Phosphorite ore blending optimization method based on multi-modal large model and intelligent interaction hardware

Abstract

The invention discloses a phosphorite ore blending optimization method based on a multi-modal large model and intelligent interaction hardware. The phosphorite ore blending optimization method comprises the following specific steps: 1) collecting related parameters; 2) dynamically optimizing and inputting the multi-modal large model; and 3) dynamically optimizing and outputting the multi-modal large model. The invention relates to the field of methods, in particular to a phosphorite ore blending optimization method based on a multi-modal large model and intelligent interaction hardware, and particularly relates to a phosphorite ore blending optimization system and method based on the multi-modal large model. Through fusion of deep learning, conditional constraint planning and intelligent interaction hardware, multi-target dynamic optimization of phosphate ore blending is realized, and the method is suitable for low-cost ore blending scheme generation under multi-index constraints of phosphorus pentoxide (PO), magnesium oxide (MgO), arsenic (As) and calcium oxide (CaO).

Description

Technical Field

[0001] The present invention relates to the field of methods, and particularly to an optimization method for phosphate rock blending based on a multimodal large model and intelligent interaction hardware Background Art

[0002] Currently, the present invention belongs to the field of intelligent technologies for phosphate rock industry, and particularly relates to an optimization system and method for phosphate rock blending based on a multimodal large model. By integrating deep learning, conditional constraint programming, and intelligent interaction hardware, it realizes the multi-objective dynamic optimization of phosphate rock blending, and is applicable to the generation of low-cost blending schemes under the constraints of multiple indexes such as phosphorus pentoxide (P2O5), magnesium oxide (MgO), arsenic (As), and calcium oxide (CaO). Summary of the Invention

[0003] The purpose of the present invention is to provide an optimization method for phosphate rock blending based on a multimodal large model and intelligent interaction hardware in view of the above deficiencies

[0004] The optimization method for phosphate rock blending based on a multimodal large model and intelligent interaction hardware of the present invention specifically comprises the following steps 1) Collect relevant parameters The collected parameters are phosphate rock data, blending quantity, inventory, unit price, and real-time working conditions Among them, the phosphate rock data are phosphorus pentoxide (P2O5), magnesium oxide (MgO), arsenic (As), and calcium oxide (CaO) 2) Dynamically optimize the input of the multimodal large model , The objective function MinimizeC is to minimize the total cost C, where c i is the unit cost of each phosphate rock i, and x i is the ratio of each phosphate rock i

[0005] In the constraint conditions in P i represents the P2O5 content of each phosphate rock i, and x i is the ratio of each phosphate rock. After multiplication and summation, the total sum is greater than or equal to the constraint condition value P min ; indicates that the sum of the ratios of each phosphate rock is 1 3) Dynamically optimize the output of the multimodal large model. Input the standardized constraint conditions, clear objective requirements, and clear instructions into the large model for optimizing the formulation scheme. Search for the optimal solution of the genetic algorithm (GA) through the sequential quadratic programming algorithm (SQP) and perform refined adjustment to output the ratios of each phosphate rock, and output the Pareto Front scheme

[0006] Furthermore, the constraints are: P2O5≥26%; MgO≤2.2%; As≤12ppm; 39%≤CaO≤41%.

[0007] Furthermore, this method is based on combining linear programming algorithm (LP), genetic algorithm (GA), sequential quadratic programming algorithm (SQP) and blind hill climbing to balance global search and local convergence efficiency.

[0008] Furthermore, an X-ray fluorescence spectrometer (XRF) or laser-induced breakdown spectrometer (LIBS) is integrated to obtain phosphate rock grade data (P2O5, MgO, As, CaO, Na2O, K2O, Al2O3 and Fe2O3) in real time, and the ERP system and real-time database are integrated to achieve real-time updates of the grade, inventory and unit price of each mineral source.

[0009] Furthermore, it has built-in multimodal large models, industry knowledge base, constraint specifications, clear target requirements and instructions, solidified and stored solutions, and supports voice command input.

[0010] Furthermore, it also has an adaptive genetic algorithm, whose crossover rate and mutation rate are dynamically adjusted with the number of iterations.

[0011] Furthermore, it includes a memory and a processor; the memory is used to store a computer program; the processor is used to implement the phosphate ore blending optimization method based on a multimodal large model and intelligent interactive hardware as described in any one of claims 1 to 6 when executing the computer program.

[0012] Furthermore, a computer program is stored on the storage medium. When the computer program is executed by the processor, the phosphate ore blending optimization method based on a multimodal large model and intelligent interactive hardware as described in any one of claims 1 to 6 is implemented.

