A method and system for controlling a gas-induced semi-solid process based on visual recognition

Abstract

The present invention discloses a method and system for controlling a gas-induced semi-solid process based on visual recognition, which is used to adjust the gas flow rate and gas temperature according to the solid content of the slurry. The system uses image processing technology to identify and extract the features of bubbles, extracts the number of bubbles, the size of bubbles, the bubble breaking time, and the number of broken bubbles, calculates the reference solid content parameter of the slurry, sets the target solid content parameter, and dynamically adjusts the gas flow rate and temperature according to the deviation between the calculated reference solid content parameter of the slurry and the target solid content parameter. The system combines algorithms such as multi-feature fusion and circular local binary pattern (LBP) texture feature extraction to achieve precise identification of bubbles and real-time monitoring of the slurry state, and proposes a set of efficient control logics for gas flow rate and temperature. Through this system, precise control of the semi-solid processing process can be achieved, improving product quality and production efficiency.

Description

Technical Field

[0001] The present invention relates to the field of semi-solid alloy manufacturing, and particularly to a method and system for controlling a gas-induced semi-solid process based on visual recognition. Background Art

[0002] Gas-Induced Semi-Solid Processing (GISS) is a metal processing method that forms semi-solid slurries by introducing gas into molten metal. This process combines the advantages of casting and forging, and can improve the mechanical properties of materials while maintaining good fluidity. The shear force of the gas-induced semi-solid process is weak. When used for preparing slurries with a large volume, as the solid content increases, it will lead to unevenly prepared slurries and the accumulation of newly formed crystal nuclei. At this time, it is necessary to increase the gas flow rate to avoid the accumulation of newly formed crystal nuclei. However, if a high gas flow rate is used at the beginning, it may cause the solution to splash. Therefore, it is necessary to adjust the gas flow rate and gas temperature according to the solid content of the slurry. However, for the slurry being prepared, it is difficult to directly observe the solid content of the slurry with the naked eye or an instrument, so the gas flow rate and gas temperature cannot be controlled in real time. Summary of the Invention

[0003] Aiming at the problem that the gas flow rate and gas temperature cannot be adjusted according to the solid content of the slurry in the prior art, the purpose of the present invention is to provide a control method and system for a gas-induced semi-solid process based on visual recognition, so as to at least partially solve the above problems.

[0004] To achieve the above purpose, the technical solution of the present invention is as follows:

[0005] Step S1: Use a camera to monitor the slurry in real time, and continuously collect frames of slurry liquid surface image information at the same interval within a preset time t a and perform frame-by-frame analysis on the obtained slurry liquid surface image information to identify the bubbles in the image.

[0006] Preferably, the time t a is 10s - 20s; the acquisition interval can be set to 0.5s - 1s.

[0007] Camera selection: Select a camera suitable for the process requirements. Generally, factors such as resolution, frame rate, and adaptability need to be considered.

[0008] Image acquisition: Install the camera and adjust its position so that it can clearly capture the slurry liquid surface. Ensure that the field of view of the camera covers the entire slurry liquid surface and can stably collect images.

[0009] Image preprocessing: Preprocess the collected images, including removing noise, enhancing contrast, etc., to obtain preprocessed images, so as to improve the accuracy and efficiency of subsequent image processing.

[0010] Image analysis: Using image processing techniques, divide the slurry liquid surface into multiple regions and identify the bubble conditions in each region. For the unique image features of the slurry liquid surface identified this time, the present invention proposes an image analysis method applicable to the slurry liquid surface. The specific image analysis method is as follows:

[0011] Step s11, threshold segmentation.

[0012] Adopt a suitable threshold segmentation algorithm (such as Otsu algorithm, adaptive threshold, etc.) to convert the collected slurry liquid surface image into a binary image, which is convenient for detecting the contour of bubbles.

[0013] Step s12, shape feature extraction.

[0014] The bubble regions in the image are usually irregular circles and are connected. Each connected domain in the binary image is a potential bubble region. Use an image edge detection algorithm (such as Canny algorithm) to extract the contour of the potential bubble region in the binary image, and then calculate the shape features of the contour, such as Hu moments or Zernike moments, to obtain the shape features of the potential bubble region.

[0015] Step s13, color feature extraction.

[0016] In the collected image, there is an obvious color difference between the color of the bubbles and the color of the slurry. Multiply the binary image obtained in step s11 by the preprocessed image, and the resulting image is the color map of the potential bubble region. For each connected domain in the color map, its color features can be extracted, which are the color features of the potential bubble region. Each potential bubble region can be converted to a suitable color space (such as HSV, LAB, etc.) to extract the color features of the potential bubble region. Methods such as color histogram or color moment can be used.

[0017] Step s14, extract the texture features of the bubble region using a circular LBP operator.

[0018] In the collected image, the texture features of the bubbles are different from those of the slurry. Multiply the binary image obtained in step s11 by the preprocessed image, and the resulting image is the color map of the potential bubble region. For each connected domain in the color map, its texture features can be extracted. Methods such as local binary pattern (LBP) can be used to extract the texture features of the bubble surface. And since the bubbles are mostly circular or elliptical in shape, a circular LBP operator is used here to extract the texture features to improve the recognition rate.

[0019] The calculation formula of circular LBP is as follows. Suppose we select a circular neighborhood with a radius of R, with the central pixel as the center, and evenly select O sampling points on the circumference:

[0020] Calculate the LBP value: For each sampling point, compare its gray value with the gray value of the central pixel. If the gray value of the sampling point is greater than or equal to the gray value of the central pixel, the weight value at this position is 1, otherwise it is 0. Connect the obtained O weight values in clockwise or counterclockwise order to form a binary number, and this binary number is the LBP value of the central pixel.

