A quantitative screening method for combined collectors based on double matching of mineral crystal chemistry and molecular structure of reagents

Abstract

This invention provides a quantitative screening method for combined collectors based on the dual matching of mineral crystal chemistry and reagent molecular structure, relating to the fields of mineral processing engineering, flotation reagent molecular design, and intelligent screening technology. The method includes: constructing a surface crystal chemistry model of typical cleavage planes of the target ore system and extracting crystal chemistry parameters; quantifying the structural parameters of candidate collector molecules and calculating their adsorption energy and cohesive energy; calculating the spatial matching degree and energy matching degree based on the above parameters; constructing a machine learning prediction model to initially screen candidate combined collectors using the combined collector ratio and dual matching degree as input variables, and the overall separation difference as the objective function; verifying the measured separation data through flotation, using it as external feedback samples, and iteratively correcting the model parameters with the goal of minimizing the prediction error, outputting the optimal formulation after the accuracy reaches the target. This invention breaks through the limitations of traditional empirical trial and error, significantly shortens the research and development cycle, and significantly improves flotation separation indicators.

Description

Technical Field

[0001] This invention relates to the fields of mineral processing engineering, flotation reagent molecular design and intelligent screening technology, and in particular to a quantitative screening method for combined collectors based on the dual matching of mineral crystal chemistry and reagent molecular structure. Background Technology

[0002] Over 90% of my country's primary ilmenite resources are highly concentrated in the Panzhihua-Xichang vanadium-titanium magnetite mining area, which accounts for 97% of the country's total primary ilmenite reserves. This region is a core strategic resource base ensuring the security of my country's titanium industry chain. As mining gradually shifts to deeper levels, the ore type has changed from easily separable gabbro to extremely difficult-to-separate olivine-pyroxene type. In this system, the target mineral ilmenite and the gangue minerals ilmenite and olivine have highly similar physicochemical properties at their surface active sites. Traditional single fatty acid collectors suffer from poor selectivity and insufficient collecting capacity, leading to difficulties in flotation separation and making it difficult to meet industrial production standards for ilmenite concentrate grade and recovery.

[0003] Combined collectors, with their synergistic adsorption effect, possess outstanding advantages such as low reagent consumption, high tolerance to pulp environment, and a balance between strong collecting capacity and high selectivity, making them a core technological direction for solving the flotation problems of difficult-to-process metal oxide ores. However, the current industry development of combined collectors relies entirely on a blind trial-and-error model of "experience-based prediction + massive flotation experiments": commonly used collectors for oxide ores cover multiple categories such as chelates, carboxylic acids, and sulfonic acids, with thousands of binary reagent combinations and tens of thousands of ternary combinations; if actual ores are used to conduct flotation verification one by one, the experimental workload is enormous, the entire reagent development cycle takes 6 to 12 months, the labor, material, and time costs are high, and the screening efficiency is extremely low.

[0004] In recent years, quantitative structure-activity relationship (QSAR) models have begun to be applied in the field of mineral processing reagents. Patent CN117457089A discloses a high-throughput mineral flotation reagent screening method based on density functional theory and graph neural networks, but this method only screens single collector molecules and does not involve combined collector systems; Patent CN109590115B discloses a multi-ligand collector containing soluble metal salts, chelating agents, and fatty acids, but it only discloses a reagent formulation with specific ratios and does not provide a general screening method for combined collectors; Patent CN113369020A discloses a flotation collector that improves the separation efficiency of ilmenite from its gangue, which also directly gives the reagent composition without revealing the screening principle of combined reagents.

[0005] Current QSAR studies generally only consider the structural parameters of the collector molecules themselves, completely ignoring the decisive influence of the crystal chemical properties of the mineral surface on reagent adsorption. This fails to establish a quantitative matching relationship between the mineral and the reagent, and therefore cannot be applied to the efficient screening of combined collectors. In summary, current technologies cannot simultaneously solve the core bottlenecks of long development cycles, poor formulation targeting, and low screening efficiency in flotation reagents for olivine-type ilmenite, making it difficult to meet the urgent needs for green, efficient, and low-cost industrial development of vanadium-titanium strategic resources. Summary of the Invention

[0006] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide a quantitative screening method for combined collectors based on the dual matching of mineral crystal chemistry and reagent molecular structure. This invention solves the problems of blind trial and error, only considering reagent molecules, lack of quantitative matching standards, long screening cycle, and poor sorting indicators in existing technologies.