[0013] The advantages of the present invention are: the present invention belongs to the field of intelligent technology of phosphate rock industry, and specifically relates to a phosphate rock blending optimization system and method based on a multimodal large model. By integrating deep learning, conditional constraint planning and intelligent interactive hardware, multi-objective dynamic optimization of phosphate rock blending is realized, and it is suitable for generating low-cost blending solutions under multiple index constraints of phosphorus pentoxide (P2O5), magnesium oxide (MgO), arsenic (As) and calcium oxide (CaO). BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic flow chart of the present invention. DETAILED DESCRIPTION

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Components of the embodiments of the present invention usually described and illustrated in the drawings here can be arranged and designed in various different configurations.

[0016] Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0017] It should be noted that similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0018] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper", "lower", "inner", "outer", etc. indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, or the orientations or positional relationships in which the inventive product is customarily placed during use, it is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. In addition, in the description of the present invention, if terms such as "first", "second", etc. are only used for distinguishing descriptions and cannot be construed as indicating or implying relative importance.

[0019] In the description of the embodiments of the present invention, it should also be noted that unless otherwise clearly specified and limited, if terms such as "set", "connect" are understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, and it can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0020] As Figure 1 shown, the phosphate ore blending optimization method based on the multimodal large model and intelligent interactive hardware is as follows: (1) Collect relevant parameters; The collected parameters are phosphate ore data, blending amount, inventory, unit price, and real-time working conditions; Among them, the phosphate ore data is phosphorus pentoxide (P2O5), magnesium oxide (MgO), arsenic (As), and calcium oxide (CaO). 2) Dynamic optimization input of multimodal large models; , , the objective function MinimizeC is to minimize the total cost C, where c i is the unit cost of each phosphate rock i, x i is the ratio of each phosphate rock i.

[0021] Sum the cost of each phosphate rock; Among the constraints, Medium P i represents the P2O5 content of each phosphate rock i, x i is the ratio of each phosphate rock, multiplied and then added, the total is greater than or equal to the constraint value P min ; It means that the sum of the proportions of each phosphate rock is 1; 3) Dynamic optimization output of a multimodal large model: standardized constraints, clear target requirements (such as specific values or prices after the required ore allocation), and clear instructions are input into the large model to optimize the formulation scheme. The sequential quadratic programming algorithm (SQP) algorithm is used to search and fine-tune the optimal solution of the genetic algorithm (GA) to output the proportions of the various phosphate ores and output the Pareto Front solution.

[0022] Furthermore, the constraints are: P2O5≥26%; MgO≤2.2%; As≤12ppm; 39%≤CaO≤41%.

[0023] Furthermore, this method is based on combining linear programming algorithm (LP), genetic algorithm (GA), sequential quadratic programming algorithm (SQP) and blind hill climbing to balance global search and local convergence efficiency.

[0024] Furthermore, an X-ray fluorescence spectrometer (XRF) or laser-induced breakdown spectrometer (LIBS) is integrated to obtain phosphate rock grade data (P2O5, MgO, As, CaO, Na2O, K2O, Al2O3 and Fe2O3) in real time, and the ERP system and real-time database are integrated to achieve real-time updates of the grade, inventory and unit price of each mineral source.

[0025] Furthermore, it has built-in multimodal large models, industry knowledge base, constraint specifications, clear target requirements and instructions, solidified and stored solutions, and supports voice command input.

[0026] Furthermore, it also has an adaptive genetic algorithm, whose crossover rate and mutation rate are dynamically adjusted with the number of iterations.

[0027] Example 1: 1. By integrating an X-ray fluorescence spectrometer (XRF) or a laser-induced breakdown spectrometer (LIBS), obtain ore grade data (P2O5, MgO, As, CaO, Na2O, K2O, Al2O3, and Fe2O3) in real time and enter the real-time database / inventory for selection and matching.

[0028] 2. The operator selects the ore blending requirements for phosphate rock and sets the constraint values: P2O5 ≥ 26%, MgO ≤ 2.2%, As ≤ 12 ppm, 39% ≤ CaO ≤ 41%.

[0029] 3. The large model outputs 3 sets of Pareto optimal solutions and gives the satisfaction of the constraint conditions for each solution.

[0030] 4. The operator determines the final solution based on the phosphate rock data, ore blending quantity, inventory, cost target, and real-time working conditions.

[0031] Example 2:

[0032] The above table shows the phosphate rock grade, unit price, and inventory in the ore pile. The ore blending constraint conditions and target requirements are: P2O5 ≥ 26%, MgO ≤ 2.2%, As ≤ 12 ppm, 39% ≤ CaO ≤ 41%, and the cost is the lowest.

