A method for detecting particle size composition of transported ore particles considering bad posture

Abstract

The present invention provides a method for detecting the particle size composition of a transported ore particle group taking into account bad postures, comprising the following steps: vibrating feeding; image acquisition; image processing and segmentation; ore particle size calculation; target detection; posture-based particle size correction; and particle size composition calculation. The method of the present invention is aimed at the pre-processing link of ore particle processing. By adding a single camera during the feeding or product transportation process of the crushing / sorting link, and incorporating the shape proportion of the ore particles into the empirical formula for particle posture distribution, each ore particle is uniformly corrected, thereby realizing fast online and accurate particle size composition detection of ore particle groups.

Description

Technical Field

[0001] The invention relates to the technical field of material particle size detection, and in particular to a method for detecting the particle size composition of a transported ore particle group taking into account a bad posture. Background Art

[0002] Ore refers to a mineral aggregate from which useful components can be extracted or which itself has certain properties that can be utilized. It is composed of ore minerals (also called useful minerals) and temporarily useless gangue minerals. The particle size composition of ore particles is an important indicator of crushing effect, and it is also a key factor affecting the subsequent sorting effect. Traditional particle size detection (100-0.5mm) often uses manual screening, which is time-consuming and labor-intensive. With the rapid development of artificial intelligence and image processing technology, the application of machine vision technology in mineral processing has become a new trend. Relevant scholars have arranged cameras above the material conveyor belt to realize the particle size composition detection of the transported materials through image segmentation. Invention patent CN113240663A discloses a conveyor belt ore particle size detection method, which uses an image acquisition device to shoot mineral particles on the conveyor belt, and uses a segmentation algorithm that combines an improved HED network edge detection and a watershed algorithm based on distance transformation to segment the ore image, and then calculates the particle size composition of the particles. Due to the irregularity of the shape of ore particles, in the detection of ore flow particle size composition for the transportation link, some particles are in an unfavorable posture (such as sideways or upright), resulting in certain errors in the visual particle size composition measurement. Invention patent CN202510032791.4 provides an ore particle size detection device and method based on a flexible vibrating disk, which disperses the ore particles through a vibrating disk and classifies the particles by shape. Subsequently, particle size correction is performed for ore particles of different shapes. However, since multiple cameras need to be arranged to identify shape information, online detection cannot be achieved. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a method for detecting the particle size composition of a transported ore particle group taking into account the poor posture, integrating the empirical formula of particle posture distribution, and realizing rapid and accurate online particle size composition detection of ore particle groups, which has the advantages of high detection accuracy and fast detection speed.

[0004] The present invention adopts the following technical solution: a method for detecting the particle size composition of a transported ore particle group taking into account the bad posture, comprising the following steps:

[0005] Step 1: Vibration feeding

[0006] The electromagnetic vibrating feeder is used to evenly transport the ore materials from the silo to the conveyor belt with controllable frequency and amplitude through the vibration force generated by electromagnetic drive; it can effectively avoid the accumulation of ore materials during the transportation process, while ensuring the uniform distribution and proper posture of the materials;

[0007] Step 2: Image acquisition

[0008] Use an industrial camera to photograph the ore material on the conveyor belt and transmit it to the image processing system in the computer;

[0009] The image acquisition device consists of an industrial camera, which is mainly used to photograph ore materials on the conveyor belt. The industrial camera consists of two main parts: lens and interface. The lens part can adjust the focal length, aperture and viewing angle to adapt to different shooting distances, lighting environments and field of view requirements. The interface part is responsible for transmitting the captured image data to the computer through high-speed and stable electrical signals to ensure real-time transmission and integrity of the data.

[0010] Step 3: Image Processing and Segmentation

[0011] The image processing and segmentation device includes a computer, which is used to pre-process and segment the collected ore images: first, the collected images are filtered using a bilateral filtering algorithm to effectively remove noise while retaining the edge information of the ore particles, providing a clear image for subsequent processing; then, the DeepLabV3+ algorithm is used to segment the image and accurately extract individual ore particles;

[0012] Step 4: Calculation of ore particle size

[0013] According to the image of a single ore particle segmented by the image segmentation algorithm, the length l and width b of its minimum circumscribed rectangle are calculated, and the two-axis geometric mean diameter D is used to represent the particle size of a single particle, as shown in formula (4):

