A method for synergistic calcium reduction and quality improvement of magnesite

Abstract

This invention relates to a method for synergistic calcium reduction and quality improvement of magnesite. The method first establishes the intrinsic relationship between each ore particle and the grade of various chemical components, including MgO, based on the differences in surface texture, luster, roughness, and color of the various minerals in the raw magnesite ore. Standard samples of magnesite concentrate, tailings, and intermediate products are then selected, thereby establishing an artificial intelligence (AI) pre-selection model. The AI ​​pre-selection model is used to intelligently pre-select and remove tailings from coarse-grained magnesite. The intermediate products obtained from the intelligent pre-selection undergo reverse flotation for desilication, forward flotation for calcium reduction, and magnetic separation for iron reduction to obtain magnesite concentrate and tailings. This method leverages the synergistic effect of intelligent pre-selection and traditional flotation for calcium reduction, overcoming the limitations of flotation for calcium reduction and significantly reducing the calcium content of the magnesite concentrate product. This method has the advantages of simple operation, low production cost, and high efficiency.

Description

Technical Field

[0001] This invention relates to a method for reducing calcium content and improving the quality of magnesite, specifically a mineral processing technology that combines intelligent sorting and magnetic levitation for calcium reduction. Background Technology

[0003] Due to the increasing depletion of high-quality resources, current industrial magnesite beneficiation processes must reduce the MgO grade of the raw ore to >43%, while still requiring a CaO grade of <1%. This is because a CaO grade >1% leads to low concentrate yield, high impurity content, and low product prices, failing to meet the minimum economic requirements for industrial production. Despite this, most magnesite beneficiation plants in the market currently only manage to break even, with low profit margins or even losses. The reason for this lies in the fact that magnesium-calcium separation in magnesite beneficiation remains a significant industry challenge. Current technologies for magnesium extraction and calcium reduction still have many shortcomings, resulting in a large amount of low-grade magnesite resources remaining unutilized. However, according to relevant statistics, under current technological conditions, low-grade magnesite resources, which cannot be industrially applied due to low MgO or high CaO grades, account for more than 60% of the total reserves. These are generally stripped and piled up as waste rock during mining. This not only results in a huge waste of resources but also has a serious impact on the environment. Therefore, the efficient and high-value development and utilization of low-grade magnesite resources has become an urgent problem to be solved.

[0004] The main factor restricting the development and utilization of low-grade magnesite is the high calcium impurity content in the ore, which cannot meet the raw material requirements for producing high-end refractory materials and other magnesium-based materials. Therefore, it is necessary to reduce the calcium impurity content through beneficiation methods. However, in this flotation system, the floatability of magnesium and calcium-bearing minerals is similar and their intergrowth relationship is complex, making calcium reduction very difficult. In recent years, there has been considerable research on magnesite desilication, mainly focusing on the separation of magnesite from silica-containing minerals such as quartz, chlorite, and talc, and some progress has been made. Compared with desilication technology, the research progress on magnesite calcium reduction technology has been very slow, especially in industrial applications, where no substantial breakthroughs have been achieved. Existing technologies suffer from low yields, large reagent consumption, and poor technical and economic indicators, making it difficult for most beneficiation plants designed with magnesite calcium reduction processes to operate normally. Currently, the recovery rate of the valuable component MgO in magnesite beneficiation plants involving calcium reduction operations at home and abroad is generally only 30-40%. Due to the low MgO recovery rate, many beneficiation plants have chosen to abandon the positive flotation process for calcium reduction and instead directly purchase ores with very low CaO grades for production.

[0005] The difficulty in reducing the calcium content of magnesite lies in the fact that magnesite and calcium-bearing gangues such as dolomite are both carbonate minerals, and the standard samples of the two minerals, such as... Figure 1As shown, the crystal forms are consistent, with identical anions and partially identical cations, and the properties of calcium and magnesium are extremely similar, making it difficult to achieve selective inhibition of dolomite and selective collection of magnesite. Furthermore, no reagent has been publicly reported to achieve effective separation of magnesium and calcium in industrial applications; in other words, current reagents lack selectivity, and their inhibitory effect on magnesite is comparable to that on dolomite, thus limiting the effectiveness of flotation separation. Currently, in addition to flotation, methods for reducing calcium in magnesite include acid leaching and carbonation. While the latter two methods achieve higher calcium removal rates, they suffer from complex processes, long workflows, high costs, and environmental risks. Summary of the Invention

[0006] In view of the shortcomings of the above-mentioned technology, the purpose of this invention application is to provide a method for synergistic calcium reduction and quality improvement of magnesite. This method starts from the perspective of improving the feeding of traditional magnetic levitation combined desilication and calcium reduction process, introduces a new generation of intelligent pre-selection technology, and forms a synergistic effect with reverse flotation and magnetic separation to significantly reduce the calcium content of magnesite concentrate products.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows.

