Abnormal early warning method of billet forging temperature based on infrared thermogram

Abstract

The invention relates to the technical field of image processing, and in particular to an abnormal temperature warning method for billet forging based on infrared thermograms, comprising: continuously collecting infrared thermograms of billets at multiple angles, preprocessing them into initial images, obtaining multiple continuous initial images at each angle, determining a first correction factor of the pixel point according to the grayscale difference of the same pixel point and all the pixels in its first neighborhood in these images and the area of ​​the first neighborhood, obtaining multiple initial images at each angle and its adjacent angles, determining a second correction factor of the pixel point according to the gradient difference of the same pixel point and all the pixels in its second neighborhood in these images, combining the observation angle of the image and the distance between the pixel point and a collection device, correcting the temperature value of the pixel point by using the first correction factor and the second correction factor, judging whether the temperature is abnormal, and improving the consistency of temperature measurement and the accuracy of temperature anomaly recognition.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method for early warning of abnormal billet forging temperature based on infrared thermograms. Background Art

[0002] The billet is a metal material that has been preliminarily processed, usually as a semi-finished product for manufacturing the final product. In the process of transforming these billets into finished products, they will go through a series of key steps, including smelting, casting, heat treatment and forging. First, in the smelting process, the original metal raw material (such as titanium ore) is heated to a liquid state. Then, in the casting stage, the molten metal is poured into a specific mold to cool and solidify into the initial shape. Subsequently, a series of heat treatment processes (such as annealing, normalizing, quenching and tempering) are used to enhance the physical properties of the billet. Finally, equipment such as the 80MN free forging hydraulic unit is used to apply high pressure to the billet to deform it into the desired part shape.

[0003] Temperature control is crucial to ensure the quality of billet forging. If the temperature is too high or too low, it may cause quality problems in the product, such as cracks, burns or excessive grain size, which will affect the strength and durability of the finished product. For temperature control, some advanced monitoring methods are used in current industrial practice, such as infrared online automatic temperature measurement systems and multi-camera monitoring systems. For example, the Chinese patent document with the existing authorization announcement number CN116228548B discloses an intelligent measurement method for boiler furnace temperature, including: collecting infrared temperature images of the boiler shell, converting the temperature analysis features of the multi-angle temperature images into a time series of features, constructing a matching integration of the time series by combining the shooting angle relationship through DTW, and obtaining the furnace temperature through the furnace material heat conduction formula based on the integrated data.

[0004] However, when collecting and analyzing infrared thermograms of the billet surface, the temperature value at the same location may vary due to natural factors, such as thermal radiation fluctuations. The same location may show different temperature readings in infrared thermograms taken from different angles. The temperature difference in this case will lead to inaccurate temperature readings, which in turn affects the accuracy of judging abnormal temperature conditions on the billet surface. Summary of the invention

[0005] In order to solve the problem that the temperature readings of pixels at the same position are biased due to factors such as fluctuations in thermal radiation and differences in viewing angles, which affects the accuracy of judging the abnormality of the billet surface temperature, the present invention proposes a billet forging temperature abnormality early warning method based on infrared thermogram, comprising:

[0006] Continuously collect infrared thermal images of the blank at multiple angles during the forging process, and pre-process the infrared thermal images to obtain the corresponding initial images;

[0007] Acquire a plurality of consecutive initial images at each angle, and determine a first correction factor for each pixel of each initial image at each angle according to the grayscale difference between the same pixel and all pixels in its first neighborhood in the plurality of consecutive initial images, and the area of ​​the first neighborhood of the pixel;

[0008] Acquire multiple initial images of each angle and its adjacent angles, and determine the second correction factor of each pixel of each initial image at each angle according to the gradient difference between the same pixel and all the pixels in its second neighborhood in the multiple initial images, combined with the angle between the angle and the facing direction, and the distance between the pixel and the acquisition device; the first neighborhood is larger than the second neighborhood;

[0009] The temperature value of each pixel point in each initial image at each angle is corrected using the first correction factor and the second correction factor of the pixel point, and based on the corrected temperature values ​​of all pixels in each initial image at each angle, it is determined whether there is a temperature anomaly at the time of acquisition of the initial image.

