A method, system and storage medium for predicting pulp liquid level based on dynamic characteristics of flotation foam

Abstract

The present invention provides a method, system and storage medium for predicting the pulp level based on the dynamic characteristics of flotation foam, which relates to the technical field of production detection in the flotation process. Under the working conditions of the flotation machine, the present invention continuously collects flotation foam images and pulp levels; extracts texture features from each flotation foam image to obtain texture features, and generates a texture distribution coefficient for the texture features related to the pulp level; uses the Sobel operator to identify the edge regions of each flotation foam in the flotation foam image, and conducts comprehensive analysis to generate a foam distribution index; analyzes two adjacent flotation foam images to generate a dynamic deviation degree of the flotation foam; constructs a pulp level prediction model, and uses the pulp level prediction model to predict the pulp level. It solves the technical problem that the prior art cannot monitor the pulp level in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of production detection in the flotation process, and specifically provides a method, a system and a storage medium for predicting the pulp level based on the dynamic characteristics of flotation foam. Background Art

[0002] Mineral flotation is a method of adding flotation reagents to pulp under specific process conditions, filling with air, and generating a large number of bubbles through stirring, and then recovering the foam containing useful minerals to achieve the beneficiation goal. In the traditional flotation control process, the control of the pulp level usually depends on manual experience. Although this method can meet the requirements to a certain extent, its accuracy is insufficient. Although the pulp level can also be accurately obtained through a level sensor, the deployment of the level sensor requires a large amount of cost. Moreover, due to the complex chemical reactions involved in the flotation process, even if a large amount of cost is paid for the deployment of the level sensor, it is extremely easy to be damaged. Therefore, in industrial production, level sensors are generally not deployed in general flotation machines. Due to the limitations of on-site conditions, there is no good method to solve the problem of real-time monitoring of the pulp level.

[0003] Flotation foam has characteristics such as a large number, adhesion, mixing, and irregular shape. Although these foam characteristics are highly related to the height of the pulp level, they are difficult to quantitatively describe. Moreover, the foam characteristics are complex and diverse. It is difficult to determine which specific characteristics have sufficient correlation with the pulp level, and these characteristics do not have a simple linear relationship with the pulp level. How to establish a connection between these foam characteristics and the pulp level is also a major technical difficulty.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and therefore it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a method and a system for predicting the pulp level based on the dynamic characteristics of flotation foam to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A method for predicting the pulp level based on the dynamic characteristics of flotation foam, the specific steps including:

[0008] Step 1: Continuously collect flotation foam images and corresponding pulp levels in different frames during the operation of a flotation machine under known working conditions; the working conditions of the flotation machine include the scraper speed, the air intake, and the slurry concentration;

[0009] Step 2: Use the gray-level co-occurrence matrix to extract texture features from each flotation foam image to obtain texture features, analyze the correlation between each texture feature and the pulp level, and generate a texture distribution coefficient based on the texture features related to the pulp level;

[0010] Step 3: Use the Sobel operator to identify the edge region of each flotation foam in the flotation foam image, determine the flotation foam region based on the edge region of each flotation foam, and comprehensively analyze the flotation foam region and the edge region of the flotation foam to generate a foam distribution index;

[0011] Step 4: Calculate the dynamic deviation degree of the flotation foam in the flotation foam image at the current frame based on the texture distribution coefficient and the foam distribution index of the current frame and the previous frame of flotation foam images;

[0012] Step 5: Use the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame as the model input, and the corresponding pulp level as the label, construct and train a pulp level prediction model, obtain the model input at the current moment, and input it into the pulp level prediction model to predict the pulp level.

[0013] Furthermore, the texture features include contrast, dissimilarity, homogeneity, energy, and entropy.

[0014] The specific formula for calculating the correlation between each texture feature and the pulp level is:

[0015] ;

[0016] Where is the correlation between the th texture feature and the pulp level, is the eigenvalue of the th texture feature of the th frame of flotation foam image, is the total number of frames of the collected flotation foam images, is the average eigenvalue of the th texture feature of the collected flotation foam images, is the average value of the pulp levels of the collected flotation foam images, is the pulp level of the th frame of flotation foam image, is the frame index of the flotation foam image, and , is the index of the texture feature, and .