[0008] A method for synergistic calcium reduction and quality improvement of magnesite, the method comprising the following steps:

[0009] Step 1: Crush the magnesite ore to obtain medium crushed (-120mm) or fine crushed (-30mm) products; screen the crushed products to obtain magnesite ore particles of -8mm size, which are retained for reverse flotation and magnetic separation operations.

[0010] Step 2: Select +8mm magnesite ore particles from the crushed product of Step 1, and rinse them with water for 20-50 minutes to remove any soft contamination from other impurities on their surface.

[0011] Step 3: Using a combination of manual sorting, microscopic detection and chemical analysis, we explored the correspondence between the appearance of different particles of +8mm magnesite ore and the grade of various chemical components, including MgO, contained in the particles in terms of mineral genetic characteristics such as surface texture, gloss, roughness and color. We identified standard samples of magnesite concentrate, tailings and intermediate products, and established a sorting model based on the genetic characteristics of these standard samples using intelligent sorting technology.

[0012] Step 4: Determine the upper limit of the particle size for selection in the intelligent pre-selection stage based on the particle size of the standard sample. Let's assume it's Amm. This upper limit is the minimum particle size at which most monomers of magnesite minerals can be liberated.

[0013] Step 5: Correct the model parameters through multiple sorting tests until the sorting model can support the production of products that meet specific performance requirements.

[0014] Step 6: The magnesite samples with a particle size of 8-A mm are subjected to intelligent sorting in batch production to obtain magnesite concentrate, tailings and intermediate products. The grade of each valuable component is analyzed and the recovery rate is calculated.

[0015] Step 7: Combine the intermediate product from Step 6 with the -8mm magnesite ore from Step 1, and perform reverse flotation desilication, forward flotation calcium reduction, and magnetic separation iron reduction to obtain magnesite concentrate and tailings. Perform grade analysis on each valuable component and calculate the recovery rate.

[0016] Step 8: Combine the magnesite concentrate and tailings from Steps 6 and 7 to obtain the final sorting product.

[0017] Furthermore, step three is detailed below:

[0018] 1) The "manual selection + microscopic inspection" method is adopted. Standard samples are selected and classified according to the differences in appearance parameters such as surface texture, gloss, roughness and color. The categories include magnesite concentrate, tailings and intermediate products.

[0019] 2) Obtain the intrinsic quality of each standard sample through chemical analysis and process mineralogical analysis, and correlate the standard samples with the intrinsic quality;

[0020] 3) Collect 360-degree full-shape stereo images of each standard sample, label the categories, and construct a dataset of all standard samples;

[0021] 4) Train the constructed deep learning model based on the dataset to obtain a sorting model for intelligent classification of samples.

[0022] In step four, the value of A is typically set to 20–35 mm.

[0023] Step five is as follows:

[0024] 1) For each sorting test, the test samples are intelligently classified using the sorting model obtained in step 3 to obtain classified test samples, and chemical and process mineralogical analyses are performed on the test samples.

[0025] 2) When the analysis and testing results meet the specific index requirements, the sorting model at this time shall be used as the final sorting model; otherwise, repeat step 3 to update the sorting model until the analysis and testing results meet the specific index requirements.

[0026] Furthermore, the microscopic detection described in step three includes two or more methods such as optical microscopy, scanning electron microscopy with energy dispersive spectroscopy, and electron probe microanalysis.

[0027] As a further improvement of the present invention, the selected particle size range of 8 to A mm should be designed to ensure that the magnesite is liberated as much as possible, because this is a prerequisite for effective separation in mineral processing. The liberation of the magnesite is characterized by the degree of liberation of the magnesite, which is 80% to 100%. The value of A is generally set to 20 to 35 mm.

