Analysis method and system suitable for relationship between particle morphology distribution characteristics and sphere diameter of ore grinding product

Abstract

The invention relates to a method and system for analyzing the relationship between particle morphology distribution characteristics and ball diameters of ore grinding products, magnetic iron ore particles are crushed and ground in a ball grinding machine by adopting steel balls with different sizes and single ball diameters, the crushed and ground particles are subjected to size grading, calculation and statistics are carried out by scanning intergranular fracture proportions in different size grading through an electron microscope, and the particle morphology distribution characteristics of the ore grinding products are obtained. Then establishing a function relationship between the intergranular fracture ratio of the grinded magnetite particles with different particle size levels and the diameter of the steel ball; it is indicated that an incidence relation exists between the intergranular fracture proportion and the diameter of the steel ball in the ball mill, so that the ball diameter can be further adjusted through the change of the intergranular fracture proportion of the particles generated by ore grinding, and the crushing and grinding process is optimized.

Description

Technical Field

[0001] The invention relates to the technical field of ball selection for ball mills for ore crushing and grinding, and in particular to an analysis method and system suitable for analyzing the relationship between particle morphology distribution characteristics and ball diameters of grinding products. Background Art

[0002] The ball mill is an indispensable crushing equipment in the ore crushing and grinding project. The ore particles are simultaneously impacted, sheared and ground in the mill. The ore particles will be crushed by multiple methods such as impact crushing and grinding crushing at the same time during the crushing process. When the ball mill is working, the steel balls inside it will impact and crush the ore particles by lifting them to a certain height. The movement of particles in the grinding process is complex. Part of the crushing energy generated by the steel balls is used for particle crushing, and most of it is used for energy consumption between steel balls and between internal liners. The crushing energy generated by the steel balls mainly comes from the weight of the steel balls themselves, and the weight is determined by the size of the steel balls, that is, the crushing energy of the steel balls is related to the diameter of the steel balls.

[0003] After the magnetite ore that meets the grinding grade is crushed by the ball mill, it forms magnetite ore particles with smaller particle size. The surface morphology of these particles includes intercrystalline fracture and transcrystalline fracture. By analyzing the relationship between the surface morphology of these particles and the diameter of the steel balls added to the ball mill, we can further analyze and study the law of ball diameter selection for the ball mill, and establish the basic conditions for the analysis plan for the ratio of different ball diameters. Summary of the invention

[0004] The purpose of the present invention is to solve the above problems and provide an analysis method for the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product.

[0005] The technical solution of the present invention is: a method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of a grinding product, comprising the following steps:

[0006] a. Use a ball mill to grind the magnetite ore that meets the grinding particle size. The diameters of multiple steel balls in the ball mill are unified into one size, with a total weight of m. The unified ball diameter size facilitates the comparison of the relationship between each size and the proportion of intergranular fracture;

[0007] b. Screen and grade the magnetite ore particles obtained after ball milling. The analysis after the graded particles can better understand the rules;

[0008] c. Scan the surface morphology of the particles in the ground magnetite particles of each particle size level by electron microscope, calculate and record the proportion of intergranular fractures in each particle size level;

[0009] d. Replace several steel balls with different diameters. The total weight of the steel balls after replacement is m. Repeat steps a, b, and c to record the intergranular fracture ratio of magnetite particles of each size level under different ball diameters. The total weight of each ball milling is basically the same. Under the same working conditions of the ball mill, the overall grinding energy of the steel balls on the magnetite particles can be controlled to be basically the same;

[0010] e. Based on the data obtained in step d, plot points and fit curves in the coordinate system to obtain the function between the ratio of intergranular fractures in magnetite particles of different particle sizes and the diameter of the steel ball. This function reflects the relationship between the ratio of intergranular fractures in magnetite particles and the ball diameter of the ball mill. It can be used to guide the selection of ball diameter and its matching scheme during ball mill grinding, and improve the efficiency and effect of ball milling.

[0011] Preferably, the ball mill in step a is an experimental conical ball mill, and the particle size of the magnetite entering the mill is -2.000 mm, with reference to the particles entering the mill of the industrial ball mill, which is conducive to providing guidance for industrial production.