[0010] The above technical solution can obtain the temperature distribution information of the blank at different viewing angles by continuously collecting infrared thermal images from multiple angles, providing temperature data from multiple angles and multiple time points, and providing a rich information basis for subsequent temperature correction and abnormality detection. By analyzing the grayscale difference and neighborhood area of ​​the same pixel in a continuous time series, the time consistency of the temperature reading is improved and the error caused by thermal radiation fluctuations is reduced.

[0011] By analyzing the gradient differences of the same pixel at different observation angles and combining the effects of angle and distance, the temperature reading deviation caused by the difference in viewing angle can be corrected, the spatial consistency of the temperature reading is improved, and the error caused by different observation angles is reduced, making the temperature measurement more accurate and consistent. The design of the first neighborhood being larger than the second neighborhood can consider the influence of the time series in a larger range and the influence of the spatial gradient in a smaller range, thereby more comprehensively correcting the temperature reading. In addition, the correction factors of the two dimensions of time and space are comprehensively considered to correct the temperature value of each pixel, thereby improving the overall accuracy and consistency of the temperature measurement. Based on the corrected temperature value, it is possible to more accurately determine whether there is a temperature anomaly, issue a warning in time, and avoid quality problems caused by temperature anomalies.

[0012] Preferably, the first correction factor satisfies the following relationship:

[0013] ;

[0014] In the formula, is the angle No. The first image of The first correction factor for pixels, For normalization operation, , are all preset weights. is the adjustment parameter used to adjust the sequence number of the initial image to make it continuous, is the angle No. The first image of The average gray value of all pixels in the first neighborhood of a pixel, Indicates angle No. The first image of The gray value of a pixel, Angle No. The first image of The area of ​​the first neighborhood of a pixel.

[0015] The above technical solution can correct the temperature reading deviation caused by thermal radiation fluctuations and improve the temporal consistency of temperature measurement by calculating the grayscale difference of the same pixel and its neighborhood in multiple consecutive initial images. By considering the influence of the neighborhood area, the error caused by uneven local temperature distribution can be reduced and the spatial consistency of temperature measurement can be improved.

[0016] Preferably, the second correction factor satisfies the following relationship:

[0017] ；

[0018] In the formula, is the angle No. The first The second correction factor for pixels, is the first The distance between the pixel and the infrared thermal imager, is the shortest distance between all pixels of the initial image and the acquisition device. is the farthest distance between all pixels of the initial image and the acquisition device. is the natural exponential function, For normalization operation, is the adjustment parameter used to adjust the angle to make it continuous. , The angles are No. The first The gradient value of the pixel and the The average gradient value of all pixels in the second neighborhood of a pixel.

[0019] The above technical solution effectively corrects the temperature reading deviation caused by these factors by comprehensively considering the influence of distance, angle and gradient value differences, thereby improving the accuracy and consistency of temperature measurement.

[0020] Preferably, the temperature value of each pixel point of each initial image at each angle is corrected using the first correction factor and the second correction factor of the pixel point based on the following relationship:

[0021] ;

[0022] In the formula, Angle No. The first image of The corrected temperature value of each pixel, Angle No. The first image of The first correction factor for pixels, Angle No. The first image of The second correction factor for pixels, Angle No. The first image of The temperature value of a pixel.

[0023] The first correction factor of the above technical solution mainly corrects the temperature deviation caused by time series fluctuations, and the second correction factor mainly corrects the temperature deviation caused by observation angle and distance. Combining these two correction factors can more comprehensively correct the temperature reading and improve the accuracy of temperature measurement.