[0021] Formula representation: Suppose the gray value of the central pixel is P, and the gray value of the i-th sampling point is P i , then the LBP value LBP P,R of the central pixel can be expressed as:

[0022]

[0023] where s(x) is the sign function. If x≥0, then s(x) = 1, otherwise s(x) = 0. P i -P represents the gray value difference between the i-th sampling point and the central pixel.

[0024] Through the above formula, the LBP value of each pixel in the image can be calculated, thereby realizing the description and analysis of the image texture.

[0025] Step s15, feature fusion and recognition.

[0026] Fuse the extracted color, shape, and texture features. You can use weighted summation or feature concatenation to fuse each feature into a comprehensive feature vector. Feature fusion is to combine multiple features obtained from different feature extraction steps (color, shape, texture) to generate a comprehensive feature vector. This can make full use of the advantages of various features and improve the accuracy and robustness of bubble recognition.

[0027] Basic principle of weighted summation: Weight each feature vector according to a certain weight to generate a new feature vector. Normalize each feature vector to ensure that the dimensions of each feature are consistent. Determine the weight of each feature, which can be set according to experience or learned through training data.

[0028] F 融合 = w 1 *F 颜色 + w 2 *F 形状 + w 3 *F 纹理

[0029] where, F融合 is the fused feature vector, w 1 , w 2 , w 3 are the weights of each feature, and their sum is 1. Initially, each weight can be set to 1 / 3. F 颜色 , F 形状 , F 纹理 are the color feature, shape feature, and texture feature extracted using the circular LBP operator, respectively.

[0030] Through the feature fusion method, multiple features such as color, shape, and texture can be effectively combined to extract the fused bubble features, generate a comprehensive feature vector containing rich information, input the comprehensive feature vector into the trained classifier, and the training of the classifier is also completed based on the extracted fused bubble features, enabling the recognition of bubbles in the collected images.

[0031] Through the above steps, real-time monitoring of the slurry liquid level and acquisition of image information can be achieved, providing basic data and support for subsequent process control.

[0032] Step S2: Divide the slurry liquid level into multiple regions, identify and record the number of bubbles, bubble size, number of broken bubbles, and break time in each region, and calculate the average bubble size and average break time.

[0033] Region division: According to actual requirements and process characteristics, divide the slurry liquid level into multiple regions [a 1 , a 2 ,... a n . It can be divided according to image features, bubble distribution, etc., ensuring that the size of each region is appropriate and can contain enough bubbles.

[0034] Bubble recognition: According to bubble characteristics, use the fused bubble features obtained in step s1 to achieve the recognition of bubbles in each region, and label the recognized bubbles for easy distinction and statistics. For example, label the recognized bubbles as bubble 1, bubble 2... bubble n.

[0035] Bubble number statistics: Count the number of recognized bubbles and record the total number of bubbles within the preset time t a in each region.

[0036] Bubble size measurement: Measure the sizes of the recognized bubbles 1, 2... n, record the size information of the bubbles in each region, obtain the bubble size by pixel counting or measuring the area of the bubble connected domain, and take its average value. The average value calculation method is the total pixel count of all bubble regions or the total area of all measured bubble connected domains, divided by the number of recognized bubbles.

[0037] Record of the number of broken bubbles: Bubbles 1, 2,..., n are identified. When a bubble breaks, it is marked as a broken bubble. For example, when bubble 1 breaks, bubble 1 is marked as a broken bubble and the number of broken bubbles is incremented by 1; if bubble 2 does not break, it is not marked; when bubble 3 breaks, bubble 3 is marked as a broken bubble and the number of broken bubbles is incremented by 1 again, and so on. The number of broken bubbles within a preset time t in each area is recorded. a within the area.

[0038] Record of the bubble break time: The bubble break time refers to the time elapsed from the formation of the bubble to its final rupture and disappearance. In a liquid, a bubble gradually grows over time until it reaches a certain size and then ruptures and disappears. The bubble break time can reflect the stability and durability of the bubble, and the break time point can be determined by monitoring the change or disappearance of the bubble morphology. When a bubble is formed, taking bubble 1 as an example, the time point of its appearance is recorded as the formation time of bubble 1; when this bubble 1 ruptures and disappears, the time point of its break is recorded as the break time of bubble 1. The break time of bubble 1 is the break time point of bubble 1 minus the formation time point of bubble 1, and this process is carried out for all bubbles. The average break time of all broken bubbles is recorded to eliminate the influence of the instability of individual bubbles.

[0039] The relationship between the solid content of the slurry and the above various parameters is as follows:

[0040] Relationship between the solid content of the slurry and the number of bubbles:

[0041] High solid content: The higher the solid content of the slurry, the higher the viscosity of the slurry usually is. In a slurry with high viscosity, it is more difficult for bubbles to form because gas diffuses slowly in a high-viscosity environment and the resistance to bubbles is greater. Therefore, in a slurry with high solid content, the number of bubbles decreases.

[0042] Low solid content: When the solid content of the slurry is low, the viscosity is low, and gas is more likely to diffuse in the slurry and form bubbles. Therefore, in a slurry with low solid content, the number of bubbles increases.

