CN109772593B - A method for predicting slurry level based on dynamic characteristics of flotation foam - Google Patents

Abstract

The invention provides a method for predicting the liquid level of ore slurry based on the dynamic characteristics of flotation foam, and relates to the field of production detection in flotation process. Including: establishing the historical database of flotation parameters, historical image database, and historical database of slurry level prediction; processing two adjacent frames of the image sequence and extracting the motion area; drawing the outline of the macro block of the frame image, searching for the calculation of the froth flow velocity The optimal position of the foam is calculated by Fourier transform of two consecutive frames of related images; the movement speed of the foam is calculated; the historical data sample library is established and fitted to obtain the function of speed and slurry level and save; corresponding to the flotation machine The slurry level at the moment is predicted and output. The method takes the image motion characteristics and foam dynamic characteristic parameters as input, builds modeling based on the SVR support vector regression algorithm, realizes the detection of the slurry level process in the flotation cell, and solves the problem that the existing technology relies on manual expert experience and cannot predict the slurry level. The problem of real-time monitoring of bits.

Description

A method for predicting slurry level based on dynamic characteristics of flotation foam

technical field

The invention relates to the technical field of production detection in flotation process, in particular to a method for predicting the liquid level of ore slurry based on the dynamic characteristics of flotation foam.

Background technique

Mineral flotation is a method in which flotation reagents are added to the pulp, filled with air, and a large number of air bubbles are generated by stirring, and then the mineral foam is recovered to achieve the beneficiation target under specific process conditions. Flotation froth is characterized by a large number, adhesion, mixing, and irregular shape. The froth morphology and flow rate characteristics are difficult to quantitatively describe. Numerous studies on flotation process have shown that the characteristic change of flotation froth can reflect the change of process control quantity and ore properties in time, so the characteristic change of flotation froth can be used as the quickest indication of flotation effect.

The size of the foam flow rate is the most concerned object among all the dynamic and static characteristics of the foam. It plays an extremely critical role in the grade-recovery optimization control system, and has a significant adjustment effect on the mineral recovery rate. During the flotation process, the on-site workers mainly rely on the slurry level of the flotation machine to judge whether the flotation conditions are normal. However, due to the limitations of the site conditions, there is no better way to solve the problem of real-time monitoring of the slurry level.

SUMMARY OF THE INVENTION

Aiming at the problems existing in the prior art, the present invention provides a method for predicting the slurry level based on the dynamic characteristics of flotation foam. Therefore, the existing technology relies on the experience of manual experts and cannot perform real-time monitoring of the predicted slurry level.

In order to achieve the above purpose, a method for predicting the slurry level based on the dynamic characteristics of flotation foam, the specific steps are as follows:

Step 1: collect characteristic parameters related to the dynamic characteristics of froth in the flotation process of the flotation machine to establish a flotation parameter historical database; the characteristic parameters include scraper rotation speed, aeration volume and pulp concentration;

Step 2: Use video surveillance technology to collect continuous flotation froth images and establish a historical image database;

Step 3: Record the below-line value of the slurry level of the flotation machine in each image in the historical image database and establish a historical database for slurry level prediction;

Step 4: Extract the foam flow velocity in each image in the historical image database, and establish a historical database of foam flow velocity. The specific steps are as follows:

Step 4.1: Find two adjacent frames of the image sequence in the historical image database, then perform grayscale processing on the two frame images, and then use the inter-frame difference method to obtain the difference image and binarize it;

Step 4.2: Threshold the image after the difference binarization, and extract the motion area in the image;

Step 4.3: Divide the frame image of the motion area into multiple image blocks of fixed size, use the macroblock registration method to analyze the motion of the image blocks, find out the position correspondence of each part of the frame image before and after, draw the outline of the macroblock, and obtain The movement speed characteristics of each sub-block;

Step 4.4: Use a three-step search method to search for the best matching position of the selected target sub-block in the adjacent frame through two stages in the frame image of the motion area, that is, the position most suitable for calculating the speed of foam movement;

Step 4.5: Determine whether the k-th frame image and the k+1-th frame image are related by means of the magnitude of the correlation function. If so, continue to step 4.6; if not, determine whether the other two adjacent images are related, until all frame images are judged to be completed;