[0024] Preferably, the first neighborhood and the second neighborhood are determined respectively according to the following methods:

[0025] For each pixel point of each initial image at each angle, an iterative self-organizing clustering algorithm is used to adaptively cluster the pixel point. After the clustering is completed, all the pixels contained in the cluster centered on the pixel point constitute the first neighborhood of the pixel point.

[0026] For each pixel point of each initial image at each angle, its eight neighboring pixels constitute the second neighborhood of the pixel point.

[0027] Preferably, the area of ​​the first neighborhood is determined based on the number of all pixels contained therein.

[0028] Preferably, the method for judging whether there is a temperature anomaly at the time of acquisition of each initial image based on the corrected temperature values ​​of all pixels in each initial image at each angle is:

[0029] Acquire the optimal processing temperature range of the blank in advance; take the pixels in the initial image whose corrected temperature values ​​exceed the optimal processing temperature range as temperature abnormality pixels; calculate the proportion of all temperature abnormality pixels in the initial image among all pixels in the initial image; if the proportion exceeds a preset proportion threshold, there is temperature abnormality at the time of acquisition of the initial image, and an early warning is issued; if the proportion does not exceed the preset proportion threshold, there is no temperature abnormality at the time of acquisition of the initial image, and no early warning is issued.

[0030] In the above technical solution, the pre-set optimal processing temperature range is determined based on material properties and process requirements to ensure that the temperature during the forging process is within a reasonable range. By comparing the corrected temperature value with the optimal processing temperature range, the pixels with abnormal temperature can be accurately identified and the proportion of abnormal pixels in the total pixels can be calculated. The degree of temperature abnormality can be quantitatively evaluated, and a real-time feedback mechanism has been established to help timely detect and deal with temperature abnormalities and avoid product quality problems.

[0031] Preferably, the method for preprocessing the infrared thermal image includes: removing noise by a filtering algorithm, enhancing contrast by an image contrast enhancement technique, and converting the infrared thermal image into a grayscale image by grayscaling.

[0032] The present invention has the following effects:

[0033] The present invention improves the comprehensiveness and accuracy of temperature data by continuously collecting and preprocessing infrared thermograms from multiple angles. By introducing a first correction factor and a second correction factor, the temperature deviation caused by time series fluctuations and observation angles and distances are corrected respectively, thereby improving the stability and consistency of temperature measurement. Abnormality detection using the comprehensively corrected temperature value can more accurately identify temperature abnormalities and issue early warnings in a timely manner, thereby overall improving the temperature control accuracy and product quality in the forging process. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0035] Figure 1 It is a schematic flow chart of the method of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0037] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0038] Reference Figure 1 The method for early warning of abnormal billet forging temperature based on infrared thermogram provided by the present invention comprises steps S1 to S4:

[0039] S1: Collect infrared thermal images of the blank from multiple angles during the forging process and perform preprocessing to obtain the corresponding initial image.

[0040] Although traditional methods can provide temperature monitoring to a certain extent, they often do not fully consider the temperature changes at the same point at different times and viewing angles. It is necessary to overcome the temperature fluctuations caused by natural factors and the data inconsistency caused by multi-angle observations in order to achieve more accurate temperature control. The present invention aims to improve the consistency and reliability of temperature measurement and improve the accuracy of temperature monitoring by conducting a comprehensive analysis of time series and multi-view data.

[0041] In this step, during the forging process of the billet, the infrared online automatic temperature measurement system in the 80MN free forging machine is used to collect infrared thermal images of the billet at all angles every second. That is, multi-angle image acquisition is performed on the forging process of the billet. The acquisition frequency is once per second, and an infrared thermal image of one angle is collected every second. These infrared thermal images can help monitor the temperature distribution of the billet and ensure the uniformity and quality of the forging process.

[0042] Since the infrared detector needs to work for a long time, its internal electronic components may introduce noise due to long-term operation, resulting in inaccurate data in the collected infrared thermal map.