[0043] Relationship between the solid content of the slurry and the stability and break time of bubbles:

[0044] High solid content: The particles in a slurry with high solid content can increase the stability of bubbles because solid particles can form a stable interface on the bubble surface to prevent the bubble from breaking. Therefore, in a slurry with high solid content, the bubbles are relatively stable. Usually, the number of broken bubbles is less compared to the case of low solid content, and the bubble break time is longer compared to the case of low solid content.

[0045] Low solid content: In a low-solid-content slurry, bubbles are less stable and more likely to break. Therefore, although bubbles are easily formed, the breakage rate is high. Usually, the number of broken bubbles is larger than that in the case of high solid content, and the bubble breakage time is shorter than that in the case of high solid content.

[0046] Relationship between the solid content of the slurry and the bubble size:

[0047] High solid content: In a high-solid-content slurry, the movement of bubbles is restricted by particles, and bubbles are not easily merged. Therefore, usually, the bubble size is smaller than that in the low-solid-content state.

[0048] Low solid content: In a low-solid-content slurry, the movement of bubbles is relatively free, and bubbles are more likely to merge into large bubbles. Therefore, usually, the bubble size is larger than that in the high-solid-content state.

[0049] Step S3: Calculate the solid content of the slurry in each area based on the number of bubbles, average bubble size, average bubble breakage time, and the number of broken bubbles, and take the average of the solid contents of all areas as the current reference solid content parameter of the slurry.

[0050] Based on the number of bubbles, size, and bubble breakage time, the solid content of the slurry can be calculated, and the gas flow rate and gas temperature can be adjusted in real time according to the calculated solid content of the slurry. The following steps can be taken:

[0051] Gas flow rate and gas temperature data: Collect gas flow rate and gas temperature data.

[0052] The relationship between the solid content of the slurry and the bubble characteristics is described by the following formula, and the gas flow rate and gas temperature are adjusted according to the change of the solid content. The relationship between the solid content of the slurry and the bubble characteristics in each area can be expressed as:

[0053] c = v 1 *N + v 2 *S + v 3 *T c + v 4 *L

[0054] where v 1 、v 2 、v 3 、v 4 are weight parameters, and the weight parameters can be continuously iteratively updated to reach the optimum. They represent the influence degrees of the number of bubbles, size, breakage time, and the number of broken bubbles on the solid content of the slurry, and the sum is 1. N represents the total number of bubbles within the preset time t a ,S represents the average value of the bubble size, T c represents the average breakage time, L represents the number of broken bubbles, and c represents the solid content of the slurry.

[0055] For each region, the solid content of the slurry is collected to obtain the solid content of the slurry in each region [c 1 , c 2 ,... c n , and the average value of the solid content of the slurry in all regions is taken as the current slurry reference solid content parameter c ave .

[0056] Step S4: Adjust the gas flow rate and gas temperature in real time according to the current slurry reference solid content parameter.

[0057] Monitor the change of solid content in real time: Using the method in step S3, calculate the average value of the solid content of the slurry in all regions as the current slurry reference solid content parameter c ave , and compare it with the set value to obtain the deviation of the solid content. The specific steps are as follows:

[0058] Step s41: Set the target solid content parameter and determine the ideal slurry solid content target value c target , and this target value can be obtained by measuring the results of multiple experiments;

[0059] Step s42: Monitor the current slurry solid content in real time. According to the measurement method in step s3, obtain the current slurry reference solid content parameter c ave ;

[0060] Step s43: Calculate the solid content deviation;

[0061] Calculate the deviation m between the current slurry reference solid content parameter and the target solid content parameter:

[0062] m = c target - c ave

[0063] Step s44: Adjust the gas flow rate and gas temperature.

[0064] First, divide the solid content deviation m into different levels, such as small deviation, medium deviation, and large deviation. Different deviation levels correspond to different adjustment strategies.

[0065] Small deviation: At this time, it means that the current slurry reference solid content parameter meets the standard, and the gas flow rate and gas temperature are not adjusted. The small deviation standard is that the ratio of the deviation m to the solid content target value c target is less than 2%.

[0066] Medium deviation: The medium deviation standard is that the ratio of the deviation m to the solid content target value c targetThe ratio is 2% - 10%. At this time, a fine-tuning strategy is adopted, and the change amounts of the gas flow rate and temperature are controlled within a small range for the purpose of making fine corrections. For example, the adjustment range of the gas flow rate is 5% to 10% of the current flow rate, and the adjustment range of the gas temperature does not exceed 10% of the current temperature.

[0067] At this time, the adjustment strategy can adopt the method of preferentially adjusting the gas flow rate because the adjustment of the flow rate is more concise and takes effect quickly. If the flow rate and temperature are adjusted simultaneously, it is easy to cause the adjustment amplitude to be too large.

[0068] Adjust the gas flow rate: If the deviation is positive (the current reference solid content parameter of the slurry is lower than the target solid content value c target ), preferentially reduce the gas flow rate F to prevent liquid splashing. If the deviation is negative (the current reference solid content parameter of the slurry is higher than the target solid content value c target ): Preferentially increase the gas flow rate F to avoid the accumulation of crystal nuclei and thus reduce the solid content.

[0069] After waiting for a period of time t1, the beating action continues during the waiting process. Repeat steps S1 to S4, and calculate the deviation between the current reference solid content parameter of the slurry and the target solid content value c target again. Preferably, the time t1 is 5s - 10s.

[0070] If the target solid content value c target is still not reached: Adjust the gas temperature Tg according to the direction and magnitude of the deviation. If the deviation is positive, that is, the current reference solid content parameter of the slurry is lower than the target solid content value c target , at this time, reduce the gas temperature Tg; if the deviation is negative, that is, the current reference solid content parameter of the slurry is higher than the target solid content value c target , at this time, increase the gas temperature Tg.