[0012] Preferably, in the step a shown, the weight of the magnetite ore particles entering the mill is 500g, the water entering the mill is 300g, and the total weight m of the steel balls is 8010g-8110g. After the magnetite ore particles out of the mill are dehydrated, they are screened and graded. The change in the diameter of the steel balls will inevitably bring about a change in the number, and it is impossible to achieve the same total weight of steel balls of different diameters. Within this range, a slight deviation in the total weight is regarded as the same total weight.

[0013] Preferably, the diameters of the steel balls in step d are 10 mm, 13 mm, 16 mm, 19 mm, 22 mm, 25 mm, 28 mm, and 32 mm, respectively.

[0014] Preferably, the particle size classification in step b is respectively -2.000+1.180mm particle size; -1.180+0.900mm particle size; -0.900+0.600mm particle size; -0.600+0.300mm particle size; -0.300+0.150mm particle size; -0.150+0.074mm particle size; -0.074+0.038mm particle size; -0.038+0.019mm particle size.

[0015] Preferably, firstly, a steel ball with a certain intermediate ball diameter is used for ball milling. In this embodiment, the surface morphology of the magnetite ore of all particle sizes after ball milling is scanned by electron microscope. According to the observation and analysis of the scanned pictures, it is determined that the particle size and the -0.038+0.019mm particle size are not statistically analyzed in step c. Among them, the particle size of -2.000+1.180mm particle size is large, and the range of its morphology photographed by the scanning electron microscope is limited, and the photographed picture has no meaning for analysis. The morphology of the particles of -0.038+0.019mm particle size is basically transgranular fracture, which has no statistical significance.

[0016] Preferably, the function obtained in step e is as follows

[0017] -1.180+0.900mm y=144.650-32.808lnx R 2 =0.99;

[0018] -0.900+0.600mm y=136.493-33.139lnx R 2 =0.97;

[0019] -0.600+0.300mm y=95.786-23.279lnx R 2 =0.98;

[0020] -0.300+0.150mm y=59.051-13.509lnx R 2 =0.94;

[0021] -0.150+0.074mm y=50.921-11.972lnx R 2 =0.98;

[0022] -0.074+0.038mm y=25.205-5.761lnx R 2 =0.99.

[0023] In the above formulas, y is the intergranular fracture ratio, x is the diameter of the steel ball, and R 2 is the function dependency.

[0024] Preferably, the step c uses magnetite fracture morphology classification and recognition software to calculate and count the intergranular fracture ratios in each particle size level, and the steps for establishing and applying the software are as follows;

[0025] 81. The magnetite ore is crushed by crushing equipment, and the crushed magnetite ore particles are collected and evenly divided into N groups. Each group is flattened and scanned by electron microscope. The N scanned pictures are imported into the computer, N ≥ 1000 pictures.

[0026] 82. Manually calibrate the surface morphology of particles in each area of ​​each scanned image. Different particle morphologies are calibrated as one of the two fracture types: intergranular fracture or transgranular fracture. The number of intergranular fracture particles and transgranular fracture particles in the image are calculated. Repeat this step to obtain N calibrated images.

[0027] 83. Establish a fully convolutional neural network model based on the SSD algorithm in the computer for learning and training the calibrated images after step 82 is completed.

[0028] 84. Import most of the calibration images completed in step 82 into the fully convolutional neural network model based on the SSD algorithm on the computer for deep learning training. After the deep learning training, an image recognition model is formed. The image recognition model is tested with the remaining calibration images. If the test accuracy meets the requirements, the deep learning training is stopped and the trained model is retained; if the test accuracy does not meet the requirements, the deep learning training and testing are continued until the test accuracy meets the requirements, and the trained image recognition model is retained;

[0029] 85. The software is established in the computer and connected to the image recognition model obtained in step 84 above.

[0030] 86. In step c, the particles of each size level are spread out and photographed by electron microscope scanning to form photos to be identified.