[0043] Therefore, this step performs the following preprocessing on the infrared thermal image collected every second:

[0044] Noise removal: The collected infrared thermal images are filtered and denoised through an adaptive filtering algorithm. The adaptive filtering algorithm will identify the noisy parts in the image and smooth them out. This is an intelligent filtering method that can automatically adjust the filtering parameters according to different areas of the image, effectively removing noise while retaining useful information.

[0045] Improve contrast: Perform histogram equalization on the collected infrared thermal images. This is a technology that enhances image contrast. By redistributing the grayscale values ​​of the image, the brightness distribution of the image is made more uniform, thereby improving the image contrast. This makes temperature changes more obvious and easier to observe.

[0046] Grayscale: Convert the infrared thermal map after noise removal and contrast enhancement into a grayscale map. The grayscale map only contains brightness information but no color information, which is more suitable for subsequent analysis and processing. The higher the temperature value of the pixel in the infrared thermal map, the larger the corresponding grayscale value in the grayscale map, that is, the brighter it is. The lower the temperature value of the pixel in the infrared thermal map, the smaller the corresponding grayscale value in the grayscale map, that is, the darker it is.

[0047] Finally, the infrared thermal image collected at each angle every second is used as an initial image after the above preprocessing, and the initial image is used for subsequent further analysis and processing.

[0048] Through this step, it can be ensured that the collected infrared thermal images have high quality and accuracy, providing a reliable data basis for subsequent analysis.

[0049] S2: Acquire multiple consecutive initial images at each angle, and determine a first correction factor of the pixel according to grayscale changes of the same pixel and all pixels in its first neighborhood in the multiple consecutive initial images, and the area of ​​the first neighborhood of the pixel.

[0050] During the forging process of the billet, the heat conduction of the billet is continuous. Therefore, for the same angle in the infrared thermogram based on time series acquisition, the temperature change of the pixel point is continuous, that is, the temperature change of a pixel point in a short time (such as a few seconds or minutes) should be smooth without drastic fluctuations. In addition, the temperature of a pixel point is not only affected by its own characteristics, but also related to the temperature distribution in its neighborhood. The temperature change in the neighborhood can provide additional information for the temperature change of the point. Since the infrared thermogram has been converted into a corresponding grayscale image, both the temperature change of a single pixel point and the temperature distribution in the neighborhood of the pixel point can be reflected in the grayscale image through the change of the grayscale value of the pixel point.

[0051] The specific operation includes: first, obtaining all the initial images of each angle in the forging process (grayscale images corresponding to the infrared thermal images), then, for any initial image of the angle, adaptively clustering each pixel of the initial image, in the clustering result, all the pixels contained in the cluster centered on the pixel constitute the first neighborhood of the pixel, and the grayscale value change of all the pixels in the first neighborhood reflects the temperature change in the neighborhood of the pixel. Finally, according to the average level of the difference between the grayscale value of the same pixel and the average grayscale value of all the pixels in its first neighborhood in multiple consecutive initial images, and the area of ​​the first neighborhood of the pixel (equal to the number of all the pixels in the first neighborhood), the first correction factor of the pixel is calculated.

[0052] In one example, let the angles be , in degrees For example, the initial image of the front side of the blank (the grayscale image corresponding to the infrared image) is obtained. During the entire forging process, a total of The initial image at that angle.

[0053] By angle No. The first image of For example, use Clustering algorithm for The first image of Pixels are adaptively clustered, including:

[0054] Set clustering related parameters: Determine the neighborhood radius: Use the elbow method to determine the best neighborhood radius. The elbow method finds an "inflection point" by plotting the clustering effect under different neighborhood radii. The neighborhood radius corresponding to this "inflection point" is the best neighborhood radius. Set the minimum number of points: Here, the minimum number of points is set to 3 (empirical value), which means that at least 3 pixels are required in the neighborhood to form a cluster.

[0055] Perform clustering operation: The first image of Pixels are used as the center, and (Iterative Self-Organizing Clustering) algorithm finds all the pixels around it that together form a cluster.