[0014] ;

[0015] Step 5: Object Detection

[0016] The useful ore and gangue in the ore particle group belong to different shape classifications due to their different chemical compositions. Therefore, the YOLO v11 target detection algorithm is used to classify each segmented ore particle in real time to distinguish between useful ore and gangue, and then determine the specific shape classification of the ore particles;

[0017] Step 6: Pose-based granularity correction

[0018] Based on the ore classification results of YOLO, determine the ore proportion coefficient R of each type of shape in formula (2) a , calculate R by formula (5)a :

[0019] ;

[0020] Where N is the number of all ore particles in the real-time image taken by the industrial camera, n a The number of ore particles of each shape in the real-time picture;

[0021] The corrected particle size value is obtained by formula (2);

[0022] ;

[0023] Where a is the shape of the ore particles, which can be divided into block, flake, long flake and column, P a is the probability of bad posture of various shapes of ore particles, R a is the proportion of various shapes in ore particles, K a is the particle size correction factor;

[0024] Among them: Block ore particles: P a 13.56%, K a is 1.113; flaky ore particles: P a 2.13%, K a 1.256; long flake ore particles: P a 2.65%, K a 1.197; columnar ore particles: P a 16.05%, K a is 1.182;

[0025] Step 7: Particle size composition calculation: Use the corrected individual ore particle size to calculate the particle size composition (volume distribution) of all particles in the captured image.

[0026] Furthermore, in step 5, the ore particle shape classification method is:

[0027] Measure the length l, width b, and height h of the ore particles, calculate their elongation E and flatness F using formula (3), and classify the ore shapes based on Table 1:

[0028] ;

[0029] Table 1 Classification of ore particle shapes

[0030] .

[0031] Furthermore, in step 6, the method for obtaining the particle size correction coefficient is:

[0032] First, the experimental device is installed: including a conveying belt, a feeding chute and a loading plate; then the useful minerals and gangue in the ore particles are collected, and multiple particles are selected respectively, and the above experimental device is used to simulate the belt transfer feeding chute before material transportation; then the distribution of the bad postures of the useful minerals and gangue particles in the ore particles falling on the loading plate after transfer and feeding is counted; finally, the particle sizes of the normal posture and the bad posture of the ore particles are counted, and the ratio of the two is calculated to obtain the particle size correction coefficient, as shown in formula (1):

[0033] ;

[0034] Where K a is the particle size correction coefficient, i is the i-th ore in the experiment, D i正常 With D i不良 are the normal posture projection particle size and the bad posture projection particle size of the i-th ore particle obtained by machine vision technology, and n is the total number of ores in the experiment.

[0035] Beneficial effect: The method of the present invention is aimed at the pretreatment stage of ore particle processing. By adding a single camera during the feeding or product transportation process of the crushing / sorting stage, and incorporating the shape ratio of the ore particles into the empirical formula of the particle posture distribution, each ore particle is uniformly corrected, thereby realizing rapid online and accurate particle size composition detection of ore particle groups. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a structural schematic diagram of the particle posture prediction and particle size detection and correction system of the present invention;

[0037] Figure 2 It is a structural schematic diagram of the online detection system for the particle size composition of transported ore particles of the present invention;

[0038] Figure 3 is a flow chart of the method of the present invention;

[0039] Figure 4 This is an example diagram of ore particle segmentation of the present invention;

[0040] Figure 5 It is a particle size distribution diagram of the comparative results of the present invention. DETAILED DESCRIPTION

[0041] The following is a detailed description of the process of using the method for detecting the particle size composition of a transported ore particle group taking into account the bad posture of the present invention in conjunction with the accompanying drawings.

[0042] like Figure 2As shown, the method for detecting the particle size composition of a group of transported ore particles taking into account the bad posture of the present invention comprises a vibrating feeding device 1, which uniformly conveys the ore material from the silo to the conveyor belt with a controllable frequency and amplitude to avoid the accumulation problem of the ore material and ensure the uniform distribution and proper posture of the material; a conveyor belt 2, which is used to transport ore particles; an image acquisition device 3, which is used to photograph the ore particles on the conveyor belt and transmit the image to a computer; and a particle size composition measuring device 4, which is used to calculate the particle size of the ore particles.

[0043] In this embodiment, for the coal preparation process, the transported materials include clean coal (useful minerals, blocky) and gangue (useless minerals, columnar) columnar gangue.

