hot area; 2. Cold area. Detailed Implementation

[0034] The present invention will be further described below with reference to embodiments, but the invention is not limited thereto.

[0035] like Figure 1 As shown, the present invention provides a method for detecting regions of interest in continuous casting leaks based on spatial features, which mainly includes the following steps:

[0036] Step 1: Visualization of Temperature Rate in Continuous Casting Crystallizer

[0037] ①Acquire the temperature signal of the crystallizer thermocouple in real time and remove the thermocouples with faults;

[0038] ② By using nonlinear interpolation, the temperature at non-thermocouple measuring points is obtained, and a thermal image of the crystallizer temperature rate is plotted to visualize the crystallizer temperature rate, such as... Figure 2 As shown.

[0039] Step 2: Spatial Feature Extraction of Connected Regions

[0040] ① The temperature rate thermal image of the crystallizer is preprocessed using a threshold segmentation algorithm, and the result is as follows: Figure 3 As shown;

[0041] ② Extract the maximum temperature rate G of the connected region max ;

[0042] ③ Obtain the area S of the connected region;

[0043] ④ Calculate the x-coordinate and y-coordinate of the centroid of the connected region according to formulas (1) and (2);

[0044]

[0045]

[0046] In the formula, X and Y are the x and y coordinates of the centroid of the connected region; S is the area of ​​the connected region in the temperature rate thermal image; M 10 M is the sum of all x-coordinates within a connected region; 01 It is the sum of all ordinates within a connected region.

[0047] Step 3: Determining Hot and Cold Zones

[0048] ①The maximum temperature rate G max Connected regions with a temperature / s greater than 0.8℃ are classified as hot regions;

[0049] ②The maximum temperature rate G max Connected regions with a temperature / s rate less than -0.3℃ are classified as cold regions.

[0050] Step 4: Detection of Region of Interest (ROI) in Continuous Casting Leakage

[0051] ① In the temperature rate thermal image, acquire the largest thermal region. If the area of ​​this thermal region is greater than 1973 pixels, then this thermal region is designated as the region of interest. At the same time, record the x-coordinate of the centroid of the region of interest. hi and the vertical axis Y hi Otherwise, return to step one and proceed to the next step for judgment;

[0052] ② According to formula (3), calculate the x-coordinate of the centroid of all cold regions. c x-coordinate of the centroid of the region of interest hi Do the difference, if X d Less than 19, and the centroid ordinate Y of the cold region is... c If the value is less than 38, the cold region will be identified as the region of interest; otherwise, return to step 1) to make the judgment at the next time step.

[0053] X d =|Xc-X hi | (3)

[0054] In the formula, X dX is the absolute value of the difference between the x-coordinate of the barycenter of the cold region and the x-coordinate of the barycenter of the region of interest; c X is the x-coordinate of the centroid of the cold region. hi The x-coordinate of the centroid of the region of interest.

[0055] Through the above steps, the hot zone 1 and cold zone 2 of interest in continuous casting can be obtained, such as... Figure 4 As shown, this region of interest can not only help on-site operators to more intuitively examine the spatial-temporal characteristics of continuous casting leakage, but also help to quickly and accurately identify continuous casting leakage and improve the accuracy of continuous casting leakage prediction.

[0056] The above-described embodiments merely illustrate the implementation of the present invention, but should not be construed as a priori limitation on the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the protection scope of the present invention.

Claims

1. A method for detecting regions of interest in continuous casting leaks based on spatial features, characterized in that, This method combines the spatial characteristics of steel leakage during continuous casting with the region of interest (ROI) detection method. By using the spatial characteristics of steel leakage from the thermographic image of the crystallizer temperature rate, the ROI is determined, helping operators to more intuitively capture the ROI and facilitating the rapid and accurate identification of steel leakage in continuous casting. The specific detection steps are as follows: 1) Visualization of temperature rate in continuous casting crystallizer ①Acquire the temperature signal of the crystallizer thermocouple in real time and remove the thermocouples with faults; ② By using nonlinear interpolation, the temperature at non-thermocouple measuring points is obtained, and a thermal image of the crystallizer temperature rate is drawn to visualize the crystallizer temperature rate. 2) Feature extraction of connected components ① The temperature rate thermal image of the crystallizer is preprocessed using a threshold segmentation algorithm; ② Extract the maximum temperature rate of the connected region G max ; ③ Obtain the area of ​​the connected region S ; ④ Calculate the abscissa of the centroid of the connected region according to formulas (1) and (2). X and ordinate Y ; （1） （2） In the formula, X and Y These are the x and y coordinates of the centroid of the connected region; S It is the area of ​​the connected region in the temperature rate thermal image; M 10 It is the sum of all x-coordinates within a connected region; M 01 It is the sum of all ordinates within a connected region; 3) Determination of hot and cold zones ①Maximum temperature rate G max Greater than G 1 The connected regions are identified as hot regions; ②Maximum temperature rate G max Less than G 2 The connected regions are identified as cold regions; 4) Detection of regions of interest in continuous casting leaks ① In the temperature rate thermal image, obtain the largest thermal region. If the area of ​​this thermal region is greater than... S min Then, this hot region is taken as the region of interest, and the x-coordinate of the centroid of the region of interest is recorded. X hi and ordinate Y hi Otherwise, return to step 1) to make the next judgment. ② According to formula (3), calculate the abscissa of the centroid of all cold regions. X c The x-coordinate of the centroid of the region of interest X hi If you do something wrong, X d Less than X lim And the centroid ordinate of the cold region Y c Less than Y lim If the cold region is identified as the region of interest, then the cold region will be determined as the region of interest; otherwise, return to step 1) to make the judgment at the next moment. X d =| X c - X hi | （3） In the formula, X d This is the absolute value of the difference between the abscissa of the centroid of the cold region and the abscissa of the centroid of the hot region of interest. X c The x-coordinate of the centroid of the cold region; X hi The x-coordinate of the centroid of the region of interest.

2. The method for detecting regions of interest in continuous casting leaks based on spatial features according to claim 1, characterized in that, Thermal zone temperature rate threshold G 1 Set to 0.8℃ / s; cold zone temperature rate threshold. G 2 The speed is set to -0.3℃ / s; the minimum area of ​​the hot region is set to 1973 pixels; the difference between the centroid abscissa of the cold region and the centroid abscissa of the hot region of interest is... X lim Set to 19; the centroid ordinate of the cold region. Y lim Set to 38.

3. The method for detecting regions of interest in continuous casting leakage based on spatial features according to claim 1, characterized in that: This method is applicable to the detection of regions of interest in continuous casting of slabs, round billets, and irregularly shaped billets.

Images