Flotation froth image feature extraction and multi-working-condition identification system and method

Abstract

The invention relates to a flotation froth image feature extraction and multi-working-condition recognition system and method, belongs to the technical field of flotation, and solves the problems of difficulty in collection of flotation froth images, lack of extreme working condition data, weak generalization ability of an image recognition model and the like in the prior art. The system comprises a flotation unit, an image acquisition unit and a feature extraction and working condition analysis unit, the flotation unit is used for flotation of ore pulp to generate flotation froth, the image acquisition unit is used for acquisition of flotation froth images under different working conditions, and the feature extraction and working condition analysis unit is used for extraction and analysis of flotation froth image features; the image acquisition unit comprises a flotation housing mechanism and a concentrate collecting tank, the flotation housing mechanism comprises a box body, and the flotation unit and the concentrate collecting tank are both arranged in the box body. According to the method, industrial images do not need to be collected on site, an extreme working condition environment can be created in a laboratory, data are rich, accurate flotation data can be obtained, and beneficial guidance is provided for actual production.

Description

Technical Field

[0001] The present invention relates to the field of flotation technology, and in particular to a flotation foam image feature extraction and multi-operating condition recognition system and method. Background Art

[0002] Flotation is currently the core purification method for mineral resources such as coal, copper-nickel sulfide ore, lead-zinc ore, etc. Its flotation effect directly affects product quality. Therefore, the monitoring and quantification of flotation effect information such as foam state, concentrate grade, and yield during flotation are crucial to achieve ore quality optimization and fine control of the production process. However, in the traditional flotation process, flotation drivers often judge and adjust the flotation effect by observing the flotation foam state. This method relies on the personal experience of the flotation driver to estimate the flotation condition, lacks a unified quantitative standard, and cannot achieve full-cycle status monitoring of flotation, which leads to large production fluctuations.

[0003] Traditional flotation foam image data collection often requires on-site industrial image collection, and its data labels also come from on-site test data. Although this method is the most intuitive and practical, due to the long industrial process, there is a lag between the foam state and the flotation index, and the data labels are distorted. In addition, the data collection process requires artificial creation of various flotation deterioration conditions, which affects the economic benefits of the ore dressing plant. Therefore, direct on-site sampling is not conducive to laboratory preliminary research and unit condition experiments. Summary of the invention

[0004] In view of the above analysis, the embodiments of the present invention aim to provide a flotation foam image feature extraction and multi-operating condition recognition system and method to solve the problems of difficulty in collecting existing flotation foam images, lack of extreme operating condition data, and weak generalization ability of image recognition models.

[0005] On the one hand, the present invention provides a flotation foam image feature extraction and multi-operating condition identification system, comprising a flotation unit, an image acquisition unit and a feature extraction and operating condition analysis unit, wherein the flotation unit is used for flotation of slurry to generate flotation foam, the image acquisition unit is used for collecting flotation foam images under different operating conditions, and the feature extraction and operating condition analysis unit is used for extracting and analyzing flotation foam image features;

[0006] The image acquisition unit comprises a flotation housing mechanism and a concentrate collecting tank. The flotation housing mechanism comprises a box body. The flotation unit and the concentrate collecting tank are both arranged in the box body.

[0007] Furthermore, the flotation housing mechanism also includes a rectangular light source and an annular shadowless light source arranged in the box body.

[0008] Furthermore, two rectangular light sources are provided, and the two rectangular light sources are respectively arranged on two opposite side walls of the box body, and the annular shadowless light source is arranged on the top of the box body.

[0009] Furthermore, the flotation unit comprises a flotation cell body, a transmission and stirring mechanism and a support base, wherein the upper portion of the transmission and stirring mechanism is connected to the flotation cell body, and the lower portion of the transmission and stirring mechanism is connected to the support base.

[0010] Furthermore, the flotation cell body comprises a flotation cell and an overflow cell, wherein the overflow cell is arranged at the top of the flotation cell and is connected to the flotation cell.