[0017] Furthermore, a preset texture feature correlation threshold is set. If , it is determined that the corresponding texture feature is related to the pulp level, and this texture feature is defined as an associated texture feature. If , it is determined that the corresponding texture feature has no association with the pulp level;

[0018] The specific formula for generating the texture distribution coefficient is:

[0019] ;

[0020] ;

[0021] Among them, is the texture distribution coefficient of the th frame of the flotation foam image, is the weight of the th key texture feature, is the th key texture feature value in the th frame of the flotation foam image.

[0022] Furthermore, the horizontal direction gradient value and the vertical direction gradient value of each pixel point in the flotation foam image are calculated using the Sobel operator, and a first gradient amplitude data set including all pixel point gradient amplitude data is generated, and the error rate between the gradient amplitude of the current pixel point and the gradient amplitude of the adjacent pixel point is judged. If the error rate is less than the set error rate threshold, the pixel value of the current pixel point is retained, otherwise it is set to 0. The part where the pixel value is retained is defined as the edge area of the flotation foam, and the closed area wrapped by the edge area is defined as the flotation foam part. The set of all flotation foam positions is the flotation foam area.

[0023] Furthermore, when calculating the horizontal direction gradient value and the vertical direction gradient value of each pixel point, the gray value of the pixel point and the gray values of its adjacent pixel points are multiplied by the horizontal direction gradient template of the Sobel operator, and the results of all multiplications are added to obtain the horizontal direction gradient value of the pixel. The gray value of the pixel point and the gray values of its adjacent pixel points are multiplied by the vertical direction gradient template of the Sobel operator, and the results of all multiplications are added to obtain the vertical direction gradient value of the pixel. The formulas for calculating the horizontal direction gradient value and the vertical direction gradient value are:

[0024] ;

[0025] ;

[0026] Among them, and are the th row and the The horizontal gradient value and vertical gradient value of column pixels; is the row and the gray value of the column pixel, , is the index of the pixel of the flotation bubble image;

[0027] The formula for generating the gradient amplitude of the pixel is:

[0028] ;

[0029] Among them, is the row and the gradient amplitude of the column pixel.

[0030] The formula for generating the error rate between the gradient amplitude of the current pixel and the gradient amplitude of the adjacent pixel is:

[0031] ;

[0032] Among them, is the average gradient amplitude of all pixels adjacent to the row and the column pixel, is the row and the error rate of the column pixel.

[0033] Furthermore, the specific logic for generating the foam distribution index is as follows: Count the number of pixels occupied by the flotation foam area, and divide the number of pixels occupied by the flotation foam area by the total number of pixels in the flotation foam image to obtain the flotation foam occupancy ratio; Count the total number of flotation foams in each flotation foam image. For each flotation foam, draw the minimum circumscribed circle of the flotation foam. Starting from the overlapping position of the flotation foam and the minimum circumscribed circle, calculate the distance between the flotation foam and the minimum circumscribed circle along the same radius of the flotation foam every clockwise, which is recorded as the deviation distance. Calculate the variance of the deviation distances of the same flotation foam, calculate the area ratio of the flotation foam to the corresponding minimum circumscribed circle as the area similarity, generate the shape regularity index based on the variance of the deviation distance and the area similarity, and calculate the average value of the shape regularity indexes of each flotation foam in the same flotation foam image; Divide the flotation foam image into M regions evenly, count the number of flotation foams in each region, calculate the variance of the number of flotation foams in the same flotation foam image, which is recorded as the foam distribution uniformity; Generate the foam distribution index based on the total number of flotation foams, the flotation foam occupancy ratio, the average value of the shape regularity index, and the foam distribution uniformity;

[0034] The specific formula for calculating the shape regularity index is:

[0035] ;

[0036] In fact, is the shape regularity index, is the area similarity, is the variance of the deviation distance;

[0037] The specific formula for generating the foam distribution index is:

[0038] ;

[0039] Among them, is the foam distribution index, is the total number of flotation foams, is the proportion of flotation foams, is the average value of the shape regularity index, is the foam distribution uniformity;

[0040] Furthermore, the specific formula for calculating the dynamic deviation degree of flotation foams is:

[0041] ;

[0042] Among them, is the dynamic deviation degree of flotation foams in the th frame of the flotation foam image, is the texture distribution coefficient of the th frame of the flotation foam image, is the texture distribution coefficient of the th frame of the flotation foam image, is the foam distribution index of the th frame of the flotation foam image; is the foam distribution index of the th frame of the flotation foam image.

