Discrimination method of crushing and grinding energy based on mineral fracture morphology characteristics

Abstract

A method for determining crushing energy based on mineral fracture morphology is proposed. First, electron microscopy is used to scan magnetite ore particles of the same size after crushing at different energies, analyzing the relationship between intergranular fracture ratio and crushing energy. When the crushing energy reaches a certain value, the intergranular fracture ratio decreases, while the transgranular fracture ratio increases. Further, electron microscopy is used to scan magnetite ore particles of different sizes after crushing at the same energy, analyzing the relationship between the intergranular fracture ratio of different particle sizes and crushing energy. Then, a functional relationship is calculated to establish the relationship between the intergranular fracture ratio of different particle sizes after crushing a single magnetite ore particle and the crushing energy. Simultaneously, a functional relationship is also calculated to establish the relationship between the intergranular fracture ratio of different particle sizes after crushing a group of magnetite ore particles and the crushing energy. The correlation R of this function is then calculated. 2 With an index >0.95, a crushing energy discrimination method based on mineral fracture morphology is proposed, providing a quantifiable basis for further research on the crushing of magnetite ore and other minerals.

Description

Technical Field

[0001] This invention relates to the technical field of particle morphology analysis after ore crushing, and in particular to a method for determining crushing and grinding energy based on the fracture morphology characteristics of minerals. Background Technology

[0002] Crushing of ore is a prerequisite for sorting and smelting. Magnetite ore, as a raw material for steelmaking, exhibits both intergranular and transgranular fractures on the surface morphology of its particles after crushing. The crushing energy applied to the magnetite ore by the crushing equipment is the cause of ore crushing and the formation of its surface morphology. However, the specific relationship between crushing energy and the surface morphology of the crushed magnetite ore particles has not been thoroughly studied in the existing technology. More specifically, the relationship between the proportion of intergranular fractures after crushing and the crushing energy is not mentioned in the literature. The inventor believes it is necessary to establish a method for judging the relationship between the two to guide research on magnetite ore crushing. Summary of the Invention

[0003] The purpose of this invention is to solve the above problems and provide a method for determining grinding energy based on the fracture morphology characteristics of minerals.

[0004] The technical solution of this invention is: a method for determining grinding energy based on mineral fracture morphology characteristics, comprising the following steps:

[0005] a. The magnetite ore sample was repeatedly crushed using a heavy hammer tester to determine the critical energy for fracture as 100.00 J. The fracture morphology of the magnetite ore at this point was obtained by scanning electron microscopy. Determining the critical energy for ore fracture provides a basis for subsequent adjustments to different crushing energies. Electron microscopy observations confirmed that the fracture type of the magnetite ore at this point was primarily intergranular fracture.

[0006] b. Adjust the hammer tester and crush single magnetite ore particles using different crushing energies. Gradually increase or decrease the crushing energy, and obtain the fracture morphology of magnetite corresponding to each crushing energy using electron microscopy. Arrange and compare them sequentially to determine the crushing energy level at which the fracture type changes from a larger proportion of intergranular fracture to a larger proportion of transgranular fracture. By adjusting the crushing energy, observe the change in the proportion of intergranular fracture and explore the preliminary relationship between the proportion of intergranular fracture and the crushing energy.

[0007] c. Select single-particle magnetite ore particles crushed under the same crushing energy, classify the crushed magnetite ore particles by sieving, obtain fracture images of magnetite ore particles of different particle sizes by scanning electron microscopy, and count the proportion of intergranular fracture; after sieving, observe the proportion of intergranular fracture in particles of different particle sizes by electron microscopy, and the proportion of intergranular fracture gradually decreases as the particle size decreases.

[0008] d. Repeat step c multiple times, and successively analyze the fracture morphology of single magnetite ore particles of different particle sizes under different crushing energies in step b, and count the proportion of intergranular fracture; explore the relationship between different crushing energies and the proportion of intergranular fracture in particles of different particle sizes, and establish its functional relationship through quantitative correspondence.

[0009] e. Summarize the data from step d, plot the points on a coordinate system, and then, based on the plotted points and fitted curves, establish a functional relationship between the intergranular fracture ratio of different particle size grades of crushed single-particle magnetite ore and the crushing energy.