[0011] Furthermore, the flotation cell body also includes a partition plate horizontally arranged in the flotation cell, and the partition plate divides the inner cavity of the flotation cell into an upper chamber and a lower chamber.

[0012] Furthermore, the flotation cell body further comprises an air inlet pipe, and two of the air inlet pipes are provided, and the two air inlet pipes are symmetrically arranged on both sides of the flotation cell.

[0013] Furthermore, an air intake channel is provided in the partition plate, one end of the air intake channel is communicated with the air intake pipe, and the other end of the air intake channel is communicated with the upper chamber.

[0014] Furthermore, a one-way valve is provided on the air intake pipe.

[0015] On the other hand, the present invention provides a method for flotation foam image feature extraction and multi-operating condition identification, which uses the above system to extract and analyze the image features of flotation foam under different operating conditions.

[0016] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0017] (1) The flotation process of the present invention is completed in the box while the industrial camera automatically obtains the flotation foam image. The feature extraction and working condition analysis unit extracts and analyzes the collected image features, thereby realizing flotation multi-working condition recognition. There is no need to collect industrial images on site, and there will be no lagging relationship between the foam state and the flotation index, and there will be no distortion of data labels. Extreme working conditions can be created in the laboratory, the data is enriched, and accurate flotation data can be obtained, providing useful guidance for actual production.

[0018] (2) The present invention provides an annular radial stator baffle around the stirring impeller, and the stirring blades are arranged in an annular radial pattern at the bottom of the umbrella-shaped top of the stirring impeller. There are gaps between the stator baffle and the stirring blades, between the stirring blades and the central axis of the umbrella-shaped top, and between the bottom of the stirring blades, the bottom of the stator baffle and the partition plate, which can enhance the shear force during the slurry stirring flotation process and provide a stable flotation environment for the slurry.

[0019] (3) The present invention quantifies the multiphase characteristic indicators of foam morphology, size distribution, and bubble color respectively, and has a feature extraction method that is universal for various foams. For different types of minerals, only a small amount of data set training is required for the final nonlinear regression model to obtain a better working condition prediction effect. Compared with the traditional "end-to-end" prediction model, the demand for data sets is greatly reduced.

[0020] In the present invention, the above-mentioned technical solutions can also be combined with each other to achieve more preferred combination solutions. Other features and advantages of the present invention will be described in the subsequent description, and some advantages can become obvious from the description, or can be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0022] Figure 1 It is a schematic diagram of the framework of the flotation foam image feature extraction and multi-operating condition recognition system of the specific embodiment 1;

[0023] Figure 2 It is a structural schematic diagram of a flotation unit of specific embodiment 1;

[0024] Figure 3 is a cross-sectional schematic diagram of a flotation unit of specific embodiment 1;

[0025] Figure 4 This is one of the structural schematic diagrams of the stirring impeller and the stator baffle of the specific embodiment 1;

[0026] Figure 5 The second structural schematic diagram of the stirring impeller and the stator baffle of the specific embodiment 1;

[0027] Figure 6 is a structural schematic diagram of an image acquisition unit in specific embodiment 1;

[0028] Figure 7 It is a schematic diagram of the connection structure between the support base and the stepping motor of the specific embodiment 1;

[0029] Figure 8 It is a flow chart of the flotation foam image feature extraction and multi-condition identification method of specific embodiment 2.

[0030] Reference numerals:

[0031] 1-flotation unit; 11-flotation tank body; 111-flotation tank; 1111-upper chamber; 1112-lower chamber; 112-overflow tank; 113-partition plate; 1131-intake channel; 114-stator baffle; 115-baffle frame; 1151-annular frame; 1152-support column; 116-intake pipe; 12-transmission stirring mechanism; 121-stirring impeller; 1211-umbrella top; 1 212-stirring blade; 122-transmission shaft; 123-coupling; 124-stepping motor; 13-support base; 131-bottom shell; 132-motor seat; 2-image acquisition unit; 21-flotation cover mechanism; 211-box; 212-rectangular light source; 213-annular shadowless light source; 214-bracket; 22-concentrate collecting tank; 23-industrial camera; 3-feature extraction and working condition analysis unit. DETAILED DESCRIPTION