[0043] The present invention further provides a pulp liquid level system based on the dynamic characteristics of flotation foams. The system is used to implement the pulp liquid level prediction method based on the dynamic characteristics of flotation foams, and specifically includes:

[0044] A data acquisition module, which is used to continuously collect flotation foam images and corresponding pulp liquid levels in different frames during the operation of a flotation machine under known working conditions; the working conditions of the flotation machine include the scraper speed, the air intake, and the slurry concentration;

[0045] A texture analysis module, which uses a gray-level co-occurrence matrix to extract texture features from each flotation foam image to obtain texture features, analyzes the correlation between each texture feature and the pulp liquid level, and generates a texture distribution coefficient according to the texture features related to the pulp liquid level;

[0046] A foam analysis module, which is used to identify the edge region of each flotation foam in the flotation foam image by using the Sobel operator, determine the flotation foam region based on the edge region of each flotation foam, and comprehensively analyze the flotation foam region and the edge region of the flotation foam to generate a foam distribution index;

[0047] A dynamic analysis module, which is used to calculate the dynamic deviation degree of the flotation foam in the flotation foam image at the current frame based on the texture distribution coefficient and the foam distribution index of the current frame and the previous frame of the flotation foam image;

[0048] A final prediction module, which is used to construct and train a pulp level prediction model with the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame as the model input and the corresponding pulp level as the label, obtain the model input at the current moment, and input it into the pulp level prediction model to predict the pulp level.

[0049] The present invention further provides a computer-readable storage medium, and a computer program executable by a processor is stored inside the storage medium. When the computer program is executed by the processor, the pulp level prediction method based on the dynamic characteristics of the flotation foam can be realized.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] The present invention quantitatively describes many characteristics of the flotation foam through image processing technology, not only overcomes the problem that the foam morphology is diverse and difficult to quantitatively describe, but also provides a solid data basis for subsequent feature correlation analysis. By calculating the correlation coefficient between each texture feature and the pulp level, the key features highly correlated with the pulp level are effectively screened out. Ensure that the extracted and analyzed flotation foam features have sufficient relevance to the pulp level, and provide a scientific theoretical basis for the prediction of the flotation foam.

[0052] The present invention not only considers the static image features of the flotation foam, but also quantifies the dynamic features of the flotation foam through the analysis of adjacent frame images. The complex relationship model between the foam features and the pulp level is successfully established through a deep learning model for the static and dynamic features, thereby greatly improving the prediction accuracy and robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the overall method flow of the present invention.

[0054] Figure 2 It is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0055] To make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with specific embodiments.

[0056] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those with ordinary skills in the field to which the present invention pertains. The "first", "second", and similar terms used in the present invention do not denote any order, quantity, or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left", "right", etc. are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0057] Embodiment:

[0058] Please refer to Figure 1 , the present invention provides a technical solution:

[0059] A method for predicting the pulp level based on the dynamic characteristics of flotation foam, the specific steps include:

[0060] Step 1: During the operation of the flotation machine under known working conditions, continuously collect flotation foam images and corresponding pulp levels in different frames; the working conditions of the flotation machine include the scraper speed, air intake, and slurry concentration;

[0061] The pulp level is obtained by an ultrasonic level sensor.

[0062] Step 2: Use the gray-level co-occurrence matrix to extract texture features from each flotation foam image to obtain texture features, analyze the correlation between each texture feature and the pulp level, and generate a texture distribution coefficient according to the texture features related to the pulp level;

[0063] The texture features include contrast, dissimilarity, homogeneity, energy, and entropy. The flotation foam image is grayscale processed to obtain the gray-level co-occurrence matrix of the flotation foam image. The gray-level co-occurrence matrix of the flotation foam image is expressed as follows:

[0064] ;

[0065] Among them, is the gray-level co-occurrence matrix, is the sum of pixels with gray level in the flotation foam image and gray level The number of times adjacent pixels pair up, and are both gray levels, and , .