[0028] As a further improvement of the present invention, the method for determining that there are no other impurities or soft adhering to the surface of the ore is to soak the ore for 10 to 20 minutes, sieve it with an 8 mm sieve, and the yield of the undersize material is not higher than 0.1%.

[0029] As a further improvement of the present invention, if the value of A / 8 is ≥3, then n particle size values ​​need to be added from 8 to A, namely A1, A2, A3...An. The purpose is to ensure that the ratio of the larger value to the smaller value of two adjacent particle size values ​​is <3, thereby ensuring that there is no obvious obstruction phenomenon during sorting on the sorting track and improving the sorting accuracy.

[0030] As a further improvement of this invention, techniques such as appearance and morphology photography, color and texture differentiation, and surface roughness detection are used to explore the differences in mineral genetic characteristics between magnesite and calcium-containing impurities (such as dolomite and apatite). Through system autonomous learning, the sorting threshold is determined and a model for accurate identification is established, thereby achieving precise sorting that is difficult to achieve with traditional chemical detection and analysis.

[0031] As a further improvement of the present invention, the reverse flotation reagent is mainly composed of hydrocarbons, alcohols and amines, the positive flotation reagent is mainly composed of fatty acids, petroleum sulfonates and oxidized paraffin soap, and the magnetic separation adopts a high gradient pulse strong magnetic separator with a field strength of 1.5T and above.

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] 1. This invention introduces a new generation of intelligent pre-selection technology, which differs from traditional X-ray beneficiation (mainly used for metal ore separation, utilizing differences in specific gravity) and color sorting (based solely on grayscale differences, resulting in either black or white). X-ray separation relies on differences in the sensitivity of materials to X-rays for identification and sorting, primarily used for the pre-sorting of metal ores. However, magnesite is mainly composed of non-metallic minerals, and the differences in the absorption and reflection capabilities of the minerals to X-rays are relatively small, resulting in poor photoelectric pre-selection test results. Color sorting is based on significant color differences between particles, establishing differences based on the varying degrees of reflection of a particular light source on the mineral surface. While color sorting can identify significant differences between black and white, it cannot distinguish between black and dark red, or yellow and yellowish-green, as their reflectivity is similar. The intelligent pre-selection in this invention overcomes the limitation of conventional flotation methods in achieving calcium reduction in magnesite by identifying subtle differences in surface texture, gloss, roughness, and color between magnesite and calcium-bearing minerals such as dolomite. This invention significantly reduces the calcium content of magnesite concentrate through the synergistic effect of a new generation of intelligent pre-selection and reverse flotation. The method has the advantages of simple operation, low production cost, and being green and efficient.

[0034] 2. The method of this invention achieves calcium reduction and quality improvement at the coarse-grained level, improving calcium reduction efficiency on the one hand and reducing the grinding volume of subsequent flotation processes on the other, saving grinding costs and increasing economic benefits. By constructing a model through screening standard samples, intelligent pre-selection and waste disposal of magnesite ore is achieved, improving the feed grade of subsequent conventional processes and facilitating cost reduction and efficiency improvement in subsequent flotation and magnetic separation operations. Flotation calcium reduction requires a high degree of difference in calcium and magnesium floatability. For calcium that exists in an isomorphous form, it is difficult to achieve magnesium-calcium separation through flotation. This invention makes intelligent pre-selection calcium reduction technology possible by exploring the differences in mineral gene characteristics between magnesite and calcium-containing minerals such as dolomite. This helps overcome the shortcomings of flotation calcium reduction capacity, achieving complementary advantages from the perspective of technological innovation, and exerting a synergistic effect in calcium reduction and quality improvement. Attached Figure Description

[0035] Figure 1 The appearance and morphology of dolomite and magnesite standard samples.

[0036] Figure 2 The appearance morphology of some typical samples selected in the first stage of the artificial intelligence pre-selection model building process.

[0037] Figure 3 The images show the appearance and morphology of three types of standard samples in the artificial intelligence pre-selection model: pre-enriched concentrate, intermediate products, and pre-disposal tailings.

[0038] Figure 4 This is a process flow diagram of reverse flotation-direct flotation-magnetic separation in Example 2.