[0031] 87. The photo to be identified is imported into the above software through software, and then the image recognition model is applied to perform image recognition, the number of transgranular fractures and intergranular fractures of the ore particles in the image is output, the proportion of intergranular fractures is calculated, and the identified and calculated data is output through the software.

[0032] Preferably, the manual calibration in step 82 is performed based on a sample of the surface morphology picture of the ore particles with intercrystalline fracture, and the fracture in the sample of the picture presents a rock candy-like pattern, and the fracture interface has clear edges and corners; the manual calibration in step 82 is performed based on a sample of the surface morphology picture of the ore particles with transcrystalline fracture, and the fracture in the sample of the picture presents a step pattern, or a river pattern, or a tear pattern, or a secondary crack pattern.

[0033] The recognition model of the software is established through deep learning, and an available model is obtained by learning and training with manually calibrated images. This model can quickly realize image recognition, improve image recognition efficiency, and increase the speed of step c.

[0034] Preferably, in step 84, 70% of the calibration images are used for deep learning training, and the remaining 30% of the calibration images are used for testing; the test accuracy rate that meets the requirements in step 84 is not less than 94%.

[0035] A system for analyzing the relationship between the morphology distribution characteristics of the particles of the grinding product and the ball diameter, the system is used to execute the above analysis method, the system at least comprises a scanning electron microscope, a computer, an ore particle fracture type identification device in the computer, and an analysis device for the relationship between the morphology distribution characteristics of the particles of the grinding product and the ball diameter in the computer,

[0036] The scanning electron microscope is used to perform step c, scanning and photographing the magnetite ore particles after ball milling to obtain a scanned image;

[0037] The computer is used to receive and store the scanned images of the scanning electron microscope, and is used to store the recognized images;

[0038] The identification device comprises

[0039] The basic operation module of the picture includes at least the sub-modules of reading, displaying, saving and reloading the picture.

[0040] The image recognition and classification module calls the image recognition model in step 84 to classify and locate the particles in the image.

[0041] Human-computer interaction module, used to manually select pictures for identification and classification;

[0042] The data display module is used to display the recognition and classification results of the image, including the location and classification of each magnetite particle target in the image, as well as the statistics of the number of intergranular fractures and transgranular fractures;

[0043] The analysis device comprises

[0044] A data collection module is used to receive data on the intergranular fracture ratio and transgranular fracture ratio of magnetite ore particles of various particle sizes at different ball diameters transmitted by the identification device;

[0045] The function generation module plots points in the coordinate system and fits the curve based on the collected data, establishes the function between the intergranular fracture ratio and the steel ball diameter in magnetite particles of different particle sizes, and completes step e. This function reflects the relationship between the fracture morphology distribution characteristics of the grinding product particles and the ball diameter, and can be used to guide the selection of the ball diameter and its matching scheme during ball mill grinding, thereby improving the efficiency and effect of ball milling.

[0046] The beneficial effects of the present invention are:

[0047] A method and system for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of a grinding product. The present invention uses steel balls of different sizes and single ball diameters to crush magnetite ore particles in a ball mill, and the crushed particles are subjected to particle size classification. The intergranular fracture ratios in different particle size classifications are calculated and counted by scanning with an electron microscope, and then a functional relationship between the intergranular fracture ratios of magnetite ore particles of different particle size grades after crushing and the steel ball diameter is established; indicating that there is a correlation between the intergranular fracture ratio and the diameter of the steel balls in the ball mill, so that the ball diameter can be further adjusted by changing the intergranular fracture ratio of particles produced by grinding to optimize the crushing and grinding process. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1-8 This is the surface morphology of magnetite particles of different particle sizes after grinding in a ball mill when the ball diameter is 16 mm;

[0049] in Figure 1 The particle size is -2.000+1.180mm; Figure 2 The particle size is -1.180+0.900mm; Figure 3 The particle size is -0.900+0.600mm. Figure 4 The particle size is -0.600+0.300mm. Figure 5 The particle size is -0.300+0.150mm. Figure 6 The particle size is -0.150+0.074mm; Figure 7 The particle size is -0.074+0.038mm; Figure 8 The particle size is -0.038+0.019mm.