[0010] f. Following the process described in steps bd above, the magnetite ore particle clusters are crushed at different crushing energies. The crushed particles at each energy level are then classified, and fracture images of each particle size are obtained using electron microscopy. The proportion of intergranular fracture is then statistically analyzed. For magnetite ore particle clusters, the crushing and electron microscopy observation methods are the same as for single-particle magnetite ore, with the energy level continuously adjusted and repeated multiple times to obtain data on the proportion of intergranular fracture after particle cluster crushing.

[0011] g. Summarize the data from step f and establish a functional relationship between the intergranular fracture ratio of different particle size classes after crushing of the magnetite ore and the crushing energy. This function can then be used to guide the adjustment of crushing energy during ore crushing.

[0012] Preferably, the crushing energies selected in step b are 226.01 J, 376.69 J, 527.36 J, 640.37 J, 753.38 J, and 866.38 J, respectively. In step b, when the crushing energy is determined to be within the range of 226.01 J to 640.37 J, the proportion of intergranular fracture in the particle fracture type of magnetite ore is relatively large. When the crushing energy reaches 753.38 J, intergranular-transgranular coupled fracture occurs. When the crushing energy reaches 866.38 J, the proportion of transgranular fracture in the particle fracture type of magnetite ore is relatively large.

[0013] Preferably, in step c, the particle size classification of the crushed single-particle magnetite ore particles observed by electron microscopy is as follows: -1.180+0.900, -0.900+0.600, -0.600+0.300, -0.300+0.150, -0.150+0.074, -0.074+0.038, -0.038+0.019, in mm; the functional relationship between the intergranular fracture ratio and the crushing energy at different particle size levels in step e is as follows:

[0014] -1.180+0.900 y=227.353-28.279lnx R 2 =0.96;

[0015] -0.900+0.600 y=198.995-25.304lnx R 2 =0.96;

[0016] -0.600+0.300 y=109.849-13.597lnx R 2 =0.95;

[0017] -0.300+0.150 y=48.031-5.975lnx R 2 =0.95;

[0018] -0.150+0.074 y=27.055-3.185lnx R 2 =0.98;

[0019] -0.074+0.038 y=21.584-2.873lnx R 2 =0.95;

[0020] -0.038+0.019 y=9.662-1.209lnx R 2 =0.94;

[0021] In the above formulas, y represents the intergranular fracture ratio, x represents the fracture energy, and R... 2 This represents functional correlation.

[0022] Preferably, the crushing energies for the magnetite particle group in step g are 332.59 J, 432.36 J, 532.14 J, 631.25 J, and 731.69 J, respectively; the particle size distribution of the crushed particles observed by electron microscopy is: -1.180+0.900, -0.900+0.600, -0.600+0.300, -0.300+0.150, -0.150+0.074, -0.074+0.038, -0.038+0.019, in mm; the functional relationship between the intergranular fracture ratio of different particle size levels and the crushing energy is as follows:

[0023] -1.180+0.900 y=289.187-37.679lnx R 2 =0.98;

[0024] -0.900+0.600 y=244.894-31.551lnx R 2 =0.98;

[0025] -0.600+0.300 y=198.682-27.897lnx R 2 =0.99;

[0026] -0.300+0.150 y=64. 844-8.844lnx R 2 =0.96;

[0027] -0.150+0.074 y=36. 708-4.984lnx R 2 =0.98;

[0028] -0.074+0.038 y=33.116-4.660lnx R 2 =0.96;

[0029] -0.038+0.019 y=14.923-2.096lnx R 2 =0.95;

[0030] In the above formulas, y represents the intergranular fracture ratio, x represents the fracture energy, and R... 2 This represents functional correlation.

[0031] Preferably, in step b, particles with a median particle size range of -10.000 to +5.000 mm after being crushed by different crushing energies are selected for electron microscopy scanning. Selecting the median range corresponding to the particle size after being crushed by different crushing energies is more representative and comparable, and the comparison results are also more reasonable.

[0032] Preferably, in steps c and f, magnetite fracture morphology classification and recognition software is used. This software is built through deep learning and uses manually labeled images for training to obtain a usable model. This model can quickly achieve image recognition, improve image recognition efficiency, and increase the speed of steps c and f. The steps for building and applying this software are as follows;

[0033] 61. The magnetite ore is crushed using a crushing device. The crushed magnetite ore particles are collected and divided into N groups. Each group is spread out and scanned with an electron microscope. The N scanned images are imported into a computer, where N ≥ 1000 images.