[0066] ;

[0067] ;

[0068] ;

[0069] ;

[0070] ;

[0071] ;

[0072] Among them, is the normalized , is the correction factor. To avoid , its value is very small and can be a decimal such as 0.0001, 0.001, etc. There is no restriction here, is the contrast of the gray level co-occurrence matrix, is the dissimilarity of the gray level co-occurrence matrix, is the homogeneity of the gray level co-occurrence matrix; is the energy of the gray level co-occurrence matrix, is the entropy of the gray level co-occurrence matrix, is the element in the th row and

[0073] The specific formula for calculating the correlation between each texture feature and the pulp level is:

[0074] ;

[0075] Among them, is the correlation between the rd texture feature and the pulp level, is the eigenvalue of the th texture feature of the th frame of the flotation foam image, is the total number of frames of the collected flotation foam images, is the average eigenvalue of the th texture feature of the collected flotation foam images, is the average value of the pulp levels of the collected flotation foam images, is the The pulp level of the frame flotation foam image is the frame index of the flotation foam image, and , is the index of the texture feature, and .

[0076] The preset texture feature correlation threshold , if , it is determined that the corresponding texture feature is related to the pulp level, and the texture feature is defined as the associated texture feature. If , it is determined that the corresponding texture feature has no association with the pulp level;

[0077] The specific formula for generating the texture distribution coefficient is:

[0078] ;

[0079] ;

[0080] Among them, is the texture distribution coefficient of the -th frame flotation foam image, is the weight of the -th key texture feature, is the -th key texture feature value in the -th frame flotation foam image.

[0081] The texture distribution coefficient reflects the comprehensive texture distribution of the flotation foam image. The flotation foam image plays an important role in analyzing the pulp level, and the texture distribution of the flotation foam is an important property of the flotation foam image and also plays an important role in analyzing the pulp level.

[0082] Step 3: Use the Sobel operator to identify the edge region of each flotation foam in the flotation foam image, determine the flotation foam region based on the edge region of each flotation foam, and comprehensively analyze the flotation foam region and the edge region of the flotation foam to generate a foam distribution index;

[0083] Use the Sobel operator to calculate the horizontal direction gradient value and the vertical direction gradient value of each pixel point in the flotation foam image, and generate a first gradient amplitude data set including all pixel point gradient amplitude data, and judge the error rate between the gradient amplitude of the current pixel point and the gradient amplitude of the adjacent pixel point. If the error rate is less than the set error rate threshold, retain the pixel value of the current pixel point, otherwise set it to 0. The part where the pixel value is retained is defined as the edge region of the flotation foam, and the closed region wrapped by the edge region is defined as the flotation foam part. The set of all flotation foam positions is the flotation foam region.

[0084] When calculating the horizontal direction gradient value and vertical direction gradient value of each pixel point, multiply the gray value of the pixel point and the gray values of its adjacent pixel points by the horizontal direction gradient template of the Sobel operator, and add up the results of all multiplications to obtain the horizontal direction gradient value of the pixel. Multiply the gray value of the pixel point and the gray values of its adjacent pixel points by the vertical direction gradient template of the Sobel operator, and add up the results of all multiplications to obtain the vertical direction gradient value of the pixel. The formulas for calculating the horizontal direction gradient value and vertical direction gradient value are as follows:

[0085] ;

[0086] ;

[0087] Among them, and are the horizontal direction gradient value and vertical direction gradient value of the pixel point in the th row and th column respectively; is the gray value of the pixel point in the th row and th column, , are the indexes of the pixel points of the flotation bubble image;

[0088] The formula for generating the gradient magnitude of the pixel point is as follows:

[0089] ;

[0090] Among them, is the gradient magnitude of the pixel point in the th row and th column.

