Iron ore concentrate grinding medium intelligent regulation and control method based on real-time data processing

Abstract

The invention relates to the technical field of industrial data processing and intelligent control, and discloses an iron ore concentrate grinding medium intelligent regulation and control method and system based on real-time data processing, and the system comprises a sensing layer, an edge calculation layer, a cloud intelligent layer and an execution layer; according to the method, multi-dimensional data are obtained in real time and processed through a professional data preprocessing process, complex process characteristics are condensed into three core indexes including a particle size distribution index, a chemical component index and an energy consumption process index, and fuzzy worker experience is converted into an accurate numerical index; a reinforcement learning algorithm is introduced to enable the system to autonomously learn an optimal control strategy, so that the system has advanced intelligence of prediction, learning and global optimization; millisecond-level real-time response of index grading and medium regulation and control is achieved through the edge computing unit, timeliness of control is ensured, depth computing and model training are carried out based on cloud capacity, and automation and closed loop of the whole process from data collection to instruction execution are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial data processing and intelligent control, and more particularly to an iron concentrate grinding medium intelligent regulation method and system based on real-time data processing. BACKGROUND

[0002] Iron ore resources in China generally have the characteristics of "poor, fine and impurity", and the target mineral in the raw ore is often densely intergrown with gangue minerals. Through grinding, the target mineral can be fully dissociated to effectively separate the iron ore from impurities and avoid harmful elements entering the molten steel during smelting to affect performance.

[0003] In the prior art, the medium regulation of the iron concentrate grinding process mainly relies on periodic shutdown inspection and manual experience decision. The operator will stop the mill according to a fixed time period, estimate the wear degree by measuring the size reduction of the grinding medium, and take samples for offline laboratory analysis to obtain the particle size distribution and chemical composition data of the grinding product. Then, based on these periodic detection results and historical experience formulas, the addition ratio and size selection of the medium are adjusted periodically, while the main operating parameters of the mill such as speed and filling rate are usually set according to the initial design value of the equipment and remain relatively stable.

[0004] The prior art also has the following defects:

[0005] Data collection relies on manual, periodic shutdown inspection and offline testing, resulting in long information acquisition period, single data dimension and serious lag, which cannot reflect the continuous dynamic changes of the production process.

[0006] The regulation and decision-making are highly dependent on the personal experience of the operator and static experience formulas, and lack systematic quantitative evaluation indicators, resulting in low regulation accuracy, poor consistency and difficulty in standardization.

[0007] The regulation strategy is fixed and rigid, and cannot be self-adaptively adjusted according to fluctuations in ore properties, and lacks forward-looking prediction of the regulation results, resulting in slow process response and limited optimization capability.

[0008] The detection, analysis, decision-making and execution are disconnected, and it is an open-loop, discrete manual operation process, resulting in long