[0034] 62. Manually calibrate the surface morphology of particles in each region of each scanned image. Different particle morphologies are calibrated as either intergranular fracture or transgranular fracture. The number of intergranular fractured particles and the number of transgranular fractured particles in the image are then calculated. This step is repeated to obtain N calibrated images.

[0035] 63. Build a fully convolutional neural network model based on the SSD algorithm in a computer to learn and train on the labeled images after step 62.

[0036] 64. Import most of the labeled images completed in step 62 into a fully convolutional neural network model based on the SSD algorithm on the computer for deep learning training. After deep learning training, an image recognition model is formed. Use the remaining labeled images to test the image recognition model. If the test accuracy meets the requirements, stop deep learning training and retain the trained model; if the test accuracy does not meet the requirements, continue deep learning training and testing until the test accuracy meets the requirements, and retain the trained image recognition model.

[0037] 65. Establish software in the computer to interface with the image recognition model obtained in step 64 above.

[0038] 66. After completing step c or f, collect the crushed ore particles that require particle morphology identification, spread them out, and take images using an electron microscope to generate images for identification.

[0039] 67. Import the photo to be identified into the above software, and then apply the image recognition model to perform image recognition. Output the number of transgranular fractures and intergranular fractures of the ore particles in the image, calculate the intergranular fracture ratio, and output the recognition and calculation data through the software for use in step e or g.

[0040] Preferably, in step 62, the manual calibration is performed based on a sample of the surface morphology image of the intergranular fractured ore particles. The sample in the image has a fracture surface with a rock candy-like pattern and clear edges and corners.

[0041] Preferably, in step 62, the manual calibration is performed based on a sample of the surface morphology image of the transcrystalline fractured ore particles. The sample of the image has a step pattern, a river pattern, a tear pattern, or a secondary crack pattern at the fracture surface.

[0042] Preferably, in step 64, 70% of the labeled images are used for deep learning training, and the remaining 30% of the labeled images are used to test the trained model.

[0043] Preferably, the accuracy rate of the compliant test in step 64 is not less than 94%.

[0044] The beneficial effects of this invention are as follows: This invention proposes a method for determining crushing energy based on the fracture morphology characteristics of minerals. First, electron microscopy is used to scan magnetite ore particles of the same size after crushing at different energies, analyzing the relationship between the proportion of intergranular fracture and the crushing energy. When the crushing energy reaches a certain value, the proportion of intergranular fracture decreases, while the proportion of transgranular fracture increases. Further, electron microscopy is used to scan magnetite ore particles of different sizes after crushing at the same energy, analyzing the relationship between the proportion of intergranular fracture at different particle sizes and the crushing energy. Then, a functional relationship is calculated to establish the relationship between the proportion of intergranular fracture at different particle sizes after crushing a single magnetite ore particle and the crushing energy. Simultaneously, a functional relationship is also calculated to establish the relationship between the proportion of intergranular fracture at different particle sizes after crushing a group of magnetite ore particles and the crushing energy. The correlation R of this function... 2 With an index >0.95, a crushing energy discrimination method based on mineral fracture morphology is proposed, providing a quantifiable basis for further research on the crushing of magnetite ore and other minerals.

[0045] In this judgment method, in order to quickly identify and judge the amount and proportion of intergranular fractures in crushed magnetite ore particles, a recognition system and software based on deep learning methods were established. This enables rapid identification and statistical analysis of the surface morphology of crushed magnetite ore particles, and quickly obtains the number and proportion of intergranular fractures, providing a faster prerequisite method and tool for establishing the above function. Attached Figure Description

[0046] Figure 1 This refers to step a of the determination method of the present invention, which describes the surface morphology of magnetite ore at critical fracture.

[0047] Figure 2 This refers to the fracture morphology of magnetite ore under different crushing energies in step b of the determination method of the present invention, wherein...

[0048] (a) shows the fracture morphology of magnetite ore at a fracture energy of 226.01 J;

[0049] (b) shows the fracture morphology of magnetite ore at a fracture energy of 376.69 J;

[0050] (c) shows the fracture morphology of magnetite ore at a fracture energy of 527.36 J;

[0051] (d) shows the fracture morphology of magnetite ore at a fracture energy of 640.37 J;

[0052] (e) shows the fracture morphology of magnetite ore with a fracture energy of 753.38 J;

[0053] (f) shows the fracture morphology of magnetite ore with a fracture energy of 866.38 J;

[0054] Figure 3This refers to the fracture morphology of magnetite ore of different particle sizes under the same crushing energy in step c of the judgment method of the present invention, wherein...