[0044] like Figure 3 As shown, the ore image processing and analysis process of the present invention includes the following steps:

[0045] 1. Vibration feeding

[0046] The electromagnetic vibrating feeder is used to evenly transport the ore materials from the silo to the conveyor belt with controllable frequency and amplitude through the vibration force generated by electromagnetic drive. It can effectively avoid the accumulation of ore materials during the transportation process, while ensuring the uniform distribution and proper posture of the materials.

[0047] 2. Image Acquisition

[0048] Use an industrial camera to shoot the ore material on the conveyor belt and transmit it to the image processing system in the computer.

[0049] 3. Image processing and segmentation module

[0050] First, the bilateral filtering algorithm is used to filter the collected image, effectively removing noise while retaining the edge information of the ore particles, providing a clear image for subsequent processing. Then, the DeepLabV3+ algorithm is used to segment the image and accurately extract individual ore particles. Figure 4 shown.

[0051] Through this series of processing steps, the image processing and segmentation module can accurately extract key information from ore images and provide reliable data support for further analysis and processing.

[0052] 4. Calculation of ore particle size

[0053] According to the image of a single ore particle segmented by the image segmentation algorithm, the length l and width b of its minimum circumscribed rectangle are calculated, and the two-axis geometric mean diameter D is used to represent the particle size of a single particle, as shown in formula (4):

[0054] .

[0055] 5. Object Detection

[0056] In this embodiment, for the coal preparation process, the transported materials include clean coal (useful minerals, blocky) and gangue (useless minerals, columnar) columnar gangue. The useful ore and gangue in the ore particle group often belong to ores of different shape classifications due to their different chemical compositions. Therefore, the YOLO v11 target detection algorithm is used to perform real-time classification of each segmented ore particle, distinguish useful ore and gangue minerals, and then determine the specific morphological classification of the ore particles.

[0057] 6. Granularity Correction

[0058] Based on the ore classification results of YOLO, Ra is calculated by formula (5):

[0059] ;

[0060] Where N is the number of all ore particles in the real-time image taken by the industrial camera, n a The number of ore particles of each shape in the real-time picture;

[0061] The method for obtaining the particle size correction coefficient is as follows: first, the experimental device is installed: including the conveyor belt 5, the feed chute 6 and the loading plate 7 (such as Figure 1 Then, the useful minerals and gangue in the ore particles are collected, and multiple particles are selected respectively. The above experimental device is used to simulate the belt transfer feeding chute before material transportation; then the distribution of the bad postures of the useful minerals and gangue particles in the ore particles falling on the carrier plate after transfer and feeding is counted; finally, the particle sizes of the normal posture and the bad posture of the ore particles are counted, and the ratio of the two is calculated to obtain the particle size correction coefficient, as shown in formula (1):

[0062] ;

[0063] Where K a is the particle size correction coefficient, i is the i-th ore in the experiment, D i正常 With D i不良 are the normal posture projection particle size and the bad posture projection particle size of the i-th ore particle obtained by machine vision technology, and n is the total number of ores in the experiment;

[0064] Calculate the correction value of the ore particles. The correction formula of this embodiment is as follows:

[0065] ;

[0066] Wherein R1 represents the proportion of blocky ore particles in all ore particles in the real-time picture, and R2 represents the proportion of columnar ore particles in all ore particles.

[0067] 7. Verify the results

[0068] The ore materials used in this experiment are 13mm-25mm block clean coal and columnar gangue. The particle size detected by the particle size detection method of the present invention is compared with the uncorrected original visual segmentation particle size and the manual screening particle size to evaluate the accuracy of the present invention. The results are shown in Table 2:

[0069] Table 2 Comparison of average particle size

[0070] .

[0071] The error proportions in different ranges are further statistically analyzed, as shown in Table 3.

[0072] Table 3 Error ratio

[0073] .

[0074] As can be seen from the table above, the maximum error is 2.3mm, indicating that the detection method has high accuracy and can achieve online real-time measurement. This high-precision detection method has significant advantages over traditional particle size detection technologies such as screening and floating methods. In addition, the particle size distribution diagram is shown in Figure 5 shown.