[0091] The formula for generating the error rate between the gradient magnitude of the current pixel point and the gradient magnitudes of adjacent pixel points is as follows:

[0092] ;

[0093] Among them, is the average gradient magnitude of all pixel points adjacent to the pixel point in the th row and th column, is the error rate of the pixel point in the th row and th column.

[0094] The specific logic for generating the foam distribution index is as follows: Count the number of pixel points in the flotation foam area. Divide the number of pixel points in the flotation foam area by the total number of pixel points in the flotation foam image to obtain the proportion of flotation foam. Count the total number of flotation foams in each flotation foam image. For each flotation foam, draw the minimum circumscribed circle of the flotation foam. Starting from the overlapping position of the flotation foam and the minimum circumscribed circle, every Calculate the distance between the flotation foam and the minimum circumscribed circle on the same radius of the flotation foam, which is denoted as the deviation distance. On the same radius, the radius intersects the boundary of the flotation foam and the minimum circumscribed circle of the flotation foam at the same time, and the deviation distance is the distance between the two intersection points. Calculate the variance of the deviation distances of the same flotation foam. Calculate the area ratio of the flotation foam to the corresponding minimum circumscribed circle as the area similarity. Generate the shape regularity index based on the variance of the deviation distance and the area similarity. Calculate the average value of the shape regularity indices of each flotation foam in the same flotation foam image. Divide the flotation foam image into M regions evenly. Count the number of flotation foams in each region. Calculate the variance of the number of flotation foams in the same flotation foam image, which is denoted as the foam distribution uniformity. Generate the foam distribution index based on the total number of flotation foams, the proportion of flotation foam, the average value of the shape regularity index, and the foam distribution uniformity. The specific formula for calculating the shape regularity index is:

[0095] ;

[0096] Actually, is the shape regularity index, is the area similarity, is the variance of the deviation distance;

[0097] The shape regularity index reflects the degree of regularity of the shape of the flotation foam. The larger the value, the more regular the shape of the flotation foam. The area similarity reflects the degree of closeness of the area of the flotation foam to the minimum circumscribed circle. The larger the value, the closer the area of the flotation foam is to the minimum circumscribed circle. The variance of the deviation distance reflects the degree of closeness of the contour of the flotation foam to the minimum circumscribed circle. The smaller the value, the closer the contour of the flotation foam is to the minimum circumscribed circle. A circle is the ideal shape of the flotation foam. The closer its area and contour are to its minimum circumscribed circle, the more regular the shape of this flotation foam is considered;

[0098] The specific formula for generating the foam distribution index is:

[0099] ;

[0100] Among them, is the foam distribution index, is the total number of flotation foams, is the proportion of flotation foam, is the average value of the shape regularity index, is the foam distribution uniformity;

[0101] reflects the size of each flotation foam on average. The larger the value, the larger each flotation foam is on average; the average value of the shape regularity index reflects the comprehensive shape regularity of the flotation foams in the flotation foam image. The larger the value, generally, the more regular the shape of the flotation foams. The foam distribution uniformity reflects the distribution uniformity of the flotation foams in the flotation foam image. The smaller the value, the more uniform the distribution of the flotation foams in the flotation foam image. The foam distribution index reflects the distribution of the flotation foams. The larger the value, the relatively larger the flotation foams and the relatively more regular the shape and distribution of the flotation foams; generally speaking, the lower the pulp level, the thicker the flotation foam layer, the less likely it is to generate large foams, and the shape and distribution of the foams tend to be more disordered.

[0102] Step 4: Calculate the dynamic deviation degree of the flotation foams in the current-frame flotation foam image based on the texture distribution coefficient and the foam distribution index of the current-frame and the previous-frame flotation foam images;

[0103] wherein, is the dynamic deviation degree of the flotation foams in the -th frame flotation foam image, is the texture distribution coefficient of the -th frame flotation foam image, is the texture distribution coefficient of the -th frame flotation foam image, is the foam distribution index of the -th frame flotation foam image; is the foam distribution index of the -th frame flotation foam image.