[0071] After waiting for a period of time t2, the beating action continues during the waiting process. Repeat steps S1 to S4, and calculate the deviation between the current reference solid content parameter of the slurry and the target solid content value c target again. Preferably, the time t2 is 10s - 20s because the control method of temperature requires waiting for the temperature change and stability, and the effect is the most accurate within this time period.

[0072] The above steps are repeated until the current reference solid content parameter of the slurry reaches the target solid content value c target or within a small deviation range. The reason for adjusting in this way is that the gas flow rate is easier to control than the temperature. Therefore, the method of changing the flow rate is preferentially used to change the solid content. When the solid content cannot reach the standard only by changing the flow rate, the method of changing the gas temperature is then used to adjust the solid content.

[0073] Large deviation: The large deviation standard is that the ratio of the deviation m to the solid content target value c target is greater than 10%. At this time, a strong adjustment strategy is adopted to quickly adjust the gas flow rate and temperature. To correct the solid content deviation as soon as possible, an adjustment strategy of simultaneously adjusting the gas flow rate and gas temperature is taken. For example, the adjustment range of the gas flow rate is 10%-50% of the current gas flow rate, and the adjustment range of the gas temperature is 10%-50% of the current temperature. The specific adjustment ratio can be set by the ratio of the deviation m to the solid content target value c target When the ratio is in the range of 10-20%, the corresponding adjustment ranges of the gas flow rate and gas temperature are 10-20%; when the ratio is in the range of 20-30%, the corresponding adjustment ranges of the gas flow rate and gas temperature are 20-30%; when the ratio is in the range of 30-40%, the corresponding adjustment ranges of the gas flow rate and gas temperature are 30-40%; when the ratio is greater than 40%, the corresponding adjustment ranges of the gas flow rate and gas temperature are 40-50%.

[0074] After waiting for a period of time t3, during which the pulping action continues, repeat steps S1 to S4, and calculate the deviation between the current pulp reference solid content parameter and the solid content target value c target Preferably, the time t3 is 5s-10s. And the above steps are cycled until the current pulp reference solid content parameter reaches the solid content target value c target or within the small deviation range.

[0075] The beneficial technical effects of the present invention are to provide a control method and system for a gas-induced semi-solid process based on visual recognition. By monitoring the bubble characteristics, calculating the current pulp reference solid content parameter, and adjusting the gas flow rate and gas temperature in real time, the solid content of the pulp can be accurately controlled, the stability and controllability of the production process can be improved, the automatic adjustment of the gas flow rate and gas temperature is realized, no manual intervention is required, the labor cost can be reduced, and the production efficiency can be improved.

[0076] A complete set of processing methods for identifying bubbles is proposed for the unique characteristics of the pulp bubble image, including steps such as threshold segmentation, shape feature extraction, color feature extraction, texture feature extraction, etc. And the circular LBP operator is used to extract texture features according to the shape features of the bubbles, providing comprehensive technical support for the accurate identification and analysis of bubble characteristics. Different adjustment methods are adopted according to different situations of the deviation between the current pulp reference solid content parameter and the target solid content, ensuring the efficiency and accuracy of the adjustment. Description of the Drawings

[0077] Figure 1 It is a flow chart of gas flow rate and temperature feedback regulation;

[0078] Figure 2 It is a flowchart for bubble feature extraction and recognition;

[0079] Figure 3 It is a flowchart for adjusting gas flow rate and gas temperature according to the current reference solid content parameter of the slurry;

[0080] Figure 4 It is a control system architecture diagram. Specific embodiments

[0081] The following further describes the specific embodiments of the present invention in conjunction with the accompanying drawings. It should be noted here that the description of these embodiments is for helping to understand the present invention, but does not constitute a limitation to the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0082] It should be noted that in the description of the present invention, the orientation or positional relationship indicated by terms such as "upper", "lower", "left", "right", "front", "rear", etc. is the description of the structure of the present invention based on the accompanying drawings, and is only for the convenience of describing the present invention simply, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present invention.

[0083] For "first" and "second" in this technical solution, they are only the appellation distinctions for the same or similar structures, or corresponding structures with similar functions, and are not the arrangement of the importance of these structures, nor do they have a sequence, or compare sizes, or other meanings.

[0084] In addition, unless otherwise clearly specified and limited, the terms "installation" and "connection" should be understood in a broad sense. For example, the connection can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and can be the communication inside two structures. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the general idea of the present invention and in connection with the specific situation of this solution.

[0085] Embodiment 1

[0086] As Figure 1 shown, the implementation steps of the present invention are as follows:

[0087] Step S1: Use a camera to monitor the slurry in real time, and continuously collect slurry liquid surface image information frames at the same interval within a preset time t a and perform frame-by-frame analysis on the obtained slurry liquid surface image information to identify the bubbles in the image.

[0088] Preferably, the time t a is 10 s - 20 s; the acquisition interval can be set to 0.5 s - 1 s.

[0089] Camera selection: Select a camera suitable for the process requirements. Generally, factors such as resolution, frame rate, and adaptability need to be considered.

[0090] Image acquisition: Install the camera and adjust its position so that it can clearly capture the slurry liquid level. Ensure that the field of view of the camera covers the entire slurry liquid level and can stably acquire images.

[0091] Image preprocessing: Preprocess the acquired images, including removing noise, enhancing contrast, etc., to obtain preprocessed images, so as to improve the accuracy and efficiency of subsequent image processing.