[0055] (a) shows the fracture morphology of magnetite ore with a grain size of -2.000 to +1.180 mm;

[0056] (b) shows the fracture morphology of magnetite ore with a grain size of -1.180 to +0.900 mm;

[0057] (c) shows the fracture morphology of magnetite ore with a grain size of -0.900 to +0.600 mm;

[0058] (d) shows the fracture morphology of magnetite ore with a grain size of -0.600 to +0.300 mm;

[0059] (e) shows the fracture morphology of magnetite ore with a grain size of -0.300 to +0.150 mm;

[0060] (f) shows the fracture morphology of magnetite ore with a grain size of -0.150 to +0.074 mm;

[0061] (g) is the fracture morphology of magnetite ore with a grain size of -0.074 to +0.038 mm;

[0062] (h) represents the fracture morphology of magnetite ore with a grain size of -0.038 to +0.019 mm;

[0063] Figure 4 This is a graph showing the relationship between the intergranular fracture ratio of different particle size levels and the crushing energy after single-particle magnetite ore is crushed, which is drawn by summarizing the data in step e of the judgment method of the present invention.

[0064] Figure 5 The graph in step g of the judgment method of this invention is a graph showing the relationship between the proportion of intergranular fracture at different particle size levels and the crushing energy after the particle group magnetite ore is crushed. Detailed Implementation

[0065] Example 1: See Figure 1-4 A method for determining grinding energy based on mineral fracture morphology characteristics includes the following steps:

[0066] a. The magnetite ore sample was repeatedly crushed using a heavy hammer tester to determine the critical energy for fracture as 100.00 J. The fracture morphology of the magnetite ore at this point was obtained by scanning electron microscopy. Determining the critical energy for ore fracture provides a basis for subsequent adjustments to different crushing energies; the crushing energy must not be lower than this critical energy. Electron microscopy observation confirmed that the fracture type of the magnetite ore at this point was primarily intergranular fracture.

[0067] b. Adjust the hammer crusher and crush single magnetite ore particles using different crushing energies, gradually increasing or decreasing the crushing energy. Obtain the fracture morphology of the magnetite corresponding to each crushing energy using electron microscopy, and compare them sequentially to determine the crushing energy level at which the fracture type changes from a larger proportion of intergranular fracture to a larger proportion of transgranular fracture. By adjusting the crushing energy, observe the change in the proportion of intergranular fracture to explore the qualitative relationship between the proportion of intergranular fracture and the crushing energy. In step b, select particles with a median particle size range of -10.000 to +5.000 mm after crushing at different energies for electron microscopy. Selecting the median particle size range corresponding to different crushing energies is more representative and comparable, and the comparison results are more reasonable. The crushing energies selected in step b are 226.01 J, 376.69 J, 527.36 J, 640.37 J, 753.38 J, and 866.38 J, respectively. In step b, it is determined that when the crushing energy is in the range of 226.01J-640.37J, the proportion of intergranular fracture in the particle fracture type of magnetite ore is relatively large. When the crushing energy reaches 753.38J, intergranular-transgranular coupled fracture occurs. When the crushing energy reaches 866.38J, the proportion of transgranular fracture in the particle fracture type of magnetite ore is relatively large.

[0068] c. Select single-particle magnetite ore particles crushed under the same crushing energy, classify the crushed magnetite ore particles by sieving, obtain fracture images of magnetite ore particles of different particle sizes by scanning electron microscopy, and count the proportion of intergranular fracture; after sieving, observe the proportion of intergranular fracture in particles of different particle sizes by electron microscopy. As the particle size decreases, the proportion of intergranular fracture gradually decreases.

[0069] d. Repeat step c multiple times, sequentially examining the fracture morphology of single magnetite ore particles from step b at different crushing energies and for different particle sizes, and statistically analyzing the proportion of intergranular fracture; explore the relationship between different crushing energies and the proportion of intergranular fracture in particles of different sizes, and establish a functional relationship through quantitative correspondence.