[0104] The dynamic deviation degree of the flotation foams reflects the dynamic change situation of the texture distribution and the comprehensive foam distribution of two adjacent flotation foam images. The larger the value, the greater this dynamic change. Generally, under certain working conditions of the flotation machine, this dynamic change at a certain pulp level will fluctuate within a certain range. The generation of the dynamic deviation degree of the flotation foams can provide an important basis for the prediction of the pulp level; reflects the texture distribution deviation between two adjacent flotation foam images. The larger the value, the greater the texture distribution deviation between two adjacent flotation foam images, reflects that the foam distribution deviation between two adjacent flotation foam images is greater. The larger the value, the greater the foam distribution deviation between two adjacent flotation foam images.

[0105] Step 5: Using the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame as the model inputs, and the corresponding pulp level as the label, construct and train a pulp level prediction model, obtain the model inputs at the current moment, and input them into the pulp level prediction model to predict the pulp level.

[0106] The pulp level prediction model uses a feedforward neural network. With the pulp level as the label, use the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame to train and optimize the pulp level prediction model; specifically including: an input layer, a hidden layer, an output layer, and an activation function. The input layer is responsible for receiving the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame; the hidden layer is used to process the data of the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame; it consists of multiple layers, each layer contains multiple time nodes, and the time nodes of each hidden layer are connected to the previous layer through weights, which are used to perform feature abstraction and non-linear transformation on the input working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame; by using the ReLu activation function, introducing a non-linear relationship enables the model to fit complex feature relationships; an independent neuron is set in the output layer, which is responsible for converting the local and high-level feature representations extracted by the hidden layer for outputting the pulp level; the root mean square error loss function is adopted; the input data is calculated through the network once to obtain the output result, calculate the loss function according to the predicted value and the true value, calculate the gradient of the loss function with respect to each weight and bias through the chain rule, and use the gradient descent algorithm to update the weights and biases of the network to minimize the loss function.

[0107] Please refer to Figure 2 , the present invention further provides a pulp level system based on the dynamic characteristics of flotation foam. The system is used to implement the pulp level prediction method based on the dynamic characteristics of flotation foam, specifically including:

[0108] A data acquisition module, which is used to continuously collect flotation foam images and the corresponding pulp levels in different frames during the working process of a flotation machine with known working conditions; the working conditions of the flotation machine include the scraper speed, the air intake, and the slurry concentration;

[0109] A texture analysis module, which uses a gray-level co-occurrence matrix to extract texture features from each flotation foam image to obtain texture features, analyzes the correlation between each texture feature and the pulp level, and generates a texture distribution coefficient according to the texture features related to the pulp level;

[0110] A foam analysis module, which is used to identify the edge regions of each flotation foam in a flotation foam image using the Sobel operator, determine the flotation foam regions based on the edge regions of each flotation foam, and comprehensively analyze the flotation foam regions and the edge regions of the flotation foams to generate a foam distribution index;

[0111] A dynamic analysis module, which is used to calculate the dynamic deviation degree of the flotation foam in the flotation foam image in the current frame based on the texture distribution coefficients and the foam distribution index of the flotation foam images in the current frame and the previous frame;

[0112] A final prediction module, which is used to construct and train a pulp level prediction model with the working conditions of the flotation machine, the dynamic deviation degree of the flotation foam, the texture distribution coefficient, and the foam distribution index in the same frame as the model inputs, and the corresponding pulp level as the label, obtain the model inputs at the current moment, and input them into the pulp level prediction model to predict the pulp level.

[0113] The present invention further provides a computer-readable storage medium, and a computer program executable by a processor is stored inside the storage medium. When the computer program is executed by the processor, the pulp level prediction method based on the dynamic characteristics of flotation foams can be implemented.

[0114] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0115] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or by a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0116] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They can be located in one place, or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0117] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all of them should be covered by the protection scope of the present application.