[0007] To achieve the above objectives, the present invention provides the following solution: A quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure includes: Construct surface crystal chemical models of typical cleavage planes of target minerals and gangue minerals in the target ore system, and extract crystal chemical parameters of each typical cleavage plane; A number of candidate collectors are obtained, the structural parameters of each candidate collector molecule are quantitatively calculated, and the adsorption energy of each candidate collector molecule on each typical cleavage surface and the cohesive energy of each candidate collector molecule are calculated. The spatial matching degree is calculated based on the structural parameters and the crystal chemical parameters. The energy matching degree is calculated based on the adsorption energy and the cohesive energy. Using the ratio parameters of the combined collector consisting of at least two of the candidate collectors, the spatial matching degree, and the energy matching degree as input variables, and the sorting comprehensive difference as the objective function, a machine learning prediction model is constructed. Using global optimization, multiple sets of combined collectors whose predicted values ​​of the objective function satisfy a preset threshold are selected as candidate combined collectors; Flotation verification tests were conducted on each of the candidate collector combinations to measure the recovery rates of the target mineral and gangue minerals, and the results were substituted into the objective function to calculate the measured objective function value for each candidate collector combination. The sample datasets containing the input variables and the measured objective function values ​​of flotation corresponding to each candidate combination collector are used as external feedback samples and input into the machine learning prediction model. With the loss function of minimizing prediction error as the optimization objective, the model parameters of the machine learning prediction model are iteratively corrected using a parameter optimization algorithm. When the prediction accuracy of the corrected machine learning prediction model reaches the preset target accuracy, the optimal combination collector formulation is output based on the prediction results.

[0008] The present invention discloses the following technical effects: This invention provides a quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure. This method significantly improves development efficiency and reduces overall costs: it reduces the complete development cycle of combined collectors from the traditional 6-12 months to 1-2 months, reduces flotation test workload by over 90%, and significantly reduces reagent, ore sample, equipment energy consumption, and manual testing costs. The method also comprehensively optimizes ilmenite flotation separation indicators: the combined collectors screened using this method possess both strong collecting ability and high selectivity. Compared to traditional single fatty acid reagents, the TiO2 grade of ilmenite concentrate increases by 2-3 percentage points, and the titanium metal recovery rate increases by 5-8 percentage points, effectively reducing costs. Low gangue inclusions; strong technical versatility and wide applicability to various mineral types: The complete quantitative screening process, dual-matching mathematical model, and machine learning framework of this invention are not limited to the ilmenite system and can be directly applied to the development of binary / multi-component combined collectors for various difficult-to-separate metal oxide ores such as hematite, cassiterite, and scheelite, with broad prospects for industrialization and promotion; green and low-carbon, in line with the development direction of clean mineral processing: By accurately matching the molecular structure of minerals and reagents, the synergistic adsorption effect of reagents is fully utilized, reducing the total reagent consumption of the combined collector by 20%~30%, and the residual amount of organic reagents in the mineral processing wastewater decreases simultaneously, reducing the subsequent wastewater treatment load, which meets the requirements of green mines and low-carbon mineral processing policies. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Picture 1 The flowchart illustrates a quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure, provided in this embodiment of the invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0013] like Picture 1 As shown, this invention provides a quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure, comprising: Step 100: Construct surface crystal chemical models of typical cleavage planes of the target mineral and gangue mineral in the target ore system, and extract the crystal chemical parameters of each typical cleavage plane; Step 200: Obtain several candidate collectors, quantitatively calculate the structural parameters of each candidate collector molecule, and calculate the adsorption energy of each candidate collector molecule on each typical cleavage surface, as well as the cohesive energy of each candidate collector molecule. Step 300: Calculate the spatial matching degree based on the structural parameters and the crystal chemical parameters; Step 400: Calculate the energy matching degree based on the adsorption energy and the cohesive energy; Step 500: Using the ratio parameters of the combined collector consisting of at least two of the candidate collectors, the spatial matching degree, and the energy matching degree as input variables, and the sorting comprehensive difference as the objective function, construct a machine learning prediction model; Step 600: Use global optimization to select multiple sets of combined collectors whose predicted values ​​of the objective function meet a preset threshold, as candidate combined collectors; Step 700: Conduct flotation verification tests on each of the candidate collector combinations, measure the recovery rates of the target mineral and gangue minerals, and substitute them into the objective function to calculate the measured flotation objective function value of each candidate collector combination. Step 800: Take the sample dataset containing the input variables and the measured objective function values ​​of flotation corresponding to each candidate combination collector as external feedback samples and input it into the machine learning prediction model; take minimizing the loss function of prediction error as the optimization objective and use the parameter optimization algorithm to iteratively correct the model parameters of the machine learning prediction model; when the prediction accuracy of the corrected machine learning prediction model reaches the preset target accuracy, output the optimal combination collector formulation based on the prediction results.