Step 4.6: Perform Fourier transform on the k-th frame image and the k+1-th frame image respectively, convert the image from the spatial domain to the frequency domain, and obtain F _k (w, v) and F _k+1 (w, v) , where w and v are the frequency domain coordinates of the image;

Step 4.7: Calculate the relative translations x ₀ and y ₀ of the two images in the x-axis and y-axis directions according to the cross-power spectrum of the k-th frame image and the k+1-th frame image Fourier map, so as to obtain the foam movement speed;

The formula for calculating the cross-power spectrum of the Fourier atlas of the k-th frame image and the k+1-th frame image is as follows:

Among them, F ₁ ^* represents the complex conjugate of F ₁ , j is an imaginary unit, and j ² =-1;

Step 4.8: Create a historical database of foam flow speed according to the obtained foam flow speed;

Step 5: According to the historical database of flotation parameters, the historical database of slurry level prediction and the historical database of froth flow rate, establish a historical data sample database;

Step 6: Fit the sample in the historical data sample library through SVR support vector regression to fit the functional relationship between the foam flow velocity and the slurry level and save it in the form of a function. The specific steps are as follows:

Step 6.1: Standardize the values of foam flow velocity and slurry level in the historical data sample library;

Step 6.2: Arrange the standardized samples in random order, take a part of the samples as training data, and use the remaining samples as test data;

Step 6.3: Fit the training data through SVR support vector regression, and test the fitting results to obtain the functional relationship between the foam flow velocity and the slurry level and save it in the form of a function;

Step 7: Use video surveillance technology to collect new continuous flotation froth images, repeat step 4, and extract the froth flow velocity in the updated image;

Step 8: Take the extracted froth flow velocity as input, bring in the obtained froth flow velocity and the slurry level function, predict and output the slurry level of the flotation machine at the corresponding moment.

Beneficial effects of the present invention:

The invention proposes a method for predicting the slurry level based on the dynamic characteristics of flotation foam. According to historical data, production site investigation and theoretical analysis of flotation production, the flow rate of the flotation tank foam has an inevitable relationship with the slurry level. At the same time, in view of the dynamic characteristics of the flotation foam image, the idea of ore slurry level detection based on visual information is proposed, and the motion characteristics of the foam image are effectively extracted by the combination of the frame difference method and the macroblock registration method. The rotation speed, aeration volume, and slurry concentration are input, and the modeling is carried out based on the SVR support vector regression algorithm. The function of detecting the slurry level in the flotation tank is realized.

Description of drawings

Fig. 1 is the flow chart of the method for predicting the slurry level based on the dynamic characteristics of flotation foam in the embodiment of the present invention;

FIG. 2 is a schematic diagram of collecting continuous flotation froth images using video monitoring technology in an embodiment of the present invention;

Wherein, (a) is a schematic diagram of the first image in the continuous flotation froth images collected by video surveillance technology; (b) is a schematic diagram of the second image in the continuous flotation froth images captured by video surveillance technology; (c) ) is the schematic diagram of the third image in the continuous flotation froth images collected by video surveillance technology; (d) is the schematic diagram of the fourth image in the continuous flotation froth images captured by video surveillance technology; (e) is the schematic diagram of the fourth image using video surveillance technology Schematic diagram of the fifth image in the continuous flotation froth images collected by monitoring technology; (f) is a schematic diagram of the sixth image in the continuous flotation froth images collected by video surveillance technology;

3 is a schematic diagram of an image obtained by performing grayscale processing and differential binarization on two adjacent frames of images collected in an embodiment of the present invention;

Among them, (a) is a schematic diagram of the first image collected after grayscale processing and differential binarization; (b) is the second image collected after grayscale processing and differential binarization Schematic diagram of the image; (c) is a schematic diagram of the image after grayscale processing and differential binarization of the third image collected; (d) The fourth image collected is subjected to grayscale processing and differential binarization Schematic diagram of the image after transformation; (e) is a schematic diagram of the image after grayscale processing and differential binarization of the fifth image collected; (f) is the sixth image collected after grayscale processing and differential Schematic diagram of the image after binarization;

4 is a schematic diagram of an image after thresholding processing of a differential binarized image in an embodiment of the present invention;