[0032] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of the present invention and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0033] Example 1

[0034] A specific embodiment of the present invention, as Figure 1 As shown, a flotation foam image feature extraction and multi-operating condition recognition system is disclosed, including a flotation unit 1, an image acquisition unit 2 and a feature extraction and operating condition analysis unit 3. The flotation unit 1 is used for flotation of slurry to generate flotation foam, the image acquisition unit 2 is used for flotation foam data collection, and the feature extraction and operating condition analysis unit 3 is used for flotation foam data analysis.

[0035] like Figure 2 and Figure 3 As shown, the flotation unit 1 includes a flotation cell 11, a transmission and stirring mechanism 12 and a support base 13. The upper part of the transmission and stirring mechanism 12 is connected to the flotation cell 11, and the lower part of the transmission and stirring mechanism 12 is connected to the support base 13. The support base 13 is arranged below the flotation cell 11.

[0036] Since the transmission stirring mechanism 12 will generate vibration during the stirring process of the ore pulp, a support base 13 is provided at the bottom of the flotation cell 11 to simultaneously support the flotation cell 11 and the transmission stirring mechanism 12, thereby providing stable support for stirring flotation to ensure accurate data collection and obtain accurate working condition analysis results.

[0037] like Figure 2 and Figure 3As shown, the flotation tank body 11 includes a flotation tank 111, an overflow tank 112 and a partition plate 113. The overflow tank 112 is arranged at the top of the flotation tank 111, and is specifically formed by an opening at the top of the flotation tank 111. Preferably, the flotation tank 111 is a cylinder with a U-shaped cross section formed by four side walls. The partition plate 113 divides the inner cavity of the flotation tank 111 into an upper chamber 1111 and a lower chamber 1112.

[0038] In order to enhance the flotation effect, Figure 3 , Figure 4 As shown in the figure, the flotation cell 11 further includes a stator baffle 114, and a plurality of stator baffles 114 are provided. The plurality of stator baffles 114 are evenly arranged in an annular shape, and the angles between adjacent stator baffles 114 are the same. The stator baffle 114 is arranged in the lower area of ​​the upper chamber 1111.

[0039] Preferably, the stator baffle 114 is a rectangular plate-shaped structure, one end of the stator baffle 114 is connected to the inner wall of the flotation tank 111, and the other end is suspended.

[0040] In this embodiment, stator baffles 114 are provided on the bottom inner wall of the flotation tank 111 . The stator baffles 114 are evenly distributed in a ring shape, that is, the evenly distributed stator baffles 114 are radially arranged around the central axis of the flotation tank 111 .

[0041] To further enhance the shearing effect, the stator baffle 114 is provided with first slots (not shown in the figure), and the first slots are evenly distributed along the height of the stator baffle 114. The first slots are bar-shaped holes, S-shaped holes, etc.

[0042] Considering the installation of the stator baffle 114, as Figure 3 , Figure 4 and Figure 5 As shown, the flotation tank body 11 further includes a baffle frame 115, and the baffle frame 115 includes a ring frame 1151 and a support column 1152. A plurality of support columns 1152 are provided, and the plurality of support columns 1152 are evenly arranged in a ring shape along the bottom of the ring frame 1151. Preferably, four support columns 1152 are provided, and the four support columns 1152 are arranged at the four corners of the bottom of the ring frame 1151. The bottom of the support column 1152 is connected to the partition plate 113. The ring frame 1151 is in contact with the inner wall of the flotation tank 111.

[0043] like Figure 3 As shown, the transmission stirring mechanism 12 includes a stirring impeller 121, a transmission shaft 122, a coupling 123 and a stepping motor 124. The stirring impeller 121 is arranged in the upper chamber 1111 and corresponds to the position of the stator baffle 114. The upper end of the transmission shaft 122 is connected to the stirring impeller 121, and the lower end of the transmission shaft 122 is connected to the output shaft of the stepping motor 124 through the coupling 123. The stepping motor 124 drives the transmission shaft 122 to rotate, thereby driving the stirring impeller 121 to rotate.