[0056] After clustering is completed, obtain The first image of The pixel point is the center of the cluster, and all the pixels contained in the cluster constitute the The first neighborhood of the pixel point, count the number of all pixels in the first neighborhood, and record this number as the The first image of The neighborhood area of ​​the pixel point can be found by this method. The first image of The pixels that are closely adjacent to each other are clustered together to better analyze the temperature changes in the local area.

[0057] Finally, determine the angle No. The first image of the The first correction factor of each pixel point specifically includes:

[0058] Get Angle No. The first image of the The gray value of a pixel is ,angle No. The first image of the The gray value of a pixel is ,angle No. The first image of the The gray value of a pixel is ;

[0059] Calculating Angles No. The first image of The average gray value of all pixels in the first neighborhood of a pixel , calculate the angle No. The first image of The average grayscale value of all pixels in the first neighborhood of a pixel is , calculate the angle No. The first image of the The average grayscale value of all pixels in the first neighborhood of a pixel is .

[0060] The angle No. The first image of The calculation formula of the first correction factor of a pixel is:

[0061]

[0062] In this formula, is the angle No. The first image of the The first correction factor for pixels, For normalization operation, 、 They are and The weight of Used to adjust the sequence number of the initial image to make it continuous. Indicates angle No. The first image of The average gray value of all pixels in the first neighborhood of a pixel, Indicates angle No. The first image of The gray value of a pixel, Angle No. The first image of The area of ​​the first neighborhood of a pixel.

[0063] In this formula, Indicated in angle Of the three consecutive initial images, The average value of the difference between the gray value of the pixel and the average gray value of all pixels in its first neighborhood. The larger the average value, the If there is a large difference between the grayscale values ​​of a pixel and all the pixels in its first neighborhood, the grayscale value of the pixel is more likely to be abnormal and needs to be corrected.

[0064] In this formula, Angle No. The first image of The first neighborhood area of ​​each pixel. In the initial image, the gray value of each pixel reflects the temperature of the point. According to the diffusion characteristics of temperature, the area with temperature anomaly is usually centered on a high or low temperature center point, forming a small area, and this abnormal temperature will gradually spread from the center to the surroundings. Pixels with smaller first neighborhood areas are more likely to be located in the center of this abnormal area. By analyzing the temperature distribution of pixels and their first neighborhoods, temperature anomalies can be more accurately identified and corrected.

[0065] In summary, if the neighborhood area of ​​a pixel is small, and the mean of the difference between the grayscale value of the pixel and the average grayscale value of all pixels in its first neighborhood is larger, and The smaller the angle, the No. The first image of The more likely a pixel is to be in a small area of ​​abnormal temperature, the more likely its temperature reading is inaccurate and requires a larger correction. On the contrary, if the area of ​​the first neighborhood of a pixel is larger, and the average value of the difference between the gray value of the pixel and the average gray value of all pixels in its first neighborhood is smaller, that is The larger the The smaller the angle, the No. The first image of the The more likely a pixel is to be within a large area with a relatively uniform temperature distribution, the more likely the temperature reading at that pixel is to be relatively normal and require a smaller correction. The smaller the value of More accurately reflects the The local grayscale features of pixels, so a larger weight is set for them. Set to 0.6, through the The size of the first neighborhood area of ​​a pixel is a rough method to judge whether the pixel is located in the temperature abnormal area, because the first neighborhood area may appear under extreme conditions. The area of ​​the first neighborhood of a pixel is large, but the pixel is in the abnormal temperature area. At this time, the reliability of the first neighborhood area is low, so Set to 0.4.

[0066] For all the initial images at each angle in the forging process, the first correction factor of each pixel point of each initial image at each angle is obtained according to the method of this step.

[0067] S3: Obtain multiple initial images of each angle and its adjacent angles, and calculate the second correction factor of each pixel in each initial image of each angle according to the gradient change of the same pixel point and its second neighborhood pixel point in multiple consecutive initial images, combined with the angle between the angle and the opposite direction, and the distance between the pixel point and the acquisition device.