[0033] The ore blending results of the multi-modal large model are as follows: Solution 1: Extreme low-cost ratio Phosphate rock B: 52.5% Phosphate rock C: 47.5% Unit price: 894.65 yuan / ton Component verification: Phosphorus: 27.9% × 52.5% + 23.9% × 47.5% = 26.0% (qualified) Magnesium: 1.75% × 52.5% + 2.4% × 47.5% ≈ 2.06% (qualified) Arsenic: 14.67 ppm × 52.5% + 8.2 ppm × 47.5% ≈ 11.60 ppm (qualified) Calcium: 41.9% × 52.5% + 39% × 47.5% ≈ 40.52% (qualified) Solution 2: Sub-optimal cost ratio Phosphate rock B: 48% Phosphate rock C: 42% Phosphate rock D: 10% Unit price: 897.76 yuan / ton Component verification: Phosphorus: 27.9%×48% + 23.9%×42% + 25.2%×10% ≈ 26.03% (meeting the standard) Magnesium: 1.75%×48% + 2.4%×42% + 2.2%×10% ≈ 2.09% (meeting the standard) Arsenic: 14.67ppm×48% + 8.2ppm×42% + 5.42ppm×10% ≈ 11.57ppm (meeting the standard) Calcium: 41.9%×48% + 39%×42% + 23.5%×10% ≈ 39.2% (meeting the standard) Option 3: High-stability ratio B phosphate rock: 60% C phosphate rock: 30% D phosphate rock: 10% Unit price: 913 yuan / ton Component verification: Phosphorus: 27.9%×60% + 23.9%×30% + 25.2%×10% ≈ 26.43% (meeting the standard) Magnesium: 1.75%×60% + 2.4%×30% + 2.2%×10% ≈ 1.99% (meeting the standard) Arsenic: 14.67ppm×60% + 8.2ppm×30% + 5.42ppm×10% ≈ 11.80ppm (meeting the standard) Calcium: 41.9%×60% + 39%×30% + 23.5%×10% ≈ 39.19% (meeting the standard) Summary: Option 1 is based on the balance between B ore (high phosphorus and low magnesium) and C ore (low price), and the calcium content is accurately controlled at 40.52%.

[0034] Option 2 introduces a small amount of D ore (low calcium characteristics) to further reduce costs. The calcium content is close to the lower limit (39.2%), which is suitable for working conditions sensitive to calcium.

[0035] Option 3 ensures component stability through the dominant ratio of B ore, and at the same time uses D ore to hedge the high magnesium risk of C ore, which is suitable for long-term stable production.

[0036] All options have passed the constraint condition verification. It is recommended to make a choice in combination with the actual situation during actual application.

[0037] Example 3:

[0038] The above table shows the phosphate ore grade, unit price and inventory in the ore pile. The blending requirement is: 22.0% ≤ P2O5 ≤ 22.5%, 5.0% ≤ MgO ≤ 5.5%, As ≤ 12 ppm, with the lowest cost.

[0039] The blending results of the multi-modal large model are as follows: Plan 1: Lowest cost ratio Ratio: E phosphate ore 37.5% G phosphate ore 62.5% Verification: Phosphorus content: 20% × 0.375 + 23.44% × 0.625 = 22.15% 20% × 0.375 + 23.44% × 0.625 = 22.15% (Meets 22.0% - 22.5%) Magnesium content: 5.8% × 0.375 + 5.32% × 0.625 = 5.5% 5.8% × 0.375 + 5.32% × 0.625 = 5.5% (Precisely matches the upper limit of 5.5%) Arsenic content: 14.41 × 0.375 + 9.46 × 0.625 = 11.32 ppm 14.41 × 0.375 + 9.46 × 0.625 = 11.32 ppm (Lower than 12 ppm) Ore unit price: 400 × 0.375 + 780 × 0.625 = 637.5 yuan Plan 2: Balanced ratio (cost and stability) Ratio: G phosphate ore 58% H phosphate ore 32% E phosphate ore 10% Verification: Phosphorus content: 23.44% × 0.58 + 21.6% × 0.32 + 20% × 0.1 ≈ 22.5% 23.44% × 0.58 + 21.6% × 0.32 + 20% × 0.1 ≈ 22.5% (Precisely matches the upper limit) Magnesium content: 5.32% × 0.58 + 5.5% × 0.32 + 5.8% × 0.1 ≈ 5.43% 5.32% × 0.58 + 5.5% × 0.32 + 5.8% × 0.1 ≈ 5.43% (Meets the range) Arsenic content: 9.46×0.58 + 11.51×0.32 + 14.41×0.1 ≈ 10.61 ppm (Compliance) Unit price of ore: 780×0.58 + 650×0.32 + 400×0.1 = 700.4 yuan Option 3: Precise ratio of high - phosphorus ores Ratio: G phosphate rock 48.9% H phosphate rock 51.1% Verification: Phosphorus content: 23.44%×0.489 + 21.6%×0.511 ≈ 22.5% (Precisely match the upper limit) Magnesium content: 5.32%×0.489 + 5.5%×0.511 ≈ 5.41% (Within the compliance range) Arsenic content: 9.46×0.489 + 11.51×0.511 ≈ 10.51 ppm (Compliance) Unit price of ore: 780×0.489 + 650×0.511 ≈ 713.6 yuan Conclusion 1. Option 1 achieves the lowest cost through the combination of low - price E ore and high - phosphorus G ore, but the phosphorus content is close to the lower limit, and the ratio fluctuation needs to be strictly monitored.