[0039] Figure 5 This is a process flow diagram of reverse flotation-direct flotation-magnetic separation in Example 3. Detailed Implementation

[0040] Example 1

[0041] The establishment of the artificial intelligence pre-selection model is as follows:

[0042] 300 kg of magnesite ore from a certain region in Xinjiang was taken, with MgO grade of 45.01%, CaO grade of 1.36%, SiO2 grade of 1.52%, and Fe2O3 grade of 0.85%. It was crushed to -30 mm using a crusher. The crushed product was screened, and the 8-30 mm product obtained by screening was rinsed with water for 40 minutes to remove any soft adhering contaminants from the surface.

[0043] In the raw ore within the particle size range of 8–30 mm, a first manual sorting of ore samples (2000 particles combined into one standard sample) was conducted based on the appearance of the ore product, including surface texture, gloss, roughness, and color. These samples were used as concentrate and tailings standards, respectively. Microscopic testing (optical microscope and scanning electron microscope + energy dispersive spectroscopy) and chemical analysis revealed (analysis results are shown in Table 1) that the SiO2 and Fe2O3 grades in both the concentrate (manual standard sample) and tailings (manual standard sample) met the impurity content requirements for both the concentrate and tailings standards. However, the grades of MgO and CaO in the concentrate were still very close to those in the raw ore. The selected standard samples showed poor separation effect for MgO and CaO, and the CaO grade in the concentrate standard samples needed to be further reduced.

[0044] Table 1. Chemical analysis results of manual standards in the 18–30 mm particle size range.

[0045]

[0046] Based on the particle size and test results of the selected standard samples, and considering further reducing the particle size of the standard samples to increase the liberation degree of each mineral, the upper limit of the particle size of the standard samples selected in the intelligent pre-selection stage was further reduced to 20 mm, and the standard samples were re-selected. Through a combination of "manual selection + microscopic detection + chemical analysis," the various standard samples required by the artificial intelligence sorting model were determined. First, a second selection of standard samples was conducted based on the differences in appearance parameters such as surface texture, gloss, roughness, and color of the ore products. The selected typical samples and the chemical analysis test results are as follows: Figure 2 As shown in Table 2.

[0047] Table 28 shows the chemical analysis results of some typical samples selected in the first batch of products in the 20mm particle size range.

[0048]

[0049]

[0050] The chemical and process mineralogical analyses of the second batch of standard samples showed that some standard samples already met the separation index requirements for concentrate, tailings, and intermediate products. Therefore, the upper limit of particle size for the intelligent pre-selection stage was determined to be 20 mm. Repeating the above process, through 30 rounds of "manual selection + microscopic observation + analysis," the target morphology of the intelligent pre-selection concentrate, intermediate product, and tailings standard samples gradually became clear. Based on the differences in appearance and the results of chemical analysis, three types of standard samples were selected: pre-enriched concentrate, intermediate product, and pre-disposal tailings. Each type of standard sample was required to have at least 2000 particles. Figure 3 As shown in Table 3, the chemical analysis results of the three types of standard samples are as follows.

[0051] The pre-enriched concentrate standard sample is characterized by a rough appearance, pure white color, and a grid-like texture. The MgO grade is ≥46%, while the SiO2, CaO, Fe2O3, and Al2O3 grades are all ≤0.66%.

[0052] The standard sample of pre-disposal tailings is tailings with a relatively smooth appearance, grayish-white color, and no obvious texture. It is required that the MgO grade is ≤40.4%, the SiO2 grade is ≥1%, the CaO grade is ≥4%, and the Fe2O3 and Al2O3 grades are all ≥0.45%.

[0053] Apart from the two types mentioned above, all other particles are used as intermediate product models.

[0054] Table 3 Chemical content of each component in standard samples of pre-enriched concentrate, intermediate products, and pre-disposal tailings

[0055]

[0056] A 360-degree full-shape stereoscopic image of each standard sample was acquired using an image acquisition system, and the categories were labeled to construct a dataset of all standard samples.

[0057] Finally, a deep learning model is built and trained based on the dataset to obtain a sorting model for intelligent classification of samples.