[0068] During the billet forging process, an infrared thermal imager is used to collect infrared thermal images from different angles. Due to the different collection angles, the pixels at the same position will receive thermal radiation of different intensities, resulting in deviations in temperature readings. Since factors such as fluctuations in thermal radiation and differences in viewing angles can easily lead to deviations in directly measured temperatures, it is necessary to correct the temperature of each pixel in the initial image to improve the accuracy and consistency of temperature measurement.

[0069] This step mainly considers the following two aspects:

[0070] The first is the effect of distance on temperature readings: the farther away the pixel is from the infrared thermal imager, the less accurate its temperature reading is, because the radiation intensity received by the infrared thermal imager decreases with the increase of distance. According to the Stefan-Boltzmann law, the radiation intensity is inversely proportional to the square of the distance. Therefore, the farther the pixel is from the infrared thermal imager, the weaker its radiation signal is, resulting in a decrease in the accuracy of the temperature reading. Pixels at a distance may be subject to more noise interference due to weak signals, resulting in inaccurate temperature readings.

[0071] Second, the effect of the observation angle on the temperature reading: When using an infrared thermal imager for temperature measurement, the observation angle has a great influence on the accuracy of the temperature reading. Generally speaking, the image taken at the positive angle (when the angle is 90 degrees, facing the camera) is more accurate, because the received radiation signal is the strongest at this time. If the observation angle deviates from the positive direction, the received radiation signal will be weakened, and the accuracy of the temperature reading will decrease, resulting in inaccurate temperature readings. This is because the radiation signal will be scattered and attenuated during the propagation process. The greater the angle of deviation from the positive direction, the more obvious this effect is. For example, an infrared thermal imager is used to monitor the temperature of a piece of metal billet during the forging process: when the infrared thermal imager is facing the metal billet (at an angle of 90 degrees), the temperature reading obtained is very accurate, and the actual temperature distribution of the metal billet can be clearly seen. If the infrared thermal imager is tilted at a certain angle (such as 45 degrees or 30 degrees), the received radiation signal will be weakened, resulting in low or inaccurate temperature readings, and the true temperature of the metal billet cannot be accurately judged. Therefore, it is necessary to introduce a correction factor to correct the deviation of the temperature reading to improve the accuracy of the measurement.

[0072] Therefore, considering that both distance and observation angle will affect the received radiation signal strength and thus the accuracy of the temperature reading, this step can effectively correct these deviations and improve the accuracy and consistency of temperature measurement by introducing a second correction factor.

[0073] For any angle, all initial images of the angle and all initial images of the adjacent angles before and after the angle are obtained. For any pixel point of any initial image of the angle, its eight neighborhood pixels constitute the second neighborhood of the pixel point. According to the average level of the difference between the gradient value of the same pixel point and the average gradient value of all pixels in its second neighborhood in the initial images of consecutive adjacent angles, the distance between the pixel point and the infrared thermal imager, and the angle between the angle of the initial image and the positive direction, the second correction factor of each pixel point of each initial image of each angle is calculated.

[0074] By angle Take for example, get all the initial images of the angle, for angle No. The first pixel point, calculate the gradient value of the pixel point ,angle No. The first The average gradient of all pixels in the second neighborhood of a pixel is ,angle No. The first The distance between a pixel and the infrared thermal imager is , calculate the angle No. The distance between all pixels of the initial image and the infrared thermal imager, and the closest distance is recorded as , and the farthest distance is recorded as .

[0075] Get the angle in the same way All initial images and angles All initial images of , calculate the angle No. The first The gradient value of a pixel is ,angle No. The first The average gradient of all pixels in the second neighborhood of a pixel is , calculate the angle No. The first The gradient value of a pixel is ,angle No. The first The average gradient of all pixels in the second neighborhood of a pixel is .