Claims

1. A method for predicting slurry level based on dynamic characteristics of flotation foam, characterized in that: The specific steps include: Step 1: During the operation of a flotation machine under known working conditions, continuously collect flotation foam images and corresponding slurry levels under different frames; the flotation machine working conditions include scraper speed, air intake and slurry concentration; Step 2: Use the gray-level co-occurrence matrix to extract the texture features of each flotation foam image to obtain the texture features, analyze the correlation between each texture feature and the slurry level, and generate the texture distribution coefficient according to the texture features associated with the slurry level; Step 3: Use the Sobel operator to identify the edge area of ​​each flotation bubble in the flotation bubble image, determine the flotation bubble area based on the edge area of ​​each flotation bubble, and perform a comprehensive analysis of the flotation bubble area and the edge area of ​​the flotation bubble to generate a foam distribution index; Step 4: Based on the texture distribution coefficient and the foam distribution index of the flotation foam image of the current frame and the previous frame, the flotation foam dynamic deviation degree of the flotation foam image of the current frame is calculated; Step 5: Taking the flotation machine working conditions, flotation foam dynamic deviation, texture distribution coefficient and foam distribution index in the same frame as the model input and the corresponding slurry level as the label, construct and train the slurry level prediction model, obtain the model input at the current moment, and input it into the slurry level prediction model to predict the slurry level; The texture features include contrast, dissimilarity, homogeneity, energy and entropy, The specific formula used to calculate the correlation between each texture feature and the slurry level is: Among them, r j is the correlation between the jth texture feature and the slurry level, W ij is the characteristic value of the jth texture feature of the i-th frame of flotation foam image, N is the total number of frames of flotation foam images collected, is the average eigenvalue of the jth texture feature of the N collected flotation foam images, is the average value of the pulp level of N flotation froth images collected, D i is the pulp level of the i-th frame of flotation foam image, i is the flotation foam image frame number index, and i∈[1,N], j is the index of texture features, and j∈[1,5]; The Sobel operator is used to calculate the horizontal gradient value and the vertical gradient value of each pixel in the flotation foam image, and a first gradient amplitude data set including the gradient amplitude data of all pixel points is generated, and the error rate between the gradient amplitude of the current pixel point and the gradient amplitude of the adjacent pixel point is determined. If the error rate is less than the set error rate threshold, the pixel value of the current pixel point is retained, otherwise it is set to 0, and the portion with the retained pixel value is defined as the edge area of ​​the flotation foam, and the closed area wrapped by the edge area is defined as the flotation foam, and the set of all flotation foam positions is the flotation foam area.

2. A method for predicting slurry level based on dynamic characteristics of flotation foam according to claim 1, characterized in that: The specific logic for judging whether the texture feature is consistent with the slurry level is: Preset texture feature association threshold r0, if |r j |>r0, then the corresponding texture feature is judged to be associated with the slurry level, and the texture feature is defined as the associated texture feature. j |≤r0, then it is judged that the corresponding texture feature has no correlation with the slurry level; The specific formula for generating the texture distribution coefficient is: Among them, WF i is the texture distribution coefficient of the flotation foam image of the i-th frame, r norm (k) is the weight of the kth key texture feature, W ik is the kth key texture feature value in the flotation foam image of the i-th frame.

3. The method for predicting slurry level based on dynamic characteristics of flotation foam according to claim 1, characterized in that: When calculating the horizontal gradient value and vertical gradient value of each pixel, the grayscale value of the pixel and the grayscale values ​​of its adjacent pixels are multiplied with the horizontal gradient template of the Sobel operator, and all the results of the products are added to obtain the horizontal gradient value of the pixel. The grayscale value of the pixel and the grayscale value of its adjacent pixels are multiplied with the vertical gradient template of the Sobel operator, and all the results of the products are added to obtain the vertical gradient value of the pixel. The formula for calculating the horizontal gradient value and the vertical gradient value is: Among them, S(l,m) and G(l,m) are the horizontal gradient value and vertical gradient value of the pixel point in the lth row and the mth column respectively; H(l,m) is the gray value of the pixel point in the lth row and the mth column, and l and m are the indexes of the pixel points in the flotation bubble image; The formula for generating the gradient amplitude of a pixel is: Where T(l,m) is the gradient amplitude of the pixel at the lth row and the mth column; The formula for generating the error rate between the gradient amplitude of the current pixel and the gradient amplitude of the adjacent pixel is: Wherein, T(o,p) is the average gradient amplitude of all pixels adjacent to the pixel in the lth row and the mth column, and E(l,m) is the error rate of the pixel in the lth row and the mth column.