[0058] Magnesite from a certain location in Xinjiang was used, with an MgO grade of 45.01% and a CaO grade of 1.36%. The ore was mined from the mine, crushed to a particle size of -20mm ("-" represents less than), and mixed and separated to obtain two parallel ore samples, 200kg each, which were used for magnesite upgrading in Examples 2 and 3, respectively.

[0059] Example 2

[0060] The first parallel ore sample was processed using the technical route proposed in this invention, which involves using artificial intelligence pre-selection combined with reverse flotation-direct flotation-magnetic separation to desilicate, reduce calcium, and upgrade magnesite. First, the raw ore was crushed and screened at 8mm to obtain -8mm and 8-20mm products. The -8mm product was retained for reverse flotation and magnetic separation. The 8-20mm ore was then subjected to AI pre-selection.

[0061] Based on the artificial intelligence pre-selection model established in Example 1, an artificial intelligence sorting machine was used to perform AI pre-selection on 8-20mm ore. All 8-20mm ore particles were sorted into concentrate 1, intermediate product and calcium-bearing tailings. The grade of each valuable component was analyzed and the recovery rate was calculated. The results are shown in Table 4. The calcium-bearing tailings were directly treated as waste, concentrate 1 was directly used as product, and the intermediate product and the -8mm ore obtained from the previous screening were combined as feed to carry out the reverse flotation-direct flotation-magnetic separation process.

[0062] Table 48-20mm ore products artificial intelligence sorting test results

[0063]

[0064] The process flow of reverse flotation-direct flotation-magnetic separation is as follows: Figure 4 As shown, the specific process is as follows:

[0065] The intermediate product obtained by AI sorting and the -8mm ore product obtained by screening the raw ore are combined as feed material for reverse flotation-direct flotation-magnetic separation.

[0066] The feedstock is ball-milled to a fineness of approximately 80% (-0.074mm particles by mass). Water is added to adjust the slurry to a solids content of 34%, which is then transferred to a mixing tank. Sulfuric acid (800g / t relative to the feedstock) and water glass (400g / t relative to the feedstock) are added to adjust the pH to 6.0. Then, reverse flotation collector BK433 (160g / t relative to the feedstock) is added for further slurry adjustment. The adjusted slurry then enters the reverse flotation system for roughing stage I. The resulting frothy product is reverse flotation tailings I, which is thickened in a thickener. BK433 (75g / t relative to the feedstock) is added to the product in the roughing stage I for further slurry adjustment, followed by reverse flotation roughing stage II. The resulting froth product flows into the reverse flotation scavenging tank for reverse flotation scavenging. Reverse flotation collector BK433 (50 kg / t relative to the feed material) is added to the roughing tank for slurry conditioning, and reverse flotation roughing III is performed. The resulting froth product flows into the scavenging tank and is combined with the froth product from reverse flotation roughing II for reverse flotation scavenging. Water glass (50 g / t relative to the feed material) and reverse flotation collector BK433 (25 g / t relative to the feed material) are added to the scavenging tank for reverse flotation scavenging. The froth product obtained in the scavenging tank is reverse flotation tailings II, which is combined with reverse flotation tailings I to form total reverse flotation tailings, which enters the thickening bucket for thickening. The product in the reverse flotation scavenging tank is returned to the reverse flotation roughing II operation. The product in the reverse flotation roughing III tank is used as feed for direct flotation.

[0067] The direct flotation feed is fed into the direct flotation roughing agitator. Sodium carbonate (300 g / t relative to the feed) and sodium hexametaphosphate (400 g / t relative to the raw ore) are added to adjust the pH of the pulp to 9.5. Then, direct flotation collector BK434 (800 g / t relative to the feed) is added to condition the pulp, and direct flotation roughing is performed. The tailings from the direct flotation roughing are retained for direct flotation scavenging. The resulting direct flotation roughing froth product undergoes three cleaning processes, specifically as follows:

[0068] Sodium hexametaphosphate (100 g / t relative to the raw ore) and BK434 collector (200 g / t relative to the feed) are added to the froth product from the roughing flotation to adjust the slurry for direct flotation cleaning I. Sodium hexametaphosphate (60 g / t relative to the feed) and BK434 collector (120 g / t relative to the feed) are added to the tailings from direct flotation cleaning I to adjust the slurry for direct flotation scavenging. The scavenged tailings are designated as direct flotation tailings 2. The froth product obtained from direct flotation scavenging is returned to the roughing operation. Sodium hexametaphosphate (50 g / t relative to the feed) and BK434 collector are further added to the froth product obtained from direct flotation cleaning I. 4. (The dosage is 100g / t relative to the feed material) slurry is prepared and subjected to direct flotation cleaning II. The tailings obtained from direct flotation cleaning II are returned to direct flotation cleaning I. Sodium hexametaphosphate (20g / t relative to the feed material) and direct flotation collector BK434 (40g / t relative to the feed material) are added to the froth product obtained from direct flotation cleaning II for further slurry preparation, and direct flotation cleaning III is performed. The tailings obtained from direct flotation cleaning III are returned to direct flotation cleaning II. The froth product obtained from direct flotation cleaning III enters the high-intensity magnetic separation operation. Under the conditions of magnetic field strength of 1.7T and pulse of 125 times / min, magnetic products (magnetic tailings) and non-magnetic products (concentrate 2) are obtained.

[0069] The process of scavenging the rougher tailings of the direct flotation is as follows: sodium hexametaphosphate (80 g / t relative to the raw ore) and direct flotation collector BK434 (160 g / t relative to the feed material) are added to the rougher tailings of the direct flotation to adjust the slurry, and direct flotation scavenging I is carried out. The tailings obtained from direct flotation scavenging I are returned to the rougher operation of direct flotation. Direct flotation collector BK434 (60 g / t relative to the feed material) is added to the froth product obtained from direct flotation scavenging I to adjust the slurry, and direct flotation scavenging II is carried out. The froth product obtained from direct flotation scavenging II is returned to the rougher operation. The tailings obtained from direct flotation scavenging II, namely tailings 1, are combined with direct flotation fine scavenging tailings 2 to form the total direct flotation tailings for discharge and stockpiling.

[0070] Concentrate 1 and Concentrate 2 were combined to form the final magnesite concentrate. The grades of each component in the product, as determined by testing and calculation, are shown in Table 5.

[0071] Table 5 shows the combined indicators of concentrate 1 and concentrate 2 for the final magnesite concentrate product.

[0072] Process route MgO grade, % CaO grade, % MgO recovery rate, % The process of this invention 47.36 0.34 65.34

[0073] Example 3

[0074] The second parallel ore sample was processed using a traditional method, namely reverse flotation for desilication, direct flotation for calcium reduction, and magnetic separation for iron removal. Figure 5 As shown.

[0075] The raw ore is first crushed to obtain a product with a particle size of -3mm.

[0076] The feedstock is ball-milled to a fineness of approximately 80% (-0.074 mm by mass). Water is added to adjust the slurry to a solids content of 34%, which is then transferred to a mixing tank. Sulfuric acid (800 g / t relative to the feedstock) and water glass (400 g / t relative to the feedstock) are added to adjust the pH to 6.0. Then, reverse flotation collector BK433 (150 g / t relative to the feedstock) is added for further slurry adjustment. The adjusted slurry then enters the reverse flotation system for roughing stage I. The resulting froth product is reverse flotation tailings I, which is thickened in a thickener. BK433 (40 g / t relative to the feedstock) is then added to the reverse flotation roughing stage I for further slurry adjustment, followed by reverse flotation roughing stage II. The resulting foam product flows into a scavenging tank for reverse flotation scavenging. Reverse flotation collector BK433 (30 g / t relative to the feed material) is added to the reverse flotation roughing II tank for slurry conditioning, and reverse flotation roughing III is performed. The resulting foam product flows into a scavenging tank and is combined with the foam product from reverse flotation roughing II for reverse flotation scavenging. Water glass (50 g / t relative to the feed material) and reverse flotation collector BK433 (25 g / t relative to the feed material) are added to the scavenging tank for reverse flotation scavenging. The foam product obtained in the scavenging tank is reverse flotation tailings II, which is combined with reverse flotation tailings I to form total reverse flotation tailings for discharge and stockpiling. The product in the reverse flotation scavenging tank is returned to the reverse flotation roughing II operation. The product obtained from reverse flotation roughing III is used as feed material for direct flotation.