[0044] It can be understood that the transmission shaft 122 passes through the through hole on the partition plate 113 , and a tetrafluoroethylene sealing ring is provided at the connection between the through hole and the transmission shaft 122 to realize the rotational sealing of the transmission shaft 122 .

[0045] like Figure 3 , Figure 4 and Figure 5 As shown, the stirring impeller 121 includes an umbrella-shaped top 1211 and stirring blades 1212, and a plurality of stirring blades 1212 are provided, and the plurality of stirring blades 1212 are evenly arranged in a ring along the bottom of the umbrella-shaped top 1211. Preferably, the stirring blade 1212 is a rectangular sheet structure, and 4 stirring blades 1212 are provided. There is a gap between the outer side of the stirring blade 1212 and the stator baffle 114. The outer side of the stirring blade 1212 is flush with the edge of the umbrella-shaped top 1211, and there is a gap between the inner side of the stirring blade 1212 and the central axis of the umbrella-shaped top 1211. There is a gap between the bottom of the stirring blade 1212 and the partition plate 113.

[0046] Understandably, in order to enhance the shearing effect of the slurry, the stirring blade 1212 is provided with second slots (not shown in the figure), and the second slots are evenly distributed along the height of the stirring blade 1212. The second slots are strip-shaped holes, S-shaped holes, etc.

[0047] In this embodiment, an annular radial stator baffle 114 is provided around the stirring impeller 121, and the stirring blades 1212 of the stirring impeller 121 are located at the bottom of the umbrella-shaped top 1211 of the stirring impeller 121 in an annular radial arrangement. There are gaps between the stator baffle 114 and the stirring blades 1212, between the stirring blades 1212 and the central axis of the umbrella-shaped top 1211, and between the bottom of the stirring blades 1212, the bottom of the stator baffle 114 and the partition plate 113, which can enhance the shear force during the slurry stirring flotation process and provide a stable flotation environment for the slurry.

[0048] Considering that the flotation process requires air intake, such as Figure 2 and Figure 3 As shown, the flotation cell body 11 further includes an air inlet pipe 116, and two air inlet pipes 116 are provided. The two air inlet pipes 116 are symmetrically arranged on both sides of the flotation cell 111. An air inlet channel 1131 is provided in the partition plate 113, and correspondingly, two air inlet channels 1131 are also provided, which are respectively connected to the air inlet pipe 116. Preferably, the air inlet channel 1131 is L-shaped, one end of which is connected to the air inlet pipe 116, and the other end is connected to the upper chamber 1111. A one-way valve is provided on the air inlet pipe 116, and the other end of the air inlet pipe 116 is connected to an external air pump. The one-way valve allows the gas of the air pump to enter the upper chamber 1111.

[0049] like Figure 7As shown, the support base 13 includes a bottom shell 131 and a motor base 132. The motor base 132 is disposed in the bottom shell 131, and the stepper motor 124 is connected to the motor base 132. The output shaft of the stepper motor 124 passes through the top of the bottom shell 131 and is connected to the coupling 123.

[0050] It should be noted that the flotation unit 1 further includes a motion controller and an encoder, which are arranged in the bottom shell 131. The stepper motor 124 and the encoder are both connected to the motion controller, and the motion controller is used to servo-control the rotation speed of the stepper motor 124. The power switch and speed knob of the motion controller protrude from the front side wall of the bottom shell 131 for easy operation.

[0051] like Figure 6 As shown, the image acquisition unit 2 includes a flotation housing mechanism 21 and a concentrate collecting tank 22. The flotation housing mechanism 21 includes a box 211, a rectangular light source 212 and an annular shadowless light source 213. Two rectangular light sources 212 are provided. The two rectangular light sources 212 are respectively arranged on two opposite side walls of the box 211, and the annular shadowless light source 213 is arranged on the top of the box 211. The flotation unit 1 and the concentrate collecting tank 22 are both arranged in the box 211, and the concentrate overflowing from the overflow tank 112 enters the concentrate collecting tank 22.