[0092] Image analysis: Using image processing technology, divide the slurry liquid level into multiple regions and identify the bubble conditions in each region. For the unique image features of the identified slurry liquid level in this case, the present invention proposes an image analysis method applicable to the slurry liquid level. The image analysis method is as Figure 2 shown below:

[0093] Step s11, threshold segmentation.

[0094] Adopt a suitable threshold segmentation algorithm (such as the Otsu algorithm, adaptive threshold, etc.) to convert the acquired slurry liquid level image into a binary image, which is convenient for detecting the bubble contour.

[0095] Step s12, shape feature extraction.

[0096] The bubble regions in the image are usually irregular circles and are connected. Each connected domain in the binary image is a potential bubble region. Use an image edge detection algorithm (such as the Canny algorithm) to extract the contour of the potential bubble region in the binary image, and then calculate the shape features of the contour, such as Hu moments or Zernike moments, to obtain the shape features of the potential bubble region.

[0097] Step s13, color feature extraction.

[0098] In the acquired image, there is an obvious color difference between the color of the bubbles and the color of the slurry. Multiply the binary image obtained in step s11 by the preprocessed image, and the resulting image is the color image of the potential bubble region. For each connected domain in the color image, its color features can be extracted, which are the color features of the potential bubble region. Each potential bubble region can be converted to a suitable color space (such as HSV, LAB, etc.), and the color features of the potential bubble region can be extracted. Methods such as color histogram or color moment can be used.

[0099] Step s14: Extract the texture features of the bubble region using the circular LBP operator.

[0100] In the acquired image, the texture features of the bubbles are different from those of the slurry. Multiply the binary image obtained in step s11 by the preprocessed image, and the resulting image is the color map of the potential bubble region. For each connected component in the color map, its texture features can be extracted. Methods such as Local Binary Pattern (LBP) can be used to extract the texture features of the bubble surface. Since bubbles are mostly circular or elliptical in shape, the circular LBP operator is used here to extract the texture features to improve the recognition rate.

[0101] The calculation formula of circular LBP is as follows. Assume that we select a circular neighborhood with a radius of R, with the central pixel as the center, and evenly select O sampling points on the circumference:

[0102] Calculate the LBP value: For each sampling point, compare its gray value with the gray value of the central pixel. If the gray value of the sampling point is greater than or equal to the gray value of the central pixel, the weight value at this position is 1, otherwise it is 0. Connect the obtained O weight values in clockwise or counterclockwise order to form a binary number, and this binary number is the LBP value of the central pixel.

[0103] Formula representation: Assume that the gray value of the central pixel is P, and the gray value of the i-th sampling point is P i , then the LBP value LBP P,R of the central pixel can be expressed as:

[0104]

[0105] where s(x) is the sign function. If x≥0, then s(x)=1, otherwise s(x)=0. P i -P represents the gray value difference between the i-th sampling point and the central pixel.

[0106] Through the above formula, the LBP value of each pixel in the image can be calculated, thus realizing the description and analysis of the image texture.

[0107] Step s15: Feature fusion and recognition.

[0108] Fuse the extracted color, shape, and texture features. The various features can be fused into a comprehensive feature vector using weighted summation or feature concatenation. Feature fusion is to combine the multiple features obtained from different feature extraction steps (color, shape, texture) to generate a comprehensive feature vector. This can make full use of the advantages of various features and improve the accuracy and robustness of bubble recognition.

[0109] Basic principle of weighted summation: Each eigenvector is weighted and summed according to a certain weight to generate a new eigenvector. Normalize each eigenvector to ensure that the dimensions of each feature are consistent. Determine the weight of each feature, which can be set according to experience or learned through training data.

[0110] F 融合 = w 1 * F 颜色 + w 2 * F 形状 + w 3 * F 纹理

[0111] Among them, F 融合 is the fused eigenvector, w 1 , w 2 , w 3 are the weights of each feature, and their sum is 1. Initially, each weight can be set to 1 / 3. F 颜色 、F 形状 、F 纹理 are the color feature, shape feature, and texture feature extracted using the circular LBP operator respectively.

[0112] Through the feature fusion method, multiple features such as color, shape, and texture can be effectively combined to extract the fused bubble features and generate a comprehensive eigenvector containing rich information. Input the comprehensive eigenvector into the trained classifier. The training of the classifier is also completed based on the extracted fused bubble features, and the bubbles in the collected images can be recognized.

[0113] Through the above steps, real-time monitoring of the slurry liquid level and acquisition of image information can be achieved, providing basic data and support for subsequent process control.

[0114] Step S2: Divide the slurry liquid level into multiple regions, identify and record the number of bubbles, bubble size, number of broken bubbles, and their breaking time in each region, and calculate the average bubble size and average breaking time.

[0115] Region division: According to actual requirements and process characteristics, divide the slurry liquid level into multiple regions [a 1 , a 2 ,... a n . It can be divided according to image features, bubble distribution, etc., ensuring that the size of each region is appropriate and can contain enough bubbles.

[0116] Bubble recognition: Based on the characteristics of bubbles, the fused bubble features obtained in step S1 are used to recognize the bubbles in each region, and the recognized bubbles are tagged for easy distinction and statistics. For example, the recognized bubbles are marked as bubble 1, bubble 2... bubble n.

[0117] Bubble quantity statistics: The recognized bubbles are counted, and the total number of bubbles within the preset time t in each region is recorded. a within.