[0014] Furthermore, the target ore system is an olivine-pyroxene type ilmenite system; The target mineral is ilmenite; The gangue minerals include: pyroxene and olivine; The typical cleavage surfaces include: ilmenite cleavage surfaces, pyroxene cleavage surfaces, and olivine cleavage surfaces.

[0015] Furthermore, the crystal chemical parameters include: The area occupied by single active sites on the mineral surface, surface charge density, and surface frontier orbital energy levels.

[0016] Furthermore, the candidate collector includes: Chelating collectors, carboxylic acid collectors, and sulfonic acid collectors.

[0017] Specifically, three typical cleavage planes—ilmenite (104), spodumene (210), and olivine (010)—with the largest exposed area and most critical interaction with the reagents in the olivine-pyroxene type ilmenite system were selected. An atomic-level surface crystal chemical model was constructed using Materials Studio software, with a vacuum layer thickness of 20 Å, a plane wave cutoff energy of 400 eV, and a Brillouin zone k-point grid of 3×3×1. The band structure, charge density distribution, electronic state density, and frontier orbital energy levels of the three mineral cleavage planes were calculated using density functional theory (DFT). Key crystal chemical parameters such as the key bond lengths of ilmenite Ti-O, Fe-O, spodumene Mg-O, and olivine Mg-O, surface charge density, and the energy difference between the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) were extracted.

[0018] Furthermore, the structural parameters include: Collector polar head cross-sectional area, polar group diameter, bonding atom charge, highest occupied molecular orbital energy level, lowest unoccupied molecular orbital energy level, hydrocarbon chain length, and hydrophilic-hydrophobic balance value.

[0019] Specifically, we collected three major categories of collectors commonly used in the flotation of oxidized ores: chelating agents, carboxylic acids, and sulfonic acids. Using Gaussian 16 software at the B3LYP / 6-31G(d,p) basis set level, we performed geometric optimization and frequency calculations on the above collector molecules to obtain a stable lowest energy conformation after confirming the absence of imaginary frequencies. We quantitatively calculated the structural parameters of each collector molecule and established a database of collector molecular structural parameters. The pure mineral flotation recovery data came from publicly published literature and our team's independent experiments to ensure that the data is traceable and reproducible.

[0020] More specifically, the collector described in this invention must meet the following conditions: Chelating collectors: The molecules contain functional groups that can form stable five- or six-membered chelates with titanium and iron ions, including but not limited to hydroxamic acid groups, phosphonic acid groups, and arsenoic acid groups; Carboxylic acid collectors: straight-chain or branched fatty acids with carbon chain lengths of C8-C22 containing carboxyl functional groups in their molecules, and their oxidized, sulfonated, or hydroxylated modified products; Sulfonic acid collectors: alkyl sulfonates, aryl sulfonates, or petroleum sulfonates whose molecules contain sulfonic acid functional groups and have carbon chain lengths of C10-C18.

[0021] The chelating collectors include, but are not limited to, benzohydroxyxamic acid, salicylic acid, octylhydroxyxamic acid, decylhydroxyxamic acid, cycloalkylhydroxyxamic acid, styrenephosphonic acid, 2-ethylhexylphosphonic acid mono-2-ethylhexyl ester, and aminotrimethylenephosphonic acid; The carboxylic acid collectors include, but are not limited to, sodium oleate, sodium stearate, sodium linoleate, oxidized paraffin soap, tall oil fatty acids, sodium castor oil, sodium palmitate, and sodium lauryl oleate. The sulfonic acid collectors include, but are not limited to, sodium dodecyl sulfonate, sodium dodecylbenzene sulfonate, sodium petroleum sulfonate, and sodium naphthalene sulfonate formaldehyde condensate.