[0052] It should be noted that if Figure 6 As shown, the rectangular light source 212 passes through a bracket 214 , and the bracket 214 is a locking hinge for adjusting the irradiation angle of the rectangular light source 212 .

[0053] Preferably, the box body 211 is a rectangular box body with a door. The interior of the box body 211 is evenly coated with black paint to prevent interference from external light sources.

[0054] like Figure 6 As shown, the image acquisition unit 2 also includes an industrial camera 23 and a light source controller. The industrial camera 23 is arranged in the middle of the annular shadowless light source 213 and passes through the annular shadowless light source 213, and is equipped with a variable zoom and aperture industrial lens to collect flotation foam images. The light source controller is used to adjust the color temperature, hue and brightness of the rectangular light source 212 and the annular shadowless light source 213, so as to adjust the optimal image sampling environment, reduce the influence of brightness, color temperature and hue on ash recognition, and highlight the foam texture characteristics.

[0055] Furthermore, the industrial camera 23 is arranged on a linear guide rail, so that the industrial camera 23 can move in the vertical direction. The vertical adjustment structure is not limited to the linear guide rail, and it only needs to be able to realize the vertical adjustment of the industrial camera 23.

[0056] Based on the above-mentioned flotation unit 1, normal flotation conditions and extreme flotation conditions are constructed, and multi-condition experiments are carried out to obtain flotation foam image data sets and multi-condition labels. Image features are extracted with the help of machine vision algorithms to extract flotation foam image color, size, texture, half-life, flow rate and other features. Image feature recognition and analysis use machine learning modeling to identify and classify the extracted features, and the results output through regression analysis include the grade of concentrate, the amount of reagent added and the entrainment rate. The model is trained through a multi-condition database to achieve accurate analysis of the foam image and output the corresponding conditions to achieve real-time online detection of various parameters of the slurry.

[0057] Specifically, the collected foam image sequence is subjected to image cropping, image enhancement and other preprocessing, and the images are subjected to CNN feature extraction to realize foam state recognition based on foam morphology, wherein the CNN network backbone is pre-trained with open source conventional data sets to reduce the demand for foam training sets; at the same time, the foam image is segmented based on the watershed algorithm, and the individual bubbles are analyzed to obtain information such as bubble size and centroid coordinates, and the change of bubble centroid is calculated through the front and back time indexes to obtain the bubble motion velocity field, and finally the foam state, bubble size, bubble velocity and color feature data are subjected to a nonlinear regression model based on concentrate grade and reagent dosage, thereby realizing flotation multi-condition recognition.

[0058] In this embodiment, the flotation unit 1 is arranged in the box 211, and the flotation foam is imaged by the industrial camera 23. The feature extraction and working condition analysis unit 3 extracts and analyzes the features of the collected image, thereby realizing flotation multi-working condition identification. There is no need to collect industrial images on site, and there will be no lagging relationship between the foam state and the flotation index, and there will be no distortion of data labels. Extreme working conditions can be created in the laboratory, the data is enriched, and accurate flotation data can be obtained, providing useful guidance for actual production.

[0059] It is worth noting that the system of this embodiment is a miniaturized system suitable for laboratory operation. It can construct multiple working conditions in the laboratory and extract and analyze flotation foam images. There is no need to collect data on site. It is easy to operate and can create extreme working conditions in the laboratory. The data is substantial and accurate.

[0060] Example 2

[0061] Another specific embodiment of the present invention is as follows Figure 8 As shown, a method for flotation foam image feature extraction and multi-condition identification is disclosed, using the flotation foam image feature extraction and multi-condition identification system of Example 1, and the steps include:

[0062] Step 1: Air tightness test.