[0118] Bubble size measurement: The recognized bubbles 1, 2... n are measured for size, and the size information of the bubbles in each region is recorded. The bubble size is obtained by pixel counting or measuring the area of the bubble connected domain, and its average value is taken. The average value calculation method is the total pixel count of all bubble regions or the total connected domain area of all measured bubbles, divided by the number of recognized bubbles.

[0119] Record of the number of broken bubbles: When bubbles 1, 2... n are recognized, when a bubble breaks, it is marked as a broken bubble. For example, when bubble 1 breaks, bubble 1 is marked as a broken bubble, and the number of broken bubbles is incremented by 1; if bubble 2 does not break, it is not marked; when bubble 3 breaks, bubble 3 is marked as a broken bubble, and the number of broken bubbles is incremented by 1 again, and so on. The number of broken bubbles within the preset time t in each region is recorded. a within.

[0120] Record of the bubble break time: The bubble break time refers to the time elapsed from the formation of the bubble to its final rupture and disappearance. In a liquid, a bubble gradually grows over time until it reaches a certain size and then ruptures and disappears. The bubble break time can reflect the stability and durability of the bubble, and the break time point can be determined by monitoring the change or disappearance of the bubble morphology. When a bubble is formed, taking bubble 1 as an example, the time point of its appearance is recorded as the formation time of bubble 1; when the bubble 1 ruptures and disappears, the time point of its rupture is recorded as the break time of bubble 1. The break time of bubble 1 is the break time point of bubble 1 minus the formation time point of bubble 1, and this process is carried out for all bubbles. The average break time of all broken bubbles is recorded to eliminate the influence of individual bubble instability.

[0121] The relationship between the solid content of the slurry and the above various parameters is as follows:

[0122] Relationship between the solid content of the slurry and the number of bubbles:

[0123] High solid content: The higher the solid content of the slurry, the higher the viscosity of the slurry usually is. In a slurry with high viscosity, it is more difficult for bubbles to form because gas diffuses slowly in a high-viscosity environment and the resistance to bubbles is greater. Therefore, in a slurry with high solid content, the number of bubbles will decrease.

[0124] Low solid content: When the solid content of the slurry is low, the viscosity is low, and gas is more likely to diffuse in the slurry and form bubbles. Therefore, in the slurry with low solid content, the number of bubbles will increase.

[0125] Relationship between the solid content of the slurry and the stability and break-up time of bubbles:

[0126] High solid content: The particles in the high-solid-content slurry increase the stability of the bubbles because the solid particles can form a stable interface on the bubble surface to prevent the bubbles from breaking. Therefore, in the high-solid-content slurry, the bubbles are relatively stable. Generally, the number of broken bubbles is less than that in the case of low solid content, and the bubble break-up time is longer than that in the case of low solid content.

[0127] Low solid content: In the low-solid-content slurry, the bubbles are less stable and easier to break. Therefore, although the bubbles are easy to form, the break-up rate is high. Generally, the number of broken bubbles is more than that in the case of high solid content, and the bubble break-up time is shorter than that in the case of high solid content.

[0128] Relationship between the solid content of the slurry and the bubble size:

[0129] High solid content: In the high-solid-content slurry, the movement of the bubbles is restricted by the particles, and the bubbles are not easy to merge. Therefore, generally, the bubble size is smaller than that in the low-solid-content state.

[0130] Low solid content: In the low-solid-content slurry, the bubbles move relatively freely, and the bubbles are more likely to merge into large bubbles. Therefore, generally, the bubble size is larger than that in the high-solid-content state.

[0131] Step S3: Calculate the solid content of the slurry in each area according to the number of bubbles, average bubble size, average bubble break-up time, and the number of broken bubbles, and take the average value of the solid contents of all areas as the current reference solid content parameter of the slurry.

[0132] According to the number of bubbles, size, and bubble break-up time, the solid content parameter of the slurry can be calculated, and the gas flow rate and gas temperature can be adjusted in real time according to the calculated solid content parameter of the slurry. The following steps can be taken:

[0133] Gas flow rate and gas temperature data: Collect the gas flow rate and gas temperature data.

[0134] The relationship between the solid content of the slurry and the bubble characteristics is described by the following formula, and the gas flow rate and gas temperature are adjusted according to the change of the solid content. The relationship between the solid content of the slurry and the bubble characteristics in each area can be expressed as:

[0135] c = v 1 *N + v2 *S + v 3 *T c +v 4 *L

[0136] Among them, v 1 , v 2 , v 3 , v 4 are weight parameters, and the weight parameters can be continuously iteratively updated according to the beating process to achieve the optimal value. They represent the influence degrees of the number of bubbles, the bubble size, the breaking time, and the number of broken bubbles on the solid content of the slurry, and the sum is 1. N represents the total number of bubbles within the preset time t a , S represents the average value of the bubble size, T c represents the average breaking time, L represents the number of broken bubbles, and c represents the solid content of the slurry.

[0137] For each region, the solid content of its slurry is collected to obtain the solid content of the slurry in each region [c 1 , c 2 ,... c n , and the average value of the solid content of the slurry calculated for all regions is taken as the current slurry reference solid content parameter c ave .

[0138] Step S4: Adjust the gas flow rate and gas temperature in real time according to the current slurry reference solid content parameter.

[0139] The adjustment logic of the gas flow rate and gas temperature is as Figure 3 shown.