Among them, (a) is a schematic diagram of the first differential binarized image after thresholding; (b) is a schematic diagram of the second differential binarized image after thresholding; (c) is the third image. Schematic diagram of the image after thresholding of the differential binarized image; (d) is a schematic diagram of the image after thresholding of the fourth differential binarized image; (e) is the fifth differential binarized image subjected to thresholding Schematic diagram of the image after processing; (f) is a schematic diagram of the image after thresholding processing of the sixth differential binarized image;

5 is a schematic diagram of an image after a thresholded image is drawn with a macroblock outline according to an embodiment of the present invention;

Among them, (a) is a schematic diagram of the first thresholded image after macroblock outline drawing; (b) is an image schematic diagram of the second thresholded image after macroblock outline drawing; (c) is the third threshold value image (d) is a schematic diagram of the fourth thresholded image after macroblock outline drawing; (e) is the fifth thresholded image after macroblock outline drawing Schematic diagram; (f) is a schematic diagram of the image after the sixth thresholded image is drawn by the macroblock outline;

6 is a schematic diagram of collecting new continuous flotation froth images using video surveillance technology in an embodiment of the present invention;

Wherein, (a) is a schematic diagram of the first image in the new continuous flotation froth image collected by video surveillance technology; (b) is the second image in the new continuous flotation froth image collected by video surveillance technology A schematic diagram of an image.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The specific embodiments described herein are only used to explain the present invention, and are not intended to limit the present invention.

A method for predicting the slurry level based on the dynamic characteristics of flotation foam, the process is shown in Figure 1, including the following steps:

Step 1: Collect characteristic parameters related to the dynamic characteristics of froth in the flotation process of the flotation machine to establish a flotation parameter historical database; the characteristic parameters include scraper rotation speed, aeration volume and pulp concentration.

In this embodiment, the collected historical data of parameters related to flotation dynamic characteristics are shown in Table 1.

Table 1 Historical data of related parameters (part of the sample library)

Step 2: Use video surveillance technology to collect continuous flotation froth images and establish a historical image database.

In this embodiment, the collected continuous flotation froth images are shown in FIG. 2 .

Step 3: Record the below-line value of the slurry level of the flotation machine in each image in the historical image database and establish a historical database of slurry level prediction.

In this embodiment, the slurry level prediction historical database constructed by the below-line value of the slurry level of the flotation machine in each image is shown in Table 1.

Step 4: Extract the foam flow velocity in each image in the historical image database, and establish a historical database of foam flow velocity. The specific steps are as follows:

Step 4.1: Find two adjacent frames of the image sequence in the historical image database, then perform grayscale processing on the two frame images, and then use the inter-frame difference method to obtain the difference image and binarize it;

In this embodiment, the grayscale processing is performed on the images in the historical image database, and the result after performing differential binarization is shown in FIG. 3 .

Step 4.2: Threshold the image after the difference binarization, and extract the motion area in the image.

In this embodiment, the result of performing thresholding processing on the difference binarized image is shown in FIG. 4 .

Step 4.3: Divide the frame image of the motion area into multiple image blocks of fixed size, use the macroblock registration method to analyze the motion of the image blocks, find out the position correspondence of each part of the frame image before and after, draw the outline of the macroblock, and obtain The movement speed characteristics of each sub-block.

In this embodiment, the image on which the outline of the macroblock is drawn is shown in FIG. 5 .

Step 4.4: Use a three-step search method to search for the best matching position of the selected target sub-blocks in adjacent frames in two stages in the frame image of the motion area, that is, the position that is most suitable for calculating the speed of foam movement.

Step 4.5: Determine whether the k-th frame image and the k+1-th frame image are related by means of the magnitude of the correlation function. If so, continue to step 4.6; if not, determine whether the other two adjacent images are related, until all frame images All judged to be complete.

Step 4.6: Perform Fourier transform on the k-th frame image and the k+1-th frame image respectively, convert the image from the spatial domain to the frequency domain, and obtain F _k (w, v) and F _k+1 (w, v) , where w and v are the frequency-domain coordinates of the image.