[0076] Calculate the angle No. The first The second correction factor for pixels:

[0077] In the formula, is the angle No. The first The second correction factor for pixels, is the natural exponential function, For normalization operation, is the adjustment parameter used to adjust the angle to make it continuous. Angle No. The first The gradient value of a pixel, Angle No. The first The average gradient value of all pixels in the second neighborhood of a pixel.

[0078] In this formula, is the angle No. The first The distance between a pixel and the infrared thermal imager is the normalized difference between the closest distance and the farthest distance between all pixels in the initial image and the infrared thermal imager. No. The first The farther the pixel is from the camera, the larger the ratio is and the larger the second correction factor is. This is because the radiation signal received by the distant pixel is weaker and the temperature reading is more inaccurate.

[0079] In this formula, This item indicates the current angle The normalized difference from the positive direction (90 degrees). The more the current angle deviates from the positive direction, The larger the ratio, The smaller the overall value of The larger the value of , the larger the second correction factor will be. This is because an angle that deviates from the positive direction will cause the received radiation signal to weaken and the temperature reading to be inaccurate.

[0080] In this formula, Indicated in angle ,angle , and angle The first The first The normalized average of the differences between the gradient value of a pixel and the gradient values ​​of all pixels in its second neighborhood. The gradient value reflects the temperature change between the pixel and its neighborhood. If the gradient value difference is large, it means that the pixel may be subject to noise or other interference and requires a larger correction. Normalize the results to ensure that the correction factor is within a reasonable range.

[0081] According to the above operation, the angle can be obtained The second correction factor for each pixel in each initial image is taken into account. The second correction factor comprehensively considers the distance between the pixel and the infrared thermal imager, the angle between the observation angle and the positive direction, and the gradient difference between the pixel and the field pixel. The influence of these factors on the temperature reading is analyzed, and the correction factor is adjusted through exponential decay and normalization processing. This method can effectively correct the temperature reading deviation caused by different observation conditions, improve the accuracy and consistency of infrared thermal imager measurement, and ensure the reliability and accuracy of temperature data.

[0082] S4: Correct the temperature value of each pixel point of each initial image at each angle using the first correction factor and the second correction factor of the pixel point, and determine whether there is a temperature anomaly based on the corrected temperature value.

[0083] By angle No. The first image of the For example, the temperature value of the pixel is , calculate the corrected temperature value:

[0084]

[0085] In the formula, Angle No. The first image of the The corrected temperature value of each pixel, Angle No. The first image of the The first correction factor for pixels, Angle No. The first image of the The second correction factor for each pixel.

[0086] Since each initial image is preprocessed from the infrared thermal map, the grayscale value of the pixel in the initial image and the temperature value of the pixel in the infrared thermal map have a mapping relationship. Therefore, correcting the temperature value of each pixel is to correct the grayscale value of the pixel. According to this method, the corrected temperature value of each pixel of each initial image at each angle can be calculated.

[0087] Establish an early warning system: obtain the optimal processing temperature range of the blank in advance , is the lower limit of the optimal processing temperature, It is the upper limit of the optimal processing temperature.

[0088] For each initial image at each angle, the corrected temperature values ​​of all pixels in the initial image are compared with the optimal processing temperature range, and the corrected temperature values ​​exceeding [ ] pixels are regarded as abnormal pixels, and the corrected temperature value does not exceed [ ] as normal pixels, and count the total number of all abnormal pixels in the initial image. If the total number of abnormal pixels accounts for the total number of all pixels in the initial image, When the temperature is abnormal, it is considered to be abnormal and triggers an early warning. This step sets the preset ratio. This value is an empirical value and can be set by yourself, but considering the sensitivity and timeliness of the warning, it is usually set at ~ Inside.