[0050] Fig. 9 This is the data summary table after step d is completed;

[0051] Fig.10 A graph showing the relationship between the ratio of intergranular fracture of magnetite particles of different particle sizes and steel balls obtained after step e is completed;

[0052] Fig.11 Schematic diagram of the recognition status of the automatic particle morphology recognition device. DETAILED DESCRIPTION

[0053] Example 1: See Figure 1-11 , a method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of a grinding product, comprising the following steps:

[0054] a. Use a ball mill to grind the magnetite ore that meets the grinding particle size. The diameters of multiple steel balls in the ball mill are unified into one size, with a total weight of m. One size is convenient for comparing the relationship between each size and the proportion of intergranular fracture. ;

[0055] b. Screen and grade the magnetite ore particles obtained after ball milling; detailed analysis after particle size classification can better understand the rules;

[0056] c. Scan the surface morphology of the particles in the ground magnetite particles of each particle size level by electron microscope, calculate and record the proportion of intergranular fractures in each particle size level;

[0057] d. Replace several steel balls with different diameters. The total weight of the steel balls after replacement is m. The total weight of each ball milling is basically the same. Under the same working conditions of the ball mill, the grinding energy of the overall steel balls on the magnetite ore particles can be controlled to be basically the same. Repeat steps a, b, and c, and record the intergranular fracture ratio of magnetite ore particles of each size level under different ball diameters;

[0058] e. Based on the data obtained in step d, plot points and fit curves in the coordinate system to obtain the function between the proportion of intergranular fractures in magnetite particles of different particle sizes and the diameter of the steel ball. This function reflects the relationship between the proportion of intergranular fractures in magnetite particles and the ball diameter of the ball mill.

[0059] The ball mill in step a is an experimental conical ball mill, and the particle size of the magnetite entering the mill is -2.000 mm. Referring to the particles entering the industrial ball mill, it is helpful to provide guidance for industrial production.

[0060] In step a shown in the figure, the weight of magnetite ore particles entering the mill is 500g, the water entering the mill is 300g, and the total weight m of the steel balls is 8010g-8110g. The change in the diameter of the steel balls will inevitably lead to a change in the number of steel balls. It is impossible to achieve the same total weight of steel balls of different diameters. Within this range, a small total weight deviation is regarded as the same total weight. After removing the moisture from the magnetite ore particles after leaving the mill, they are screened and graded.

[0061] The diameters of the steel balls in step d are 10 mm, 13 mm, 16 mm, 19 mm, 22 mm, 25 mm, 28 mm, and 32 mm, respectively.

[0062] The particle size classification in step b is

[0063] -2.000+1.180mm particle size; -1.180+0.900mm particle size; -0.900+0.600mm particle size; -0.600+0.300mm particle size; -0.300+0.150mm particle size; -0.150+0.074mm particle size; -0.074+0.038mm particle size; -0.038+0.019mm particle size.

[0064] First, a steel ball with a certain intermediate diameter is used for ball milling. In this embodiment, a steel ball with a diameter of 16 mm is used for crushing and grinding. The surface morphology of the magnetite ore of all particle sizes after ball milling is scanned by electron microscope. According to the observation and analysis of the scanned pictures, see the attached Figure 1 , it is determined that in step c, the intergranular fracture ratio of the -2.000+1.180mm and -0.038+0.019mm grains is not counted. The -2.000+1.180mm grains are large in size, and the scanning electron microscope has a limited range of morphology, so the pictures taken are meaningless for analysis; the morphology of the -0.038+0.019mm grains is basically transgranular fracture, which has no statistical significance. .

[0065] The function obtained in step e is as follows

[0066] -1.180+0.900mm y=144.650-32.808lnx R 2 =0.99;

[0067] -0.900+0.600mm y=136.493-33.139lnx R 2 =0.97;

[0068] -0.600+0.300mm y=95.786-23.279lnx R 2 =0.98;

[0069] -0.300+0.150mm y=59.051-13.509lnx R 2 =0.94;

[0070] -0.150+0.074mm y=50.921-11.972lnx R 2 =0.98;

[0071] -0.074+0.038mm y=25.205-5.761lnx R 2 =0.99.