[0022] The parameters for quantitatively calculating each collector molecule described in this invention include the following parameters: Active head group parameters: polar group diameter (calculated using van der Waals radius method), polar group cross-sectional area (calculated using Monte Carlo method), bonding atom charge (using Mulliken population analysis), HOMO energy level, LUMO energy level; Hydrophobic hydrocarbon group parameters: hydrocarbon chain length (number of carbon atoms), hydrocarbon group cross-sectional area (calculated using van der Waals volume method), hydrophilic-hydrophobic balance value (HLB) (calculated using Davis method).

[0023] Furthermore, a quantitative structure-activity relationship (QSAR) prediction model for binary combined collectors was constructed using a partial least squares regression (PLS) coupled support vector machine (SVM) algorithm. The model training set included 120 sets of publicly available and reproducible reagent combination flotation data, and the independent test set included 30 sets of blank experimental data. A grid search method was used to perform global optimization in the full parameter space to screen out high-performance candidate combined collector systems with an objective function Y≥60, and the top 5 to 10 candidate formulations were prioritized for output.

[0024] Standardized pure mineral flotation verification tests were conducted on the top 5-10 candidate collector combinations obtained from the screening. The test conditions were uniformly controlled as follows: pulp concentration 10%, pulp pH range 5.5-6.5, total collector dosage 100-110 g / t, and single flotation duration 3 min. Measured flotation recoveries of ilmenite, pyroxene, and olivine were collected and substituted into the comprehensive separation evaluation objective function to calculate the actual recoveries. .

[0025] The sample dataset {S, E, x, Y} corresponding to each candidate drug is used as external feedback sample input to the PLS-SVM coupled QSAR model. A Bayesian optimization algorithm is used to iteratively refine the model, where the objective function of the Bayesian optimization is to minimize the prediction error loss function. In the formula: n is the number of flotation verification sample groups; Y pred,i Y is the model's predicted objective function value for the i-th reagent combination; Y is the measured objective function value for the i-th flotation group; the optimization objective is to minimize the loss function L, reduce the deviation between the model's predicted value and the measured flotation value, and improve the overall prediction accuracy.

[0026] Specifically, the PLS partial least squares regression linear weight coefficients: correct the linear regression weights corresponding to the input features (spatial matching degree S, energy matching degree E, and binary drug molar ratio x). , , The contribution ratio of the three core matching parameters to the sorting index Y was adjusted; at the same time, the intercept constant term b1 of the PLS model was corrected to correct the basic prediction offset.

[0027] SVM Support Vector Machine Kernel Hyperparameters and Classification Weights: A Gaussian kernel (RBF) is used, with Bayesian optimization to simultaneously adjust the kernel width γ and penalty coefficient C; the SVM output layer weight matrix W is also adjusted. sum Adjust the loss weights for high-matching and low-matching drug combination samples, and reduce the prediction error weights for high-selectivity drug samples.

[0028] Dual matching degree feature fusion balance weight: Added fusion weight coefficients α (spatial matching degree weight) and β (energy matching degree weight) to satisfy α+β=1. Bayesian optimization dynamically iterates α and β to adaptively match the crystal features of the target mineral surface and distinguish the influence of two types of factors on flotation selectivity: mineral interface spatial adaptation and interface adsorption force.

[0029] After completing one round of Bayesian parameter iteration, the QSAR model is rerun to predict candidate reagent combinations. If the overall model fitting accuracy is <94%, flotation verification samples are continuously added and iteratively corrected to adjust all the above weight coefficients until the model prediction accuracy is ≥94%. After the model converges, the 2-3 reagent combinations with the highest comprehensive Y value are selected. Molecular dynamics simulations are conducted using COMPASS II force field to analyze the synergistic adsorption configuration and interfacial bonding rules of the optimal combination collectors on the ilmenite surface. The reliability of the screening results is verified at the microscale, and finally, the optimal binary combination collector ratio for industrial application is determined.

[0030] Furthermore, quantitative screening of the benzohydroxyxamic acid-sodium oleate combined collector: Using the pure mineral system of low-magnesium ilmenite of the olivine-pyroxene type from a mine in the Panxi region as an example, the method of this invention was used for rapid screening of combined collectors.