[0070] e. Summarize the data from step d, plot the points on the coordinate system, then fit a curve based on the plotted points and draw a plot. Figure 4 A functional relationship was established between the intergranular fracture ratio of different particle sizes after crushing of single-particle magnetite ore and the crushing energy.

[0071] In step c, the particle size classification of the crushed single-particle magnetite ore particles observed by electron microscopy is as follows: -1.180+0.900, -0.900+0.600, -0.600+0.300, -0.300+0.150, -0.150+0.074, -0.074+0.038, -0.038+0.019, in mm; The functional relationship between the intergranular fracture ratio and crushing energy for different particle size grades in step e is as follows:

[0072] -1.180+0.900 y=227.353-28.279lnx R 2 =0.96;

[0073] -0.900+0.600 y=198.995-25.304lnx R 2 =0.96;

[0074] -0.600+0.300 y=109.849-13.597lnx R 2 =0.95;

[0075] -0.300+0.150 y=48.031-5.975lnx R 2 =0.95;

[0076] -0.150+0.074 y=27.055-3.185lnx R 2 =0.98;

[0077] -0.074+0.038 y=21.584-2.873lnx R 2 =0.95;

[0078] -0.038+0.019 y=9.662-1.209lnx R 2 =0.94;

[0079] In the above formulas, y represents the intergranular fracture ratio, x represents the fracture energy, and R... 2 This represents functional correlation.

[0080] As attached Figure 4As shown, the proportion of intergranular fracture in magnetite ore particles gradually decreases with increasing crushing energy. With increasing crushing energy, the proportion of intergranular fracture decreases while the proportion of transgranular fracture increases. The proportion of intergranular fracture in different particle sizes shows a logarithmic decreasing relationship with the crushing energy, and the equation has a high good fit. Therefore, the crushing energy can be determined by the proportion of intergranular fracture. However, the proportion of intergranular fracture is relatively low in the -0.074 to +0.038 mm and -0.038 to +0.019 mm particle sizes, with transgranular fracture being the dominant type. Therefore, using the proportion of intergranular fracture in fine-grained particles to determine the crushing energy will result in a significant error.

[0081] The intergranular fracture ratio of the -1.1 80 + 0.900 mm particle size varies significantly under different crushing energies, and therefore can be used as a particle size for determining crushing energy.

[0082] In step c, magnetite fracture morphology classification and identification software is used. The steps for establishing and applying this software are as follows;

[0083] 61. The magnetite ore is crushed using a crushing device. The crushed magnetite ore particles are collected and divided into N groups. Each group is spread out and scanned with an electron microscope. The N scanned images are imported into a computer, where N ≥ 1000 images.

[0084] 62. Manually calibrate the surface morphology of particles in each region of each scanned image. Different particle morphologies are calibrated as either intergranular fracture or transgranular fracture. The number of intergranular fractured particles and the number of transgranular fractured particles in the image are then calculated. This step is repeated to obtain N calibrated images.

[0085] 63. Build a fully convolutional neural network model based on the SSD algorithm in a computer to learn and train on the labeled images after step 62.

[0086] 64. Import most of the labeled images completed in step 62 into a fully convolutional neural network model based on the SSD algorithm on the computer for deep learning training. After deep learning training, an image recognition model is formed. Use the remaining labeled images to test the image recognition model. If the test accuracy meets the requirements, stop deep learning training and retain the trained model; if the test accuracy does not meet the requirements, continue deep learning training and testing until the test accuracy meets the requirements, and retain the trained image recognition model.

[0087] 65. Establish software in the computer to interface with the image recognition model obtained in step 64 above.

[0088] 66. After completing step c or f, collect the crushed ore particles that require particle morphology identification, spread them out, and take images using an electron microscope to generate images for identification.

[0089] 67. Import the photo to be identified into the above software, and then apply the image recognition model to perform image recognition. Output the number of transgranular fractures and intergranular fractures of the ore particles in the image, calculate the intergranular fracture ratio, and output the recognition and calculation data through the software for use in step e or g.

[0090] In step 62, the manual calibration is performed based on the surface morphology image of the ore particles with intergranular fracture. The fracture surface of the sample in the image has a rock candy-like pattern and clear edges and corners.

[0091] In step 62, the manual calibration is performed based on the surface morphology image of the transcrystalline fractured ore particles. The fracture surface of the sample in the image has a step pattern, a river pattern, a tear pattern, or a secondary crack pattern.