[0072] In the above formulas, y is the intergranular fracture ratio, x is the diameter of the steel ball, and R 2 is the function dependency.

[0073] The step c uses magnetite fracture morphology classification and recognition software to calculate and count the intergranular fracture ratios in each particle size level. The recognition model of the software is established through deep learning, and the available model is obtained by learning and training with manually calibrated pictures. The model can quickly realize picture recognition, improve picture recognition efficiency, and improve the speed of step c. The steps of establishing and applying the software are as follows;

[0074] 81. The magnetite ore is crushed by a crushing device, which can be a drop weight crushing test machine or a ball mill. The crushed magnetite particles are collected and evenly divided into N groups. Each group is flattened and then scanned by an electron microscope. The N scanned images are imported into a computer, N ≥ 1000 images.

[0075] 82. Manually calibrate the surface morphology of particles in each area of ​​each scanned image. Different particle morphologies are calibrated as one of the two fracture types: intergranular fracture or transgranular fracture. The number of intergranular fracture particles and transgranular fracture particles in the image are calculated. Repeat this step to obtain N calibrated images.

[0076] 83. Establish a fully convolutional neural network model based on the SSD algorithm in the computer for learning and training the calibrated images after step 82 is completed.

[0077] 84. Import most of the calibration images completed in step 82 into the fully convolutional neural network model based on the SSD algorithm on the computer for deep learning training. After the deep learning training, an image recognition model is formed. The image recognition model is tested with the remaining calibration images. If the test accuracy meets the requirements, the deep learning training is stopped and the trained model is retained; if the test accuracy does not meet the requirements, the deep learning training and testing are continued until the test accuracy meets the requirements, and the trained image recognition model is retained;

[0078] 85. The software is established in the computer and connected to the image recognition model obtained in step 84 above.

[0079] 86. In step c, the particles of each size level are spread out and photographed by electron microscope scanning to form photos to be identified.

[0080] 87. Import the photo to be identified into the above software through software, and then use the image recognition model to perform image recognition, output the number of transcrystalline fractures and intercrystalline fractures of the ore particles in the image, calculate the intercrystalline fracture ratio, and output the identified and calculated data through the software, see Fig.11 .

[0081] The manual calibration in step 82 is performed based on a sample of the surface morphology picture of the ore particles with intercrystalline fracture, in which the fracture surface in the sample is a rock candy-like pattern, and the fracture interface has clear edges and corners; the manual calibration in step 82 is performed based on a sample of the surface morphology picture of the ore particles with transcrystalline fracture, in which the fracture surface in the sample is a step pattern, or a river pattern, or a tear pattern, or a secondary crack pattern.

[0082] In step 84, 70% of the calibration images are used for deep learning training, and the remaining 30% of the calibration images are used for testing; the test accuracy rate that meets the requirements in step 84 is not less than 94%.

[0083] Embodiment 2: A system for analyzing the relationship between the morphology distribution characteristics of the particles of the grinding product and the ball diameter, the system is used to execute the analysis method in the above embodiment 1, the system at least comprises a scanning electron microscope, a computer, an ore particle fracture type identification device in the computer, and an analysis device for the relationship between the morphology distribution characteristics of the particles of the grinding product and the ball diameter in the computer,

[0084] The scanning electron microscope is used to perform step c, scan and photograph the magnetite ore particles after ball milling to obtain scanned images; according to the requirements of step d, the ore particles of different particle sizes are scanned respectively to obtain a large amount of image data;

[0085] The computer is used to receive and store the scanned images of the scanning electron microscope, and to store the recognized images; at the same time, the computer is also used to install the recognition device and the analysis device, that is, the recognition software and the analysis software, to realize the automation of the entire analysis process.

[0086] The identification device comprises

[0087] The basic operation module of the picture includes at least the sub-modules of reading, displaying, saving and reloading the picture.