In this embodiment, if the k+1 th frame image f _k+1 (x, y) is obtained after the k th frame image f _k (x, y) is translated by x ₀ and y ₀ in the x and y directions, respectively, Then there is formula (1):

f _k+1 (x, y)=f _k (xx ₀ , yy ₀ ) (1)

where x and y are the spatial domain coordinates of the image.

Fourier transform is performed on the k-th frame image and the k+1-th frame image to obtain F _k (w, v) and F _k+1 (w, v), then there is formula (2):

Step 4.7: Calculate the relative translations x ₀ and y ₀ of the two images in the x-axis and y-axis directions according to the cross-power spectrum of the k-th frame image and the k+1-th frame image Fourier map, so as to obtain the foam Movement speed.

The calculation formula of the cross-power spectrum of the Fourier atlas of the k-th frame image and the k+1-th frame image is shown in formula (3):

Wherein, F ₁ ^* represents the complex conjugate of F ₁ , j is an imaginary unit, and j ² =-1.

Step 4.8: According to the obtained foam flow speed, a historical database of foam flow speed is established.

By inverse Fourier transform of formula (3), a unit impulse function will be formed at the spatial point (x ₀ , y ₀ ), and the pulse position is the relative translation amount x ₀ and y between the two registered images ₀ , the movement speed of the foam can be obtained through this displacement.

In this embodiment, the flow velocity toward the flotation cell is defined as the positive direction, and the flow toward the center of the flotation cell is defined as the negative flow velocity. The obtained foam flow velocity historical database is shown in Table 2.

Table 2 Extraction of flotation froth flow velocity characteristics (example)

Step 5: According to the historical database of flotation parameters, the historical database of slurry level prediction and the historical database of froth flow rate, a historical data sample database is established.

In this embodiment, the established historical data sample database is shown in Table 3.

Table 3 Sample library (part)

Step 6: Fit the sample in the historical data sample library through SVR support vector regression to fit the functional relationship between the foam flow velocity and the slurry level and save it in the form of a function. The specific steps are as follows:

Step 6.1: Standardize the values of foam flow velocity and slurry level in the historical data sample library.

In this embodiment, the normalization processing is the normalization processing, and the specific processing formula is shown in formula (4):

P _i =2(PP _min )/(P _max -P _min )-1 (4)

Among them: _Pi is the processed data, P is the input data, _Pmax is the maximum value in the input data, and _Pmin is the minimum value in the input data.

Step 6.2: Arrange the standardized samples in random order, take a part of the samples as training data, and use the remaining samples as test data.

In this embodiment, the first four-fifths of the samples are taken as training data, and the last fifth of the samples are taken as test data.

Step 6.3: Fit the training data through SVR support vector regression, and test the fitting results to obtain the functional relationship between the foam flow velocity and the slurry level and save it in the form of a function.

In this embodiment, when the test error is less than 0.1%, it is considered that the training result is satisfactory, and the set of functional relationships is saved, and the obtained functional expression is shown in formula (5):

high=f(V ₀ , A, C, |V|) (5)

Among them, V ₀ is the rotating speed of the scraper, A is the aeration amount, C is the slurry concentration, and V is the slurry flow rate.

Step 7: Use video surveillance technology to collect new continuous flotation froth images, repeat step 4, and extract the froth flow velocity in the updated image.

In this embodiment, the acquired new continuous flotation froth image is shown in FIG. 6 , and step 4 is repeated, and the extracted froth flow velocity is shown in Table 4.

Table 4 Dynamic characteristics of flotation froth

Scraper speed V₀ Aeration volume A(m3/min) Pulp concentration C (%) Slurry velocity V 20 1.02 32.30 52.01

Step 8: Take the extracted froth flow velocity as input, bring in the obtained froth flow velocity and the slurry level function, predict and output the slurry level of the flotation machine at the corresponding moment.

In this embodiment, the extracted foam flow velocity is brought into the obtained function of the foam flow velocity and the slurry level, and the predicted results of the slurry level and the actual recorded results are shown in Table 5.