[0077] The feedstock for direct flotation flows into the direct flotation stirred tank. Sodium carbonate (300 g / t relative to the feed) and sodium hexametaphosphate (400 g / t relative to the raw ore) are added to adjust the pH of the pulp to 9.5. Then, direct flotation collector BK434 (800 g / t relative to the feed) is added to condition the pulp, and direct flotation roughing is performed. The tailings from the direct flotation roughing are retained for direct flotation scavenging. The resulting froth product from the direct flotation roughing undergoes four cleaning processes, specifically as follows:

[0078] Sodium hexametaphosphate (100 g / t relative to the raw ore) and BK434 collector (200 g / t relative to the feed) are added to the froth product from the roughing stage of direct flotation for slurry preparation, followed by direct flotation cleaning I. Sodium hexametaphosphate (60 g / t relative to the feed) and BK434 collector (120 g / t relative to the feed) are added to the tailings from direct flotation cleaning I for scavenging and finishing, and the tailings are designated as direct flotation tailings 2. The froth product obtained from the scavenging and finishing stages is returned to the roughing stage. Sodium hexametaphosphate (50 g / t relative to the feed) and BK434 collector (100 g / t relative to the feed) are added to the froth product from direct flotation cleaning I for slurry preparation, followed by direct flotation cleaning II. The tailings from direct flotation cleaning II are returned to the direct flotation cleaning I stage. The froth product obtained from direct flotation II is further treated with sodium hexametaphosphate (20 g / t relative to the feed) and direct flotation collector BK434 (40 g / t relative to the feed) for slurry conditioning, and then subjected to direct flotation III. The tailings obtained from direct flotation III are returned to direct flotation II. The froth product obtained from direct flotation III is further treated with sodium hexametaphosphate (20 g / t relative to the feed) and direct flotation collector BK434 (40 g / t relative to the feed) for slurry conditioning, and then subjected to direct flotation IV. The tailings obtained from direct flotation IV are returned to direct flotation III. The froth product obtained from direct flotation IV is then subjected to high-intensity magnetic separation, where magnetic products (magnetic tailings) and non-magnetic products (magnesite concentrate) are obtained under conditions of a magnetic field strength of 1.7 T and a pulse rate of 125 times / min.

[0079] The final magnesite concentrate product has the following grades: (see Table 6)

[0080] The process of scavenging the rougher tailings of the direct flotation is as follows: sodium hexametaphosphate (80 g / t relative to the raw ore) and direct flotation collector BK434 (160 g / t relative to the feed material) are added to the rougher tailings of the direct flotation to adjust the slurry, and direct flotation scavenging I is carried out. The tailings obtained from direct flotation scavenging I are returned to the rougher operation of direct flotation. Direct flotation collector BK434 (60 g / t relative to the feed material) is added to the froth product obtained from direct flotation scavenging I to adjust the slurry, and direct flotation scavenging II is carried out. The froth product obtained from direct flotation scavenging II is returned to the rougher operation. Tailings 1 and direct flotation tailings 2 obtained from direct flotation scavenging II are combined into the total direct flotation tailings and discharged and stockpiled.

[0081] Table 6. Final Magnesite Concentrate Product Indicators

[0082] Process route MgO grade, % CaO grade, % MgO recovery rate, % conventional process 46.20 0.66 59.65

[0083] A comparison of the results in Tables 5 and 6 shows that the process of this invention not only significantly improves the MgO grade and recovery rate in the concentrate, but also reduces the CaO grade by nearly 50%, significantly increasing the added value of the product. According to the magnesite standard (YB / T5208-2016), the product grade is upgraded from M46A to M47B. After subsequent calcination and purification, the unit price of the latter product (high-purity fused magnesite) will be 3000 yuan / ton higher than that of the former (general fused magnesite). Furthermore, the intelligent pre-selection + reverse flotation synergistic calcium reduction and quality improvement process for magnesite helps reduce the number of subsequent flotation calcium reduction cleaning steps, improving both the separation index and efficiency.