[0031] The specific steps are as follows: Construction and parameter extraction of crystal chemical models of mineral cleavage planes: Crystal chemical models were constructed using cleavage planes of ilmenite (104), spodumene (210), and olivine (010). Modeling was performed using Materials Studio software with a vacuum layer thickness of 20 Å, a cutoff energy of 400 eV, and a k-point grid of 3×3×1. DFT calculations yielded the following results: Ti-O bond length 1.95 Å, Fe-O bond length 2.07 Å on the ilmenite surface; Mg-O bond length 2.11 Å on the spodumene surface; and Mg-O bond length 2.14 Å on the olivine surface. The surface charge densities were +0.12 e / Å, respectively. 2 -0.08e / Å 2 -0.09e / Å 2 The area occupied by a single active site on the ilmenite (104) face was calculated from the two-dimensional lattice parameters of the surface. A s =14.5Å 2 .

[0032] Quantification of the molecular structure parameters of the collector: Benzyl hydroxamic acid and sodium oleate were selected as candidate collectors. Gaussian 16 was used for geometry optimization and frequency calculations under the B3LYP / 6-31G(d,p) basis set, and the absence of imaginary frequencies confirmed a stable conformation.

[0033] The calculated cross-sectional area of ​​the polar head group of benzyl hydroxamic acid is A = 13.6 Å. 2 HLB=9.2; the adsorption energy of benzohydroxyxamic acid on the ilmenite surface was calculated using molecular dynamics simulation (COMPASS II force field, simulation time 2ns). |E ads | =145.2 kJ / mol, intermolecular cohesive energy E coh =89.6 kJ / mol; Cross-sectional area of ​​the polar head group of sodium oleate A = 12.1 Å 2 HLB=15.3. Enter the above parameters into the database.

[0034] Double-matched chemical operator computation and QSAR model prediction: Computational spatial matching degree: ilmenite A s =14.5Å 2 , S = A / A s =13.6 / 14.5=0.94; Energy matching degree E = |E ads | / E coh =145.2 / 89.6=1.62.

[0035] Using S, E, and molar ratio x as input variables, a QSAR model was constructed using the PLS-SVM algorithm. After global optimization via grid search, the model predicted the optimal molar ratio of benzohydroxyxamic acid to sodium oleate as 1:4.5, with a predicted Y=52.7.

[0036] Pure mineral flotation verification: Experimental conditions: Pure mineral mass 2.00g, pulp concentration 10%, stirring speed 1600r / min, pH=6.0 (adjusted with sodium carbonate and sulfuric acid), total collector dosage 100g / t, flotation time 3min. Actual results: Ilmenite recovery rate 91.5%, pyroxene recovery rate 20.8%, olivine recovery rate 17.3%, measured Y=91.5-(20.8+17.3) / 2=53.4.

[0037] Model feedback correction and optimal combination determination: The measured value of Y=53.4 was fed back into the QSAR model, and the weight coefficients were corrected using a Bayesian optimization algorithm. After correction, the model returned Y=53.2 for the same set of data, and the prediction accuracy improved from 89.1% to 95.6%. The optimal combination was finally determined to be benzohydroxyxamic acid:sodium oleate = 1:4.5, which can be directly used for the flotation of ilmenite in this mine.

[0038] Furthermore, quantitative screening of salicylic acid-paraffin soap combination collectors: Using a pure mineral system of ilmenite of the olivine-pyroxene type with high titanium pyroxene content from a mine in the Panxi region as an example, the method of this invention was used for quantitative screening of combined collectors.

[0039] The specific steps are as follows: Construction and parameter extraction of crystal chemical models of mineral cleavage planes: Crystal chemical models were constructed using cleavage planes of ilmenite (104), spodumene (210), and olivine (010). Modeling was performed using Materials Studio software with a vacuum layer thickness of 20 Å, a cutoff energy of 400 eV, and a k-point grid of 3×3×1. DFT calculations yielded the following results: Ti-O bond length 1.96 Å, Fe-O bond length 2.08 Å on the ilmenite surface; Mg-O bond length 2.12 Å on the spodumene surface; and Mg-O bond length 2.15 Å on the olivine surface. The surface charge densities were +0.12 e / Å, respectively. 2 -0.08e / Å 2 -0.09e / Å 2The area occupied by a single active site on the ilmenite (104) face was calculated from the two-dimensional lattice parameters of the surface. A s =14.7Å 2 .