[0063] Use pure water to test the sealing of the flotation cell 11. Pass a certain amount of pure water into the flotation cell 111, observe the air pipe connected to the air pump, adjust the gas flow rate in advance through the flow meter and judge whether the inflation device is well sealed to ensure that there is no external interference affecting the experimental results.

[0064] Step 2: Prepare samples and debug the light source and industrial camera 23.

[0065] For example, taking copper-nickel sulfide ore as an example, samples were prepared, copper-nickel sulfide ores of different grades were prepared in advance, the copper-nickel sulfide ore was crushed and ground, and the effects of different slurry concentrations, aeration volumes, and reagent dosages on the flotation foam state were explored.

[0066] Turn on the rectangular light source 212 and the annular shadowless light source 213, adjust the color temperature, hue and brightness, so as to adjust the best image sampling environment, reduce the influence of brightness, color temperature and hue on gray recognition, and highlight the foam texture characteristics. Start the industrial camera 23, set the frame rate, sensitivity, exposure time, white balance, sampling time and other parameters to ensure that the photos taken are clear.

[0067] Step 3: Place the flotation unit 1 into the box 211 and collect the flotation foam image.

[0068] Step 3.1: Add water and start stirring.

[0069] The flotation unit 1 that has passed the test flotation cell 11 is placed in the box 211, and then 1L of water is injected into the flotation cell 11, and the stepper motor 124 is started, with the rotation speed set at 1600 r / min.

[0070] Step 3.2: Add minerals and pH adjuster.

[0071] Add the weighed mineral (such as copper-nickel sulfide ore) and pH adjuster.

[0072] Step 3.3: After stirring for 60 seconds, add the collector.

[0073] Step 3.4: Add the foaming agent after stirring for 90 seconds.

[0074] Step 3.5: Inflate and collect flotation foam images.

[0075] After stirring for a while, the air pump is turned on to start inflation. At the same time, the industrial camera 23 is used to continuously take pictures during the flotation process, ensuring that the light source is stable during the shooting process, the time intervals of the picture sets are the same, and the collected pictures are stored in a storage device.

[0076] It should be noted that this step requires collecting flotation foam images under different operating parameters. Considering the factors of pulp concentration, aeration, collector dosage, frother dosage, and pH, preferably, the specific multi-operating parameters are shown in the following table.

[0077] Variable Name scope unit gradient Slurry concentration 5％-40％ — 5％ Inflation volume 0.2-1.0 <![CDATA[m 3 / h]]> 0.2 Collector dosage 0-200 g / t 20 Foaming agent dosage 0-50 g / t 10 pH 3-12 — 3

[0078] Step 4: Image feature extraction.

[0079] According to different times and working conditions of the flotation process, the flotation foam image area that needs to be identified is selected and segmented to obtain data set I.

[0080] After completing the image segmentation, the features of the segmented images are extracted, and the watershed algorithm is used to mark and segment the bubbles. The equivalent area circle radius of each segmented area is counted as the bubble size, and the coordinates of the bubble centroid are calculated. The overall foam image is divided into blocks, and the bubble flow field velocity is calculated through the graphic sequence of the coordinates of the bubble centroids.

[0081] The bubble state labels are assigned according to the foam morphology in the data set I. Preferably, the foam is classified into state labels such as "virtual", "real", "brittle", and "sticky". Classification training is performed using a convolutional neural network to obtain a convolutional network for identifying foam state classification. The backbone of the CNN network is pre-trained with open source conventional data sets such as imagene, and some layers are frozen to reduce the demand for foam training sets.

[0082] Step 5: Operating condition analysis.

[0083] The extracted bubble flow field velocity, bubble size distribution, bubble color, and bubble state label features obtained through convolutional neural network (CNN) are subjected to multi-factor nonlinear regression algorithms such as MLP and random forest to output corresponding working conditions such as concentrate grade, reagent addition evaluation and other parameters. The current working condition data is compared with the target working condition such as target concentrate grade and other information, and combined with the reagent addition evaluation, the dosage adjustment of the flotation process is guided.