[0089] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0090] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. A method for early warning of abnormal billet forging temperature based on infrared thermogram, characterized in that: include: Continuously collect infrared thermal images of the blank at multiple angles during the forging process, and pre-process the infrared thermal images to obtain the corresponding initial images; Acquire a plurality of consecutive initial images at each angle, and determine a first correction factor for each pixel of each initial image at each angle according to the grayscale difference between the same pixel and all pixels in its first neighborhood in the plurality of consecutive initial images, and the area of ​​the first neighborhood of the pixel; The first correction factor satisfies the following relationship: ； In the formula, is the angle No. The first image of the The first correction factor for pixels, For normalization operation, 、 are all preset weights. is the adjustment parameter used to adjust the sequence number of the initial image to make it continuous, is the angle No. The first image of the The average gray value of all pixels in the first neighborhood of a pixel, Indicates angle No. The first image of the The gray value of a pixel, Angle No. The first image of the The area of ​​the first neighborhood of a pixel; Acquire multiple initial images of each angle and its adjacent angles, and determine the second correction factor of each pixel of each initial image at each angle according to the gradient difference between the same pixel and all the pixels in its second neighborhood in the multiple initial images, combined with the angle between the angle and the facing direction, and the distance between the pixel and the acquisition device; The second correction factor satisfies the following relationship: ； In the formula, is the angle No. The first The second correction factor for pixels, is the first The distance between the pixel and the infrared thermal imager, is the shortest distance between all pixels of the initial image and the acquisition device. is the farthest distance between all pixels of the initial image and the acquisition device. is the natural exponential function, is the adjustment parameter used to adjust the angle to make it continuous. , The angles are No. The first The gradient value of the pixel and the The average gradient value of all pixels in the second neighborhood of a pixel; The first neighborhood is larger than the second neighborhood; The temperature value of each pixel point in each initial image at each angle is corrected using the first correction factor and the second correction factor of the pixel point, and based on the corrected temperature values ​​of all pixels in each initial image at each angle, it is determined whether there is a temperature anomaly at the time of acquisition of the initial image.

2. The method for early warning of abnormal billet forging temperature based on infrared thermogram according to claim 1, characterized in that: The temperature value of each pixel of each initial image at each angle is corrected using the first correction factor and the second correction factor of the pixel based on the following relationship: ； In the formula, Angle No. The first image of the The corrected temperature value of each pixel, Angle No. The first image of the The first correction factor for pixels, Angle No. The first image of the The second correction factor for pixels, Angle No. The first image of the The temperature value of a pixel.

3. The method for early warning of abnormal billet forging temperature based on infrared thermogram according to claim 1, characterized in that: The first neighborhood and the second neighborhood are determined respectively according to the following methods: For each pixel point of each initial image at each angle, an iterative self-organizing clustering algorithm is used to adaptively cluster the pixel point. After the clustering is completed, all the pixels contained in the cluster centered on the pixel point constitute the first neighborhood of the pixel point. For each pixel point of each initial image at each angle, its eight neighboring pixels constitute the second neighborhood of the pixel point.

4. The method for early warning of abnormal billet forging temperature based on infrared thermogram according to claim 1, characterized in that: The area of ​​the first neighborhood is determined based on the number of all pixel points contained therein.

5. The method for early warning of abnormal billet forging temperature based on infrared thermogram according to claim 2, characterized in that: The method for judging whether there is a temperature anomaly at the time of acquisition of the initial image based on the corrected temperature values ​​of all pixels in each initial image at each angle is as follows: Acquire the optimal processing temperature range of the blank in advance; take the pixels in the initial image whose corrected temperature values ​​exceed the optimal processing temperature range as temperature abnormality pixels; calculate the proportion of all temperature abnormality pixels in the initial image among all pixels in the initial image; if the proportion exceeds a preset proportion threshold, there is temperature abnormality at the time of acquisition of the initial image, and an early warning is issued; if the proportion does not exceed the preset proportion threshold, there is no temperature abnormality at the time of acquisition of the initial image, and no early warning is issued.

6. The method for early warning of abnormal billet forging temperature based on infrared thermogram according to claim 1, characterized in that: The method of preprocessing the infrared thermal image includes: removing noise through filtering algorithm, enhancing contrast through image contrast enhancement technology, and converting the infrared thermal image into a grayscale image through grayscale.

Images