[0040] Quantification of the molecular structure parameters of the collector: Salicylic acid hydroxamic acid and oxidized paraffin soap were selected as candidate collectors. Gaussian 16 was used for geometry optimization and frequency calculations on the B3LYP / 6-31G(d,p) basis set, and the absence of imaginary frequencies confirmed a stable conformation.

[0041] The calculated cross-sectional area of ​​the polar head group of salicylic acid is A = 14.2 Å. 2 HLB=8.7; Molecular dynamics simulation (COMPASS II force field, 2ns) yielded the adsorption energy of salicylic acid on the ilmenite surface. |E ads | =141.3 kJ / mol, intermolecular cohesive energy E coh =87.5 kJ / mol; Cross-sectional area of ​​the polar head group of oxidized paraffin soap A = 12.7 Å 2 HLB=14.6. Enter the above parameters into the database.

[0042] Double-matched chemical operator computation and QSAR model prediction: Spatial matching degree S = A / A s =14.2 / 14.7=0.91; Energy matching degree E = |E ads | / E coh =141.3 / 87.5=1.58.

[0043] Using S, E, and molar ratio x as input variables, a QSAR model was constructed using PLS-SVM. After global optimization via grid search, the optimal molar ratio of salicylic acid hydroxyxamic acid to oxidized paraffin soap was predicted to be 1:3.0, with a predicted Y=49.2.

[0044] Pure mineral flotation verification: Experimental conditions: Pure mineral mass 2.00g, pulp concentration 10%, stirring speed 1600r / min, pH=6.2 (adjusted with sodium carbonate and sulfuric acid), total collector dosage 100g / t, flotation time 3min. Actual results: Ilmenite recovery rate 90.7%, pyroxene recovery rate 22.1%, olivine recovery rate 19.4%, actual value Y=90.7-(22.1+19.4) / 2=50.0.

[0045] Model feedback correction and optimal combination determination: Substituting the measured value Y=50.0 into the model for Bayesian optimization, the model's prediction accuracy improved from 88.3% to 94.8%. The optimal combination was finally determined to be salicylic acid: oxidized paraffin soap = 1:3.2, which can be used for efficient flotation of ilmenite with high titanium pyroxene content.

[0046] Furthermore, quantitative screening of styrenephosphonic acid-tal oil fatty acid combination collectors: Using the 38μm fine-grained olivine-pyroxene type pure mineral system from a mine in the Panxi region as an example, the method of this invention was used to screen phosphonic acid-fatty acid combined collectors.

[0047] The specific steps are as follows: Construction and parameter extraction of crystal chemical models of mineral cleavage planes: Crystal chemical models were constructed using cleavage planes of ilmenite (104), spodumene (210), and olivine (010). Modeling was performed using Materials Studio software with a vacuum layer thickness of 20 Å, a cutoff energy of 400 eV, and a k-point grid of 3×3×1. DFT calculations yielded the following results: Ti-O bond length 1.95 Å, Fe-O bond length 2.07 Å on the ilmenite surface; Mg-O bond length 2.11 Å on the spodumene surface; and Mg-O bond length 2.14 Å on the olivine surface. The surface charge densities were +0.12 e / Å, respectively. 2 -0.08e / Å 2 -0.09e / Å 2 The area occupied by a single active site on the ilmenite (104) face was calculated from the two-dimensional lattice parameters of the surface. A s =14.5Å 2 .

[0048] Quantification of the molecular structure parameters of the collector: Styrenicotinic acid and tall oil fatty acids were selected as candidate collectors. Geometric optimization and frequency calculations were performed using Gaussian 16 on the B3LYP / 6-31G(d,p) basis set, confirming a stable conformation with no imaginary frequencies. The calculated cross-sectional area of ​​the polar head group of styrenicotinic acid was A = 15.1 Å. 2 HLB=7.8; Molecular dynamics simulation (COMPASS II force field, 2ns) yielded the adsorption energy of styrenephosphonic acid on the ilmenite surface. |E ads | =152.6 kJ / mol, intermolecular cohesive energy E coh =92.3 kJ / mol; Tall oil fatty acid polar head group cross-sectional area A = 12.4 Å 2 HLB=13.9. Enter the above parameters into the database.

[0049] Double-matched chemical operator computation and QSAR model prediction: Spatial matching degree S = A / A s =15.1 / 14.5=0.89; Energy matching degree E = |E ads | / E coh =152.6 / 92.3=1.65.