Claims

1. A method for synergistic calcium reduction and upgrading of magnesite, characterized by, Includes the following steps: Step 1: Crush the magnesite ore to obtain medium crushed (-120mm) or fine crushed (-30mm) products. Screen the crushed products to obtain magnesite ore particles of -8mm size, which are then retained for reverse flotation and magnetic separation operations. Step 2: Select +8mm magnesite ore particles from the crushed product of Step 1, and rinse them with water for 20-50 minutes to remove any soft contamination from other impurities on their surface. Step 3: Using a combination of "manual selection + microscopic detection + chemical analysis", we explored the correspondence between the appearance of different particles of +8mm magnesite ore and the grade of various chemical components, including MgO, contained in the particles in terms of mineral genetic characteristics such as surface texture, gloss, roughness and color. We identified standard samples of magnesite concentrate, tailings and intermediate products, and established a sorting model based on the genetic characteristics of these standard samples using intelligent sorting technology. Step 4: Determine the upper limit of particle size Amm for selection in the intelligent pre-selection stage based on the particle size of the standard sample. This upper limit of particle size is the minimum particle size at which most monomers of magnesite minerals are liberated. Step 5: Correct the model parameters through multiple sorting tests until the sorting model can support the production of products that meet specific performance requirements. Step 6: The magnesite samples with a particle size of 8-A mm are subjected to intelligent sorting in batch production to obtain magnesite concentrate, tailings and intermediate products. The grade of each valuable component is analyzed and the recovery rate is calculated. Step 7: Combine the intermediate product from Step 6 with the -8mm magnesite ore from Step 1, and carry out separation operations such as reverse flotation for desilication, forward flotation for calcium reduction, and magnetic separation for iron reduction to obtain magnesite concentrate and tailings. Then, perform grade analysis on each valuable component and calculate the recovery rate. Step 8: Combine the magnesite concentrate and tailings from Steps 6 and 7 to obtain the final sorting product.

2. The method of magnesite co-ordinated calcium reduction and upgrading according to claim 1, characterized in that, Step three is as follows: 1) The "manual selection + microscopic inspection" method is adopted. Standard samples are selected and classified according to the differences in appearance parameters such as surface texture, gloss, roughness and color. The categories include magnesite concentrate, tailings and intermediate products. 2) Obtain the intrinsic quality of each standard sample through chemical analysis and process mineralogical analysis, and correlate the standard samples with the intrinsic quality; 3) Collect 360-degree full-shape stereo images of each standard sample, label the categories, and construct a dataset of all standard samples; 4) Train the constructed deep learning model based on the dataset to obtain a sorting model for intelligent classification of samples.

3. The method of magnesite co-ordinated calcium reduction and upgrading according to claim 1, characterized in that, Step five is as follows: 1) For each sorting test, the test samples are intelligently classified using the sorting model obtained in step 3 to obtain classified test samples, and chemical and process mineralogical analyses are performed on the test samples. 2) When the analysis and testing results meet the specific index requirements, the sorting model at this time shall be used as the final sorting model; otherwise, repeat step 3 to update the sorting model until the analysis and testing results meet the specific index requirements.

4. The method of synergistic calcium reduction and upgrading of magnesite according to claim 1, characterized by, In step four, the value of A is typically set to 20–35 mm.

5. The method of synergistic calcium reduction and upgrading of magnesite according to claim 1 or 3, c h a r a c t e r i z e d b y, The specific requirements for the product are as follows: magnesite concentrate with a rougher appearance, pure white color, and a grid-like texture is required, with an MgO grade ≥ 46%, and SiO2, CaO, Fe2O3, and Al2O3 grades all ≤ 0.66%; tailings with a smoother appearance, grayish-white color, and no obvious texture are required, with an MgO grade ≤ 40.4%, SiO2 grade ≥ 1%, CaO grade ≥ 4%, and Fe2O3 and Al2O3 grades all ≥ 0.45%; all other particles except the above two can be used as intermediate product models.

6. The method for synergistic calcium reduction and quality improvement of magnesite according to claim 1 or 3, characterized in that, The reagents used for reverse flotation desilication are mainly composed of hydrocarbons, alcohols, and amines, specifically selected from diesel oil, kerosene, methyl isobutyl methanol, dodecylamine, and cocoyl amine, etc.; the reagents used for direct flotation calcium reduction are mainly composed of fatty acids, petroleum sulfonates, and oxidized paraffin soaps, specifically selected from oleic acid and sodium petroleum sulfonate, etc.; the magnetic separation adopts a high-gradient pulse strong magnetic separator with a field strength of 1.5T and above.

Images