Table 5 Comparison table between predicted results of slurry level and actual recorded results

serial number Slurry level value forecast result 2.2 actual results 2.1

Compared with the existing method, the prediction method provided by the present invention has the following advantages: through the image feature extraction technology, the feature extraction is performed on the foam image by using the difference between frames and the macroblock registration method, and a prediction model of the slurry level based on the foam image is established. , realizes the prediction of the slurry level in the flotation process, and can provide assistance for the operator to adjust the flotation conditions.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, but not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand; The technical solutions described in the foregoing embodiments are modified, or some or all of the technical features thereof are equivalently replaced; therefore, these modifications or replacements do not make the essence of the corresponding technical solutions depart from the scope defined by the claims of the present invention.

Claims (3)

1. a method for predicting ore slurry liquid level based on flotation froth dynamic characteristics, is characterized in that, comprises the following steps: Step 1: Collect characteristic parameters related to the dynamic characteristics of froth in the flotation process of the flotation machine to establish a flotation parameter historical database; the characteristic parameters include scraper rotation speed, aeration volume and pulp concentration; Step 2: Use video surveillance technology to collect continuous flotation froth images and establish a historical image database; Step 3: Record the below-line value of the slurry level of the flotation machine in each image in the historical image database and establish a historical database of slurry level prediction; Step 4: Extract the foam flow velocity in each image in the historical image database, and establish a foam flow velocity historical database. The process is as follows; Step 4.1: Find two adjacent frames of the image sequence in the historical image database, then perform grayscale processing on the two frames of images, and then use the inter-frame difference method to obtain the difference image and binarize it; Step 4.2: Threshold the image after the difference binarization, and extract the motion area in the image; Step 4.3: Divide the frame image of the motion area into multiple image blocks of fixed size, use the macroblock registration method to analyze the motion of the image blocks, find out the position correspondence of each part of the frame image before and after, draw the outline of the macroblock, and obtain The movement speed characteristics of each sub-block; Step 4.4: Use the three-step search method to search for the best matching position of the selected target sub-block in the adjacent frame through two stages in the frame image of the motion area, that is, the position that is most suitable for calculating the speed of foam movement; Step 4.5: Determine whether the k-th frame image and the k+1-th frame image are related by means of the magnitude of the correlation function. If so, continue to step 4.6; if not, determine whether the other two adjacent images are related, until all frame images are judged to be completed; Step 4.6: Perform Fourier transform on the k-th frame image and the k+1-th frame image respectively, convert the image from the spatial domain to the frequency domain, and obtain F _k ( w, v ) and F _k+1 ( w, v ) , where w and v are the frequency domain coordinates of the image; Step 4.7: Calculate the relative translations x0 and y0 of the two images in the x-axis and y-axis directions according to the cross-power spectrum of the k-th frame image and the k+1-th frame image Fourier spectrum, so as to obtain the movement speed of the foam ; Step 4.8: Create a foam flow rate historical database based on the obtained foam flow rate; Step 5: According to the historical database of flotation parameters, the historical database of slurry level prediction and the historical database of froth flow rate, establish a historical data sample database; Step 6: Fit the sample in the historical data sample library through SVR support vector regression to fit the functional relationship between the foam flow velocity and the slurry level and save it in the form of a function; Step 7: Use video surveillance technology to collect new continuous flotation froth images, repeat step 4, and extract the froth flow velocity in the updated image; Step 8: Take the extracted froth flow velocity as input, bring in the obtained froth flow velocity and slurry level function, predict and output the slurry level of the flotation machine at the corresponding moment. 2. The method for predicting the slurry level based on the dynamic characteristics of flotation foam according to claim 1, wherein in the step 4.7, the cross-power spectrum of the k-th frame image and the k+1-th frame image Fourier spectrum The calculation formula is as follows:

Wherein, F ₁ * represents the complex conjugate of F ₁ , j is an imaginary unit, and j ² =-1. 3. The method for predicting the slurry level based on the dynamic characteristics of flotation foam according to claim 1, wherein the step 6 comprises the following steps: Step 6.1: Standardize the values of foam flow velocity and slurry level in the historical data sample library; Step 6.2: Arrange the standardized samples in random order, take a part of the samples as training data, and use the remaining samples as test data; Step 6.3: Fit the training data through SVR support vector regression, and test the fitting results to obtain the functional relationship between the foam flow velocity and the slurry level and save it in the form of a function.

Images