4. The method for predicting slurry level based on dynamic characteristics of flotation foam according to claim 3, characterized in that: The specific logic for generating the foam distribution index is as follows: the number of pixels occupied by the flotation foam area is counted, and the number of pixels occupied by the flotation foam area is divided by the total number of pixels of the flotation foam image to obtain the flotation foam proportion; Count the total number of flotation bubbles in each flotation bubble image, for each flotation bubble, make the minimum circumscribed circle of the flotation bubble, start from the overlapping position of the flotation bubble and the minimum circumscribed circle, calculate the distance between the flotation bubble and the minimum circumscribed circle on the same radius of the flotation bubble every 15° in the clockwise direction, record it as the deviation distance, calculate the variance of the deviation distance of the same flotation bubble, calculate the area ratio of the flotation bubble to the corresponding minimum circumscribed circle as the area similarity, generate the shape regularity index according to the variance of the deviation distance and the area similarity, calculate the average value of the shape regularity index of each flotation bubble in the same flotation bubble image; divide the flotation bubble image into M blocks, count the number of flotation bubbles in each block, calculate the variance of the number of flotation bubbles in the same flotation bubble image, record it as the uniformity of foam distribution; generate the foam distribution index according to the total number of flotation bubbles, the proportion of flotation bubbles, the average value of the shape regularity index and the uniformity of foam distribution; The specific formula for calculating the shape regularity index is: In fact, HD is the shape regularity index, SI is the area similarity, σ P is the variance of the deviation distance; The specific formula upon which the Foam Distribution Index is generated is: Among them, PF is the foam distribution index, ML is the total flotation foam quantity, Nb is the flotation foam proportion, is the shape regularity index mean, σ B It is the uniformity of foam distribution.

5. The method for predicting slurry level based on dynamic characteristics of flotation foam according to claim 4, characterized in that: The specific formula for calculating the dynamic deviation of flotation foam is: PA i =|WF i -WF i-1 |*|PF i -PF i | Among them, PA i WF is the flotation foam dynamic deviation of the i-th frame flotation foam image, i is the texture distribution coefficient of the flotation foam image in the i-th frame, WF i-1 is the texture distribution coefficient of the flotation foam image of the i-1th frame, PF i is the foam distribution index of the flotation foam image of the i-th frame; PF i-1 is the foam distribution index of the flotation foam image in the i-1th frame.

6. A slurry level system based on the dynamic characteristics of flotation foam, characterized by: The system is used to implement the method for predicting slurry level based on dynamic characteristics of flotation foam according to any one of claims 1 to 5, and specifically comprises: A data acquisition module is used to continuously acquire flotation foam images and corresponding slurry levels in different frames during the operation of a flotation machine under known working conditions; the flotation machine working conditions include scraper speed, air intake and slurry concentration; The texture analysis module uses the gray-level co-occurrence matrix to extract the texture features of each flotation foam image to obtain the texture features, analyzes the correlation between each texture feature and the slurry level, and generates the texture distribution coefficient according to the texture features associated with the slurry level; A foam analysis module is used to identify the edge area of ​​each flotation foam in the flotation foam image using the Sobel operator, determine the flotation foam area based on the edge area of ​​each flotation foam, and perform a comprehensive analysis on the flotation foam area and the edge area of ​​the flotation foam to generate a foam distribution index; A dynamic analysis module, for calculating the flotation foam dynamic deviation of the flotation foam image in the current frame based on the texture distribution coefficient and the foam distribution index of the flotation foam image in the current frame and the previous frame; The final prediction module is used to build and train a slurry level prediction model with the flotation machine working conditions, flotation foam dynamic deviation, texture distribution coefficient and foam distribution index in the same frame as model input and the corresponding slurry level as a label, obtain the model input at the current moment, and input it into the slurry level prediction model to predict the slurry level.

7. A computer-readable storage medium, characterized in that: The storage medium internally stores a computer program that can be executed by a processor, and when the computer program is executed by the processor, the method for predicting the slurry level based on the dynamic characteristics of flotation foam described in any one of claims 1 to 5 can be implemented.

Images