[0088] Image recognition and classification module, which calls the image recognition model in step 84, classifies and locates the particles in the image, counts the morphological distribution, and summarizes the various fracture types in the image. Fig.11 ;

[0089] Human-computer interaction module, used to manually select pictures for identification and classification;

[0090] The data display module is used to display the recognition and classification results of the image, including the location and classification of each magnetite particle target in the image, as well as the statistics of the number of intergranular fractures and transgranular fractures;

[0091] The analysis device comprises

[0092] The data collection module is used to receive data on the intergranular fracture ratio and transgranular fracture ratio of magnetite ore particles of various particle sizes at different ball diameters transmitted by the identification device.

[0093] The function generation module plots points in the coordinate system and fits the curve based on the collected data, establishes the function between the intergranular fracture ratio of magnetite ore particles of different particle size levels and the steel ball diameter, and completes step e.

[0094] This function reflects the relationship between the fracture morphology distribution characteristics of the grinding product particles and the ball diameter. It can be used to guide the selection of ball diameter and its matching scheme during ball mill grinding, thereby improving the efficiency and effect of ball milling.

Claims

1. A method for analyzing the relationship between the particle morphology distribution characteristics and ball diameter of a grinding product, comprising the following steps: a. Use a ball mill to grind the magnetite ore that meets the grinding particle size. The diameters of multiple steel balls in the ball mill are unified into one size, and the total weight is m; b. Screening and grading the magnetite particles obtained after ball milling; c. Scan the surface morphology of the particles in the ground magnetite particles of each particle size level by electron microscope, calculate and record the proportion of intergranular fractures in each particle size level; d. Replace several steel balls with different diameters, the total weight of the steel balls after replacement is m, repeat steps a, b, and c, and record the intergranular fracture ratio of magnetite particles of each size level under different ball diameters; e. Based on the data obtained in step d, plot points and fit curves in the coordinate system to obtain the function between the intergranular fracture ratio of magnetite ore particles of different particle size grades and the steel ball diameter.

2. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 1, characterized in that: The ball mill in step a is an experimental conical ball mill, and the inlet particle size of the magnetite is -2.000 mm.

3. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 1, characterized in that: In step a shown, the weight of magnetite ore particles entering the mill is 500 g, the water entering the mill is 300 g, and the total weight m of the steel balls is 8010 g-8110 g. After the magnetite ore particles exiting the mill are dehydrated, they are screened and graded.

4. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of a grinding product according to claim 1, characterized in that: The diameters of the steel balls in step d are 10 mm, 13 mm, 16 mm, 19 mm, 22 mm, 25 mm, 28 mm, and 32 mm, respectively.

5. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 4, characterized in that: The particle size classification in the step b is respectively -2.000+1.180mm particle size; -1.180+0.900mm particle size; -0.900+0.600mm particle size; -0.600+0.300mm particle size; -0.300+0.150mm particle size; -0.150+0.074mm particle size; -0.074+0.038mm particle size; -0.038+0.019mm particle size.

6. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 5, characterized in that: First, a steel ball of a certain intermediate diameter is used for ball milling. The surface morphology of the crushed magnetite of all particle sizes after ball milling is scanned by electron microscope. Based on the observation and analysis of the scanned images, it is determined that the intergranular fracture ratio of the -2.000+1.180mm particle size and the -0.038+0.019mm particle size is not counted in step c.

7. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 6, characterized in that: The function obtained in step e is as follows -1.180+0.900mm y=144.650-32.808lnx R 2 =0.99; -0.900+0.600mm y=136.493-33.139lnx R 2 =0.97; <h2 style=";text-align:left;direction:ltr">-0.600+0.300mm y = 95.786-23.279lnx R<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> =0.98; -0.300+0.150mm y=59.051-13.509lnx R 2 =0.94; <h2 style=";text-align:left;direction:ltr">-0.150+0.074mm y = 50.921-11.972lnx R<h2 style=";text-align:left;direction:ltr"> 2 <h2 style=";text-align:left;direction:ltr"> =0.98; -0.074+0.038mm y=25.205-5.761lnx R 2 =0.99。 In the above formulas, y is the intergranular fracture ratio, x is the diameter of the steel ball, and R 2 is the function dependency.

8. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of a grinding product according to claim 1, characterized in that: In the step c, magnetite fracture morphology classification and recognition software is used to calculate and count the intergranular fracture ratios in each particle size level. The steps for establishing and applying the software are as follows:

81. The magnetite ore is crushed by crushing equipment, and the crushed magnetite ore particles are collected and evenly divided into N groups. Each group is flattened and scanned by electron microscope. The N scanned pictures are imported into the computer, N ≥ 1000 pictures.

82. Manually calibrate the surface morphology of particles in each area of ​​each scanned image. Different particle morphologies are calibrated as one of the two fracture types: intergranular fracture or transgranular fracture. The number of intergranular fracture particles and transgranular fracture particles in the image are calculated. Repeat this step to obtain N calibrated images.

83. Establish a fully convolutional neural network model based on the SSD algorithm in the computer for learning and training the calibrated images after step 82 is completed.

84. Import most of the calibration images completed in step 82 into the fully convolutional neural network model based on the SSD algorithm on the computer for deep learning training. After the deep learning training, an image recognition model is formed. The image recognition model is tested with the remaining calibration images. If the test accuracy meets the requirements, the deep learning training is stopped and the trained model is retained; if the test accuracy does not meet the requirements, the deep learning training and testing are continued until the test accuracy meets the requirements, and the trained image recognition model is retained; 85. Software is established in the computer to interface with the image recognition model obtained in step 84 above.

86. In step c, particles of each size level are spread out and photographed by electron microscope scanning to form photos to be identified.

87. The photo to be identified is imported into the above software through software, and then the image recognition model is applied to perform image recognition, the number of transgranular fractures and intergranular fractures of the ore particles in the image is output, the proportion of intergranular fractures is calculated, and the identified and calculated data is output through the software.

9. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 8, characterized in that: The manual calibration in step 82 is performed based on a sample of the surface morphology picture of the ore particles with intercrystalline fracture, in which the fracture surface in the sample is a rock candy-like pattern, and the fracture interface has clear edges and corners; the manual calibration in step 82 is performed based on a sample of the surface morphology picture of the ore particles with transcrystalline fracture, in which the fracture surface in the sample is a step pattern, or a river pattern, or a tear pattern, or a secondary crack pattern.

10. The method for analyzing the relationship between the particle morphology distribution characteristics and the ball diameter of the grinding product according to claim 8, characterized in that: In step 84, 70% of the calibration images are used for deep learning training, and the remaining 30% of the calibration images are used for testing; the test accuracy rate that meets the requirements in step 84 is not less than 94%.

11. An analysis system for the relationship between the morphology distribution characteristics of the particles of the grinding product and the ball diameter, the system is used to execute the analysis method in claim 8, the system at least comprising a scanning electron microscope, a computer, an ore particle fracture type identification device in the computer, and an analysis device for the relationship between the morphology distribution characteristics of the particles of the grinding product and the ball diameter in the computer, The scanning electron microscope is used to perform step c, scanning and photographing the magnetite ore particles after ball milling to obtain a scanned image; The computer is used to receive and store the scanned images of the scanning electron microscope, and is used to store the recognized images; The identification device comprises The basic operation module of the picture includes at least the sub-modules of reading, displaying, saving and reloading the picture. The image recognition and classification module calls the image recognition model in step 84 to classify and locate the particles in the image. Human-computer interaction module, used to manually select pictures for identification and classification; The data display module is used to display the recognition and classification results of the image, including the positioning and classification of each magnetite particle target in the image, and the statistics of the number of intergranular fractures and transgranular fractures; the analysis device includes A data collection module is used to receive data on the intergranular fracture ratio and transgranular fracture ratio of magnetite ore particles of various particle sizes at different ball diameters transmitted by the identification device; The function generation module plots points in the coordinate system and fits the curve based on the collected data, establishes the function between the intergranular fracture ratio of magnetite ore particles of different particle size levels and the steel ball diameter, and completes step e.

Images