[0040] 2. Option 2 introduces H ore to improve stability, precisely controls the phosphorus content at the upper limit, and has better arsenic control, which is suitable for long - term stable production.

[0041] 3. Option 3 reaches the phosphorus upper limit through the precise ratio of G / H ores, is suitable for scenarios with strict requirements for phosphorus content, and has a moderate cost.

[0042] All options have passed the verification of the constraint conditions. It is recommended to make a choice according to the actual situation during practical application.

[0043] .

Claims

1. A phosphate ore blending optimization method based on a multimodal large model and intelligent interaction hardware, characterized in that The specific steps are as follows: 1) Collect relevant parameters; The collected parameters are phosphate rock data, ore blending quantity, inventory, etc. and unit price; Among them, the phosphate rock data includes phosphorus pentoxide (P2O5), magnesium oxide (MgO), arsenic (As) and calcium oxide (CaO). 2) Dynamic optimization input of multimodal large models; ； , MinimizeC is to minimize the total cost C, where c i is the unit cost of each type of phosphate rock i, and x i is the proportion of each type of phosphate rock i; It is the sum of the costs of each phosphate rock. Among the constraints, where P i represents the P2O5 content of each phosphate rock i, and x i is the ratio of each phosphate rock; Indicating that the sum of the proportions of each phosphate rock is 1; 3) Dynamic optimization output of the multimodal large model: input the constraints and target requirements into the large model to optimize the formulation scheme. The optimal solution of the genetic algorithm is searched and fine-tuned through the sequential quadratic programming algorithm to output the ratio of each phosphate rock and output the Pareto frontier solution.

2. The phosphate ore blending optimization method based on a multimodal large model and intelligent interaction hardware according to claim 1, wherein The constraints are: P2O5≥26%; MgO≤2.2%; As≤12ppm; 39%≤CaO≤41%.

3. The phosphate rock blending optimization method based on a multimodal large model and intelligent interactive hardware according to claim 1, wherein This method is based on a combination of linear programming algorithm, genetic algorithm, sequential quadratic programming algorithm and blind hill climbing method to balance global search and local convergence efficiency.

4. The phosphate ore blending optimization method based on a multimodal large model and intelligent interaction hardware according to claim 1, wherein, Integrate X-ray fluorescence spectrometer or laser induced breakdown spectrometer to obtain real-time phosphate rock grade data, integrate ERP system and real-time database to achieve real-time update of each mineral source grade, inventory and unit price.

5. The phosphate rock blending optimization method based on a multimodal large model and intelligent interactive hardware according to claim 1, wherein It has built-in multimodal large models, industry knowledge base, constraint specifications, clear target requirements and instructions, solution solidification and storage, and supports voice command input.

6. The phosphate ore blending optimization method based on the multimodal large model and intelligent interaction hardware according to claim 1, wherein, It also has an adaptive genetic algorithm whose crossover rate and mutation rate are dynamically adjusted with the number of iterations.

7. The phosphate ore blending optimization method based on the multi-modal large model and intelligent interaction hardware according to claim 1, wherein, The collected parameters also include real-time working conditions.

8. A mineral processing device, characterized in that, It comprises a memory and a processor; the memory is used to store a computer program; the processor is used to implement the phosphate ore blending optimization method based on a multimodal large model and intelligent interactive hardware as described in any one of claims 1 to 7 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and when the computer program is executed by the processor, the phosphate ore blending optimization method based on a multimodal large model and intelligent interactive hardware as described in any one of claims 1 to 7 is implemented.

Images