[0140] Monitor the change of solid content in real time: Using the method in step S3, calculate the average value of the solid content of the slurry in all regions as the current slurry reference solid content parameter c ave , and compare it with the set value to obtain the deviation of the solid content. The specific steps are as follows:

[0141] Step s41: Set the target solid content parameter and determine the ideal target value of the slurry solid content c target , and this target value can be obtained by measuring the results of multiple experiments;

[0142] Step s42: Monitor the current slurry solid content in real time, and according to the measurement method in step s3, obtain the current slurry reference solid content parameter c ave ;

[0143] Step s43: Calculate the solid content deviation;

[0144] Calculate the deviation m between the current slurry reference solid content parameter and the target solid content parameter:

[0145] m = c target - cave

[0146] Step s44: Adjust the gas flow rate and gas temperature.

[0147] First, divide the solid content deviation m into different levels, such as small deviation, medium deviation, and large deviation. Different deviation levels correspond to different adjustment strategies for gas flow rate and gas temperature.

[0148] Small deviation: At this time, it represents that the current slurry reference solid content parameter meets the standard. The small deviation standard is that the ratio of the deviation m to the solid content target value c target is less than 2%.

[0149] Medium deviation: The medium deviation standard is that the ratio of the deviation m to the solid content target value c target is 2% - 10%. At this time, a fine-tuning strategy is adopted, and the change amounts of the gas flow rate and temperature are controlled within a small range. The purpose is to make fine corrections. For example, the adjustment range of the gas flow rate is 5% to 10% of the current flow rate, and the adjustment range of the gas temperature does not exceed 10% of the current temperature.

[0150] The adjustment strategy at this time can adopt the method of preferentially adjusting the gas flow rate because the adjustment of the flow rate is more concise and takes effect faster. If the flow rate and temperature are adjusted simultaneously, it is easy to cause the adjustment amplitude to be too large.

[0151] Adjust the gas flow rate: If the deviation is positive, that is, the current slurry reference solid content parameter is lower than the solid content target value c target , preferentially reduce the gas flow rate F to prevent liquid splashing. If the deviation is negative, that is, the current slurry reference solid content parameter is higher than the solid content target value c target , preferentially increase the gas flow rate F to avoid the accumulation of crystal nuclei and thus reduce the solid content.

[0152] After waiting for a period of time t1, during which the beating action continues, repeat steps S1 to S4, and calculate the deviation between the current slurry reference solid content parameter and the solid content target value c target again. Preferably, the time t1 is 5s - 10s.

[0153] If the solid content target value c target has still not been reached: Adjust the gas temperature Tg according to the direction and magnitude of the deviation. If the deviation is positive, that is, the current slurry reference solid content parameter is lower than the solid content target value c target , then reduce the gas temperature Tg at this time; if the deviation is negative, that is, the current slurry reference solid content parameter is higher than the solid content target value c target , then increase the gas temperature Tg at this time.

[0154] After waiting for a period of time t2, during which the beating action continues, repeat steps S1 to S4, and calculate again the deviation between the current pulp reference solids content parameter and the target solids content value c target Preferably, the time t2 is 10s - 20s. Since the temperature control method requires waiting for the temperature to change and stabilize, the effect is the most accurate within this time period.

[0155] The above steps are repeated until the current pulp reference solids content parameter reaches the target solids content value c target or within a small deviation range. The reason for adjusting in this way is that the gas flow rate is easier to control compared to the temperature. Therefore, the flow rate is preferentially changed to change the solids content. When the solids content cannot reach the standard only by changing the flow rate, the gas temperature is then changed to adjust the solids content.

[0156] Large deviation: The large deviation standard is that the ratio of the deviation m to the target solids content value c target is greater than 10%. At this time, a strong adjustment strategy is adopted to quickly adjust the gas flow rate and temperature. To correct the solids content deviation as soon as possible, an adjustment strategy of simultaneously adjusting the gas flow rate and the gas temperature is taken. If the deviation is positive, reduce the gas flow rate F and simultaneously reduce the gas temperature Tg. If the deviation is negative, increase the gas flow rate F and simultaneously increase the gas temperature Tg.

[0157] For example, the adjustment range of the gas flow rate is 10% - 50% of the current gas flow rate, and the adjustment range of the gas temperature is 10% - 50% of the current temperature. The specific adjustment ratio can be set by the ratio of the deviation m to the target solids content value c target When the ratio is in the range of 10 - 20%, the corresponding adjustment ranges of the gas flow rate and the gas temperature are 10 - 20%; when the ratio is in the range of 20 - 30%, the corresponding adjustment ranges of the gas flow rate and the gas temperature are 20 - 30%; when the ratio is in the range of 30 - 40%, the corresponding adjustment ranges of the gas flow rate and the gas temperature are 30 - 40%; when the ratio is greater than 40%, the corresponding adjustment ranges of the gas flow rate and the gas temperature are 40 - 50%.

[0158] After waiting for a period of time t3, during which the beating action continues, calculate again the current pulp reference solids content parameter and the deviation from the target solids content value. Preferably, the time t3 is 5s - 10s. And the above steps are repeated until the current pulp reference solids content parameter reaches the target solids content value c target or within a small deviation range.

[0159] Example 2

[0160] Such as Figure 4As shown, the gas-induced semi-solid process control system 100 based on visual recognition includes a camera module 10 for real-time monitoring of the slurry liquid level and acquisition of image information; an image processing unit 20 for identifying bubble features in the manner of the said steps S1 - S2; a data processing unit 30 for calculating the current reference solid content parameter of the slurry in the manner of step S3; and a control unit 40 for adjusting the gas flow rate and gas temperature in the manner of step S4.

[0161] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the described embodiments. For those skilled in the art, without departing from the principle and spirit of the present invention, various changes, modifications, substitutions, and variations made to these embodiments still fall within the protection scope of the present invention.