[0084] In this embodiment, while the flotation process is completed in the box 211, the industrial camera 23 automatically obtains the flotation foam image, and the feature extraction and working condition analysis unit 3 extracts and analyzes the collected image features, thereby realizing flotation multi-working condition identification. There is no need to collect industrial images on site, and there will be no lagging relationship between the foam state and the flotation index, and there will be no distortion of data labels. Extreme working conditions can be created in the laboratory, the data is enriched, and accurate flotation data can be obtained, providing useful guidance for actual production.

[0085] This embodiment quantifies the multiphase characteristic indicators of foam morphology, size distribution, and bubble color respectively, and has a feature extraction method that is universal for various foams. For different types of minerals, it only needs to train the final nonlinear regression model with a small amount of data sets to obtain better working condition prediction results. Compared with the traditional "end-to-end" prediction model, the demand for data sets is greatly reduced.

[0086] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A flotation foam image feature extraction and multi-condition recognition system, characterized in that: It comprises a flotation unit (1), an image acquisition unit (2) and a feature extraction and working condition analysis unit (3), wherein the flotation unit (1) is used for flotation of ore pulp to generate flotation foam, the image acquisition unit (2) is used for collecting flotation foam images under different working conditions, and the feature extraction and working condition analysis unit (3) is used for extracting and analyzing the features of the flotation foam image; The image acquisition unit (2) comprises a flotation housing mechanism (21) and a concentrate collecting tank (22); the flotation housing mechanism (21) comprises a box body (211); the flotation unit (1) and the concentrate collecting tank (22) are both arranged in the box body (211).

2. The flotation foam image feature extraction and multi-condition identification system according to claim 1 is characterized in that: The flotation housing mechanism (21) further comprises a rectangular light source (212) and an annular shadowless light source (213) which are arranged in the box body (211).

3. The flotation foam image feature extraction and multi-condition identification system according to claim 2 is characterized in that: Two rectangular light sources (212) are provided, and the two rectangular light sources (212) are respectively arranged on two opposite side walls of the box body (211), and the annular shadowless light source (213) is arranged on the top of the box body (211).

4. The flotation foam image feature extraction and multi-condition identification system according to any one of claims 1 to 3, characterized in that: The flotation unit (1) comprises a flotation tank body (11), a transmission and stirring mechanism (12) and a support base (13); the upper part of the transmission and stirring mechanism (12) is connected to the flotation tank body (11), and the lower part of the transmission and stirring mechanism (12) is connected to the support base (13).

5. The flotation foam image feature extraction and multi-condition identification system according to claim 4 is characterized in that: The flotation cell body (11) comprises a flotation cell (111) and an overflow cell (112); the overflow cell (112) is arranged at the top of the flotation cell (111) and is in communication with the flotation cell (111).

6. The flotation foam image feature extraction and multi-condition identification system according to claim 5 is characterized in that: The flotation cell body (11) further comprises a partition plate (113) horizontally arranged in the flotation cell (111), wherein the partition plate (113) divides the inner cavity of the flotation cell (111) into an upper chamber (1111) and a lower chamber (1112).

7. The flotation foam image feature extraction and multi-condition identification system according to claim 6 is characterized in that: The flotation tank body (11) further comprises an air inlet pipe (116), two of which are provided, and the two air inlet pipes (116) are symmetrically arranged on both sides of the flotation tank (111).

8. The flotation foam image feature extraction and multi-condition identification system according to claim 7, characterized in that: An air intake channel (1131) is provided in the partition plate (113), one end of the air intake channel (1131) is in communication with the air intake pipe (116), and the other end of the air intake channel (1131) is in communication with the upper chamber (1111).

9. The flotation foam image feature extraction and multi-condition identification system according to claim 7, characterized in that: The air inlet pipe (116) is provided with a one-way valve.

10. A flotation foam image feature extraction and multi-condition identification method, characterized in that: The system described in any one of claims 1 to 9 is used to extract and analyze image features of flotation foam under different working conditions.

Images