[0050] Using S, E, and molar ratio x as input variables, a QSAR model was constructed using PLS-SVM. After global optimization via grid search, the optimal molar ratio of styrenephosphonic acid: tall oil fatty acid = 1:6.0 was predicted, with a predicted Y = 46.8.

[0051] Pure mineral flotation verification: Experimental conditions: Pure mineral mass 2.00 g, pulp concentration 10%, stirring speed 1700 r / min, pH=5.8 (adjusted with sodium carbonate and sulfuric acid), total collector dosage 110 g / t, flotation time 3 min. Actual results: Ilmenite recovery rate 89.8%, pyroxene recovery rate 22.3%, olivine recovery rate 19.6%, measured Y=89.8-(22.3+19.6) / 2=47.9.

[0052] Model feedback correction and optimal combination determination: The measured value of Y=47.9 was fed back into the QSAR model. After Bayesian optimization, the model's prediction accuracy improved from 87.6% to 95.2%. The optimal combination was finally determined to be styrenephosphonic acid: tall oil fatty acid = 1:6.5, which is suitable for the efficient recovery of fine-grained ilmenite.

[0053] Furthermore, quantitative screening of the octyl hydroxamic acid-sodium linoleate combined collector: Using a pure mineral system of olivine-type ilmenite with high clay content from a mine in the Panxi region as an example, the method of this invention was used to screen clay-resistant combined collectors.

[0054] The specific steps are as follows: Construction and parameter extraction of crystal chemical models of mineral cleavage planes: Crystal chemical models were constructed using cleavage planes of ilmenite (104), spodumene (210), and olivine (010). Modeling was performed using Materials Studio software with a vacuum layer thickness of 20 Å, a cutoff energy of 400 eV, and a k-point grid of 3×3×1. DFT calculations yielded the following results: Ti-O bond length 1.95 Å, Fe-O bond length 2.07 Å on the ilmenite surface; Mg-O bond length 2.11 Å on the spodumene surface; and Mg-O bond length 2.14 Å on the olivine surface. The surface charge densities were +0.12 e / Å, respectively. 2 -0.08e / Å 2 -0.09e / Å 2 The area occupied by a single active site on the ilmenite (104) face was calculated from the two-dimensional lattice parameters of the surface. A s =14.5Å 2 .

[0055] Quantification of the molecular structure parameters of the collector: Octyl hydroxamic acid and sodium linoleate were selected as candidate collectors. Geometric optimization and frequency calculations were performed using Gaussian 16 under the B3LYP / 6-31G(d,p) basis set, confirming a stable conformation with no imaginary frequencies. The calculated cross-sectional area of ​​the polar head group of octyl hydroxamic acid was A = 13.2 Å. 2 HLB=9.5; Molecular dynamics simulation (COMPASS II force field, 2ns) yielded the adsorption energy of octyl hydroxamic acid on the ilmenite surface. |E ads | =139.5 kJ / mol, intermolecular cohesive energy E coh =87.2 kJ / mol; Cross-sectional area of ​​the polar head group of sodium linoleate A = 12.3 Å 2 HLB=14.8. Enter the above parameters into the database.

[0056] Double-matched chemical operator computation and QSAR model prediction: Spatial matching degree S = A / A s =13.2 / 14.5=0.93; Energy matching degree E = |E ads | / E coh =139.5 / 87.2=1.59.

[0057] Using S, E, and molar ratio x as input variables, a QSAR model was constructed using PLS-SVM. After global optimization via grid search, the optimal molar ratio of octylhydroxamic acid:sodium linoleate was predicted to be 1:4.0, with a predicted Y=49.5.

[0058] Pure mineral flotation verification: Experimental conditions: Pure mineral mass 2.00g, pulp concentration 10%, stirring speed 1650r / min, pH=6.3 (adjusted with sodium carbonate and sulfuric acid), total collector dosage 100g / t, flotation time 3min. Actual results: Ilmenite recovery rate 90.1%, pyroxene recovery rate 21.7%, olivine recovery rate 18.5%, calculated Y=90.1-(21.7+18.5) / 2=50.1.