Claims

1. A gas-induced semi-solid process control method based on visual recognition, characterized in that: The method comprises the following steps: Step S1: Use a camera to monitor the slurry in real time. a Continuously collecting slurry liquid surface image information frames at the same interval, and analyzing the acquired slurry liquid surface image information frame by frame to identify bubbles in the image; Step S1 includes: Step s11: threshold segmentation, using a threshold segmentation algorithm to convert the image into a binary image; Step s12: shape feature extraction, each connected domain in the binary image is a potential bubble area, and the contour of the potential bubble area in the binary image is extracted using an image edge detection algorithm, and then the shape feature of the contour is calculated; Step s13: color feature extraction, multiplying the binary image obtained in step s11 with the preprocessed image to obtain a color map of the potential bubble area, and extracting the color features of the potential bubble area in the color map; Step s14: using a circular LBP operator to extract texture features of the bubble area, multiplying the binary image obtained in step s11 with the preprocessed image to obtain a color map of the potential bubble area, and using a circular LBP operator to extract texture features of the potential bubble area in the color map; Step s15: feature fusion and identification, fusing the extracted multiple features to obtain fused bubble features and generate a comprehensive feature vector; Step S2: Divide the slurry surface into multiple areas, and identify and record the number of bubbles, bubble size, number of broken bubbles and breaking time in each area, and calculate the average bubble size and average breaking time; Step S3: Calculate the slurry solid content in each area according to the number of bubbles, average bubble size, average bubble breakage time, and number of broken bubbles, and take the average value of the solid content of all areas as the current slurry reference solid content parameter; Step S4: adjusting the gas flow rate and gas temperature in real time according to the current slurry reference solid content parameters.

2. The method according to claim 1, characterized in that The preset time is 10 seconds to 20 seconds, and the collection interval is 0.5 seconds to 1 second.

3. The method according to claim 1, characterized in that Step S2 further comprises the following steps: The slurry surface is divided into multiple regions [a1, a2, ...a n ], the method of extracting fused bubble features is used to realize the recognition of bubbles in each area, and the recognized bubbles are labeled; the total number of bubbles in each area within the preset time is recorded; the size information of the bubbles in each area is recorded, and the average value is taken; the number of broken bubbles in each area is recorded; the formation time and breakage time of the bubbles are recorded, and the average breakage time of all broken bubbles is calculated.

4. The method according to claim 1, characterized in that: The calculation formula in step S3 is as follows: c=v1*N+v2*S+v3*T c +v4*L Among them, v1, v2, v3, and v4 are weight parameters, the sum of which is 1, and N represents the preset time t a The total number of bubbles in the sample, S represents the average bubble size, T c represents the average bubble breaking time, L represents the number of broken bubbles, and c represents the solid content of the slurry; The slurry solid content of each area is collected to obtain the slurry solid content of each area [c1, c2, ... c n ], and take the average value of the slurry solid content in all areas as the current slurry reference solid content parameter c ave .

5. The method according to claim 4, characterized in that Step S4 further comprises the following steps: Set the target value of slurry solid content c target , calculate the current slurry reference solid content parameter c ave , calculate the solid content target value c target With the current slurry reference solid content parameter c ave The deviation between .

6. The method according to claim 5, characterized in that According to the deviation, setting adjustment strategies for different deviation levels, including small deviation, medium deviation and large deviation; The small deviation standard is the difference between the deviation and the solid content target value c target The ratio of the deviation to the solid content target value c is less than 2%, and the medium deviation standard is the deviation to the solid content target value c target The ratio of the above deviation is 2%-10%, and the above large deviation standard is the ratio of the above deviation to the above solid content target value c target The ratio is greater than 10%.

7. The method according to claim 6, characterized in that When the deviation is small, the gas flow rate and gas temperature are not adjusted; When the deviation is medium, the gas flow rate is first adjusted within a range of 5% to 10% of the current flow rate. After waiting for a period of time t1, which is 5s-10s, steps S1 to S4 are repeated to calculate the current slurry reference solid content parameter and the solid content target value c again. target If the solid content target value c is still not reached, target Or the small deviation standard, adjust the gas temperature Tg, the adjustment range is within 10% of the current temperature, wait for a period of time t2, the time t2 is 10s-20s, repeat steps S1 to S4, and calculate the current slurry reference solid content parameter and the solid content target value c again target The deviation between When the deviation is large, adjust the gas flow rate and gas temperature at the same time. The adjustment range of the gas flow rate is 10%-50% of the current gas flow rate, and the adjustment range of the gas temperature is 10%-50% of the current temperature. After waiting for a period of time t3, repeat steps S1 to S4 to calculate the current slurry reference solid content parameters and the solid content target value c again. target The deviation between .

8. The method according to claim 7, characterized in that When the deviation is a large deviation, if the deviation is positive, the gas flow rate F is reduced while the gas temperature Tg is reduced, and if the deviation is negative, the gas flow rate F is increased while the gas temperature Tg is increased.

9. A gas-induced semi-solid process control system based on visual recognition, comprising: A camera module is used to monitor the slurry level in real time and collect image information; An image processing unit, used for identifying bubbles by adopting the method of step S1 or step S2 in any one of claims 1 to 8; A data processing unit, used to calculate the current slurry reference solid content parameter by the method of step S3 described in any one of claims 1 to 8; A control unit, for adjusting the gas flow rate and the gas temperature in the manner of step S4 as described in any one of claims 1 to 8.

Images