[0075] The applicant declares that the above is only a specific implementation of the present invention, but the protection scope of the present invention is not limited thereto. The technicians in the technical field should understand that any changes or substitutions that can be easily thought of by the technicians in the technical field within the technical scope disclosed by the present invention are within the protection scope and disclosure scope of the present invention.

Claims

1. A method for detecting the particle size composition of transported ore particles taking into account the bad posture, characterized in that: The following steps are involved: Step 1: Vibration feeding The electromagnetic vibrating feeder is used to evenly transport the ore materials from the silo to the conveyor belt with controllable frequency and amplitude through the vibration force generated by electromagnetic drive; it can effectively avoid the accumulation of ore materials during the transportation process, while ensuring the uniform distribution and proper posture of the materials; Step 2: Image acquisition Use an industrial camera to photograph the ore material on the conveyor belt and transmit it to the image processing system in the computer; The image acquisition device consists of an industrial camera, which is used to photograph the ore materials on the conveyor belt. The industrial camera consists of two main parts: the lens and the interface. The lens part can adjust the focal length, aperture and viewing angle to adapt to different shooting distances, lighting environments and field of view requirements. The interface part is responsible for transmitting the captured image data to the computer through high-speed and stable electrical signals to ensure real-time transmission and integrity of the data. Step 3: Image Processing and Segmentation The image processing and segmentation device includes a computer, which is used to pre-process and segment the collected ore images: first, the collected images are filtered using a bilateral filtering algorithm to effectively remove noise while retaining the edge information of the ore particles, providing a clear image for subsequent processing; then, the DeepLabV3+ algorithm is used to segment the image and accurately extract individual ore particles; Step 4: Calculation of ore particle size According to the image of a single ore particle segmented by the image segmentation algorithm, the length l and width b of its minimum circumscribed rectangle are calculated, and the two-axis geometric mean diameter D is used to represent the particle size of a single particle, as shown in formula (4): ； Step 5: Object Detection The useful ore and gangue in the ore particle group belong to different shape classifications due to their different chemical compositions. Therefore, the YOLO v11 target detection algorithm is used to classify each segmented ore particle in real time to distinguish between useful ore and gangue, and then determine the specific shape classification of the ore particles; Step 6: Pose-based granularity correction Based on the ore classification results of YOLO, determine the ore proportion coefficient R of each type of shape in formula (2) a , calculate R by formula (5) a : ； Where N is the number of all ore particles in the real-time image taken by the industrial camera, n a The number of ore particles of each shape in the real-time picture; The corrected particle size value is obtained by formula (2); ； Where a is the shape of the ore particles, which can be divided into block, flake, long flake and column, P a is the probability of various shapes of ore particles appearing in bad postures, R a is the proportion of various shapes in ore particles, K a is the particle size correction factor; Among them: Block ore particles: P a 13.56%, K a is 1.113; flaky ore particles: P a 2.13%, K a 1.256; long flake ore particles: P a 2.65%, K a 1.197; columnar ore particles: P a 16.05%, K a is 1.182; Step 7: Particle size composition calculation: Use the corrected particle size of a single ore particle to calculate the particle size composition of all particles in the captured image.

2. A method for detecting particle size composition of transported ore particles taking into account bad posture according to claim 1, characterized in that: In step 5, the ore particle shape classification method is: Measure the length l, width b, and height h of the ore particles, calculate their elongation E and flatness F using formula (3), and classify the ore shapes based on Table 1: ； Table 1 Classification of ore particle shapes 。 3. A method for detecting particle size composition of transported ore particles taking into account bad posture according to claim 2, characterized in that: In step 6, the method for obtaining the particle size correction coefficient is: First, the experimental device is installed: including a conveying belt, a feeding chute and a loading plate; then the useful minerals and gangue in the ore particles are collected, and multiple particles are selected respectively, and the above experimental device is used to simulate the belt transfer feeding chute before material transportation; then the distribution of the bad postures of the useful minerals and gangue particles in the ore particles falling on the loading plate after transfer and feeding is counted; finally, the particle sizes of the normal posture and the bad posture of the ore particles are counted, and the ratio of the two is calculated to obtain the particle size correction coefficient, as shown in formula (1): ； Where K a is the particle size correction coefficient, i is the i-th ore in the experiment, D i正常 With D i不良 are the normal posture projection particle size and the bad posture projection particle size of the i-th ore particle obtained by machine vision technology, and n is the total number of ores in the experiment.

Images