[0059] Model feedback correction and optimal combination determination: The measured value of Y=50.1 was fed back into the QSAR model. After Bayesian optimization, the model's prediction accuracy improved from 88.5% to 95.0%. The optimal combination was finally determined to be octyl hydroxamic acid: sodium linoleate = 1:4.2, which has excellent resistance to mud interference and is suitable for high-mud ilmenite systems.

[0060] This invention also provides a comparative example: screening of combined harvesters using traditional empirical trial-and-error methods: Using the same ilmenite system as an example, a trial-and-error method based on industry experience was employed to screen combined collectors. The specific steps are as follows: 1) Based on field experience, 5 kinds of hydroxamic acid and 8 kinds of fatty acid were selected to conduct a full combination binary ratio test, and a total of 200 sets of tests were designed.

[0061] 2) Relying solely on manual adjustment of the molar ratio, the experiment lasted 45 days, ultimately yielding the optimal combination: benzohydroxyxamic acid: sodium oleate = 1:5.

[0062] 3) Pure mineral flotation results: ilmenite recovery rate 88.7%, pyroxene recovery rate 24.3%, olivine recovery rate 21.6%, objective function Y=42.8.

[0063] Comparative conclusions: The method of this invention reduces the number of test groups by 95%, shortens the development cycle by 85%, increases the objective function Y by 10.6 units, and has significantly better selectivity and recovery rate than the traditional trial-and-error method.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0065] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure, characterized in that, include: Construct surface crystal chemical models of typical cleavage planes of target minerals and gangue minerals in the target ore system, and extract crystal chemical parameters of each typical cleavage plane; A number of candidate collectors are obtained, the structural parameters of each candidate collector molecule are quantitatively calculated, and the adsorption energy of each candidate collector molecule on each typical cleavage surface and the cohesive energy of each candidate collector molecule are calculated. The spatial matching degree is calculated based on the structural parameters and the crystal chemical parameters. The energy matching degree is calculated based on the adsorption energy and the cohesive energy. Using the ratio parameters of the combined collector consisting of at least two of the candidate collectors, the spatial matching degree, and the energy matching degree as input variables, and the sorting comprehensive difference as the objective function, a machine learning prediction model is constructed. Using global optimization, multiple sets of combined collectors whose predicted values ​​of the objective function satisfy a preset threshold are selected as candidate combined collectors; Flotation verification tests were conducted on each of the candidate collector combinations to measure the recovery rates of the target mineral and gangue minerals, and the results were substituted into the objective function to calculate the measured objective function value for each candidate collector combination. The sample datasets containing the input variables and the measured objective function values ​​of flotation corresponding to each candidate combination collector are used as external feedback samples and input into the machine learning prediction model. With the goal of minimizing the loss function of prediction error, the model parameters of the machine learning prediction model are iteratively corrected using a parameter optimization algorithm. When the prediction accuracy of the corrected machine learning prediction model reaches the preset target accuracy, the optimal combination of collector formulations is output based on the prediction results.

2. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, The target ore system is an olivine-pyroxene type ilmenite system; The target mineral is ilmenite; The gangue minerals include: pyroxene and olivine; The typical cleavage surfaces include: ilmenite cleavage surfaces, pyroxene cleavage surfaces, and olivine cleavage surfaces.

3. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, The crystal chemical parameters include: The area occupied by single active sites on the mineral surface, surface charge density, and surface frontier orbital energy levels.

4. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, The candidate collectors include: Chelating collectors, carboxylic acid collectors, and sulfonic acid collectors.

5. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, Structural parameters include: Collector polar head cross-sectional area, polar group diameter, bonding atom charge, highest occupied molecular orbital energy level, lowest unoccupied molecular orbital energy level, hydrocarbon chain length, and hydrophilic-hydrophobic balance value.

6. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, The formula for calculating the spatial matching degree is: ,in, A The cross-sectional area of ​​the polar head group of the collector. A s It refers to the area occupied by a single active site on the mineral surface.

7. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, The formula for calculating energy matching degree is: E ads E represents the adsorption energy of the collector on the mineral surface. coh It represents the cohesive energy of the collector molecule.

8. The quantitative screening method for combined collectors based on dual matching of mineral crystal chemistry and reagent molecular structure according to claim 1, characterized in that, The formula for calculating the loss function that minimizes prediction error is: ； Where n is the number of flotation verification sample groups; Y pred,i Y is the predicted objective function value of the model for the i-th reagent combination; Y is the measured objective function value of the i-th flotation group.

Images