[0092] In step 64, 70% of the labeled images are used for deep learning training, and the remaining 30% of the labeled images are used to test the trained model.

[0093] The accuracy rate of the test in step 64 is no less than 94%.

[0094] This software uses deep learning to build a usable model that is trained on manually labeled images. This model enables rapid image recognition, improving image recognition efficiency and speeding up step c.

[0095] The judgment method in this embodiment establishes a functional relationship between the intergranular fracture ratio and crushing energy of magnetite ore particles at different particle size levels after crushing by analyzing the relationship between the intergranular fracture ratio and crushing energy. In ore crushing production, this functional relationship can be used to infer the crushing energy through the intergranular fracture ratio, providing a quantifiable basis for further research on magnetite ore crushing.

[0096] Example 2: See Figure 5 Example 2 is basically the same as Example 1, and the similarities will not be repeated. The difference is that step f in Example 2 is a repetition of step bd in Example 1. The magnetite ore particles are crushed at different crushing energies, and the particles after crushing at each crushing energy are graded. Fracture images of each particle size are obtained by electron microscopy, and the proportion of intergranular fracture is statistically analyzed. For magnetite ore particles, the crushing and electron microscopy observation methods are the same as those for single-particle magnetite ore, with the energy level continuously adjusted and repeated multiple times to obtain data on the proportion of intergranular fracture after particle crushing.

[0097] g. Summarize the data from step f and establish a functional relationship between the intergranular fracture ratio of different particle size classes after crushing of the magnetite ore and the crushing energy. This function can then be used to guide the adjustment of crushing energy during ore crushing.

[0098] This embodiment can infer the crushing energy by the proportion of intergranular fracture, providing a quantifiable basis for further research on magnetite ore crushing.

Claims

1. A method for determining grinding energy based on mineral fracture morphology characteristics, comprising the following steps: a. The magnetite ore sample was repeatedly crushed using a heavy hammer tester to determine the critical energy for fracture of the magnetite ore as 100.00 J. The fracture morphology of the magnetite ore at this point was obtained by scanning electron microscopy. b. Adjust the heavy hammer tester and crush single magnetite ore particles with different crushing energies. Gradually increase or decrease the crushing energy and obtain the fracture morphology of magnetite corresponding to each crushing energy by scanning with an electron microscope. Arrange and compare them in sequence to determine the crushing energy level at which the fracture type changes from a larger proportion of intergranular fracture to a larger proportion of transgranular fracture. c. Select the particles after crushing a single magnetite ore particle under the same crushing energy, classify the crushed magnetite ore particles by screening, obtain fracture images of magnetite ore particles of different particle sizes by scanning electron microscopy, and count the proportion of intergranular fracture. d. Repeat step c multiple times, and successively analyze the fracture morphology of single magnetite ore particles of different particle sizes under different crushing energies in step b, and count the proportion of intergranular fracture. e. Summarize the data from step d, plot the points on a coordinate system, and then, based on the plotted points and fitted curves, establish a functional relationship between the intergranular fracture ratio of different particle size grades of crushed single-particle magnetite ore and the crushing energy. f. According to the above steps bd, the magnetite ore particles are crushed at different crushing energies, and the particles after crushing at each crushing energy are classified. Fracture images of each particle size class are obtained by scanning electron microscopy, and the proportion of intergranular fracture is statistically analyzed. g. Summarize the data from step f and establish a functional relationship between the intergranular fracture ratio of different particle size classes after crushing of the magnetite ore and the crushing energy.

2. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 1, characterized in that: The crushing energies selected in step b are 226.01 J, 376.69 J, 527.36 J, 640.37 J, 753.38 J, and 866.38 J, respectively.

3. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 1, characterized in that: In step c, the particle size classification of the crushed single-particle magnetite ore particles observed by electron microscopy is as follows: -1.180+0.900, -0.900+0.600, -0.600+0.300, -0.300+0.150, -0.150+0.074, -0.074+0.038, -0.038+0.019, in mm; The functional relationship between the intergranular fracture ratio and crushing energy for different particle size grades in step e is as follows: -1.180 + 0.900 y = 227.353 - 28.279 lnx R 2 =0.96; -0.900+0.600 y=198.995-25.304lnx R 2 =0.96; -0.600 + 0.300 y = 109.849 - 13.597lnx R 2 =0.95; -0.300 + 0.150 y = 48.031 - 5.975lnx R 2 =0.95; -0.150+0.074 y=27.055-3.185lnx R 2 =0.98; -0.074 + 0.038 y = 21.584 - 2.873lnx R 2 =0.95; -0.038 + 0.019 y = 9.662 - 1.209lnx R 2 =0.94; In the above formulas, y represents the intergranular fracture ratio, x represents the fracture energy, and R... 2 This represents functional correlation.

4. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 1, characterized in that: In step g, the crushing energies for the magnetite particle group were 332.59 J, 432.36 J, 532.14 J, 631.25 J, and 731.69 J, respectively. The particle size distribution of the crushed particles, as determined by electron microscopy, was: -1.180+0.900, -0.900+0.600, -0.600+0.300, -0.300+0.150, -0.150+0.074, -0.074+0.038, and -0.038+0.019 (unit: mm). The functional relationship between the intergranular fracture ratio and the crushing energy for different particle size levels is as follows: -1.180 + 0.900 y = 289.187 - 37.679 lnx R 2 =0.98; -0.900+0.600 y=244.894-31.551lnx R 2 =0.98; -0.600 + 0.300 y = 198.682 - 27.897lnx R 2 =0.99; -0.300+0.150 y=64. 844-8.844lnx R 2 =0.96; -0.150 + 0.074 y = 36.708 - 4.984lnx R 2 =0.98; -0.074+0.038 y=33.116-4.660lnx R 2 =0.96; -0.038+0.019 y=14.923-2.096lnx R 2 =0.95; In the above formulas, y represents the intergranular fracture ratio, x represents the fracture energy, and R... 2 This represents functional correlation.

5. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 1, characterized in that: In step b, particles with a median particle size range of -10.000 to +5.000 mm after being crushed by different crushing energies are selected for electron microscopy scanning.

6. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 1, characterized in that: In steps c and f, magnetite fracture morphology classification and identification software is used. The steps for establishing and applying this software are as follows; 61. The magnetite ore is crushed using a crushing device. The crushed magnetite ore particles are collected and divided into N groups. Each group is spread out and scanned with an electron microscope. The N scanned images are imported into a computer, where N ≥ 1000 images.

62. Manually calibrate the surface morphology of particles in each region of each scanned image. Different particle morphologies are calibrated as either intergranular fracture or transgranular fracture. The number of intergranular fractured particles and the number of transgranular fractured particles in the image are then calculated. This step is repeated to obtain N calibrated images.

63. Build a fully convolutional neural network model based on the SSD algorithm in a computer to learn and train on the labeled images after step 62.

64. Import most of the labeled images completed in step 62 into a fully convolutional neural network model based on the SSD algorithm on the computer for deep learning training. After deep learning training, an image recognition model is formed. Use the remaining labeled images to test the image recognition model. If the test accuracy meets the requirements, stop deep learning training and retain the trained model; if the test accuracy does not meet the requirements, continue deep learning training and testing until the test accuracy meets the requirements, and retain the trained image recognition model.

65. Establish software in the computer to interface with the image recognition model obtained in step 64 above.

66. After completing step c or f, collect the crushed ore particles that require particle morphology identification, spread them out, and take images using an electron microscope to generate images for identification.

67. Import the photo to be identified into the above software, and then apply the image recognition model to perform image recognition. Output the number of transgranular fractures and intergranular fractures of the ore particles in the image, calculate the intergranular fracture ratio, and output the recognition and calculation data through the software for use in step e or g.

7. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 6, characterized in that: In step 62, the manual calibration is performed based on the surface morphology image of the ore particles with intergranular fracture. The fracture surface of the sample in the image has a rock candy-like pattern and clear edges and corners.

8. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 6, characterized in that: In step 62, the manual calibration is performed based on the surface morphology image of the transcrystalline fractured ore particles. The fracture surface of the sample in the image has a step pattern, a river pattern, a tear pattern, or a secondary crack pattern.

9. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 6, characterized in that: In step 64, 70% of the labeled images are used for deep learning training, and the remaining 30% of the labeled images are used to test the trained model.

10. The grinding energy discrimination method based on mineral fracture morphology characteristics according to claim 6, characterized in that: The accuracy rate of the test in step 64 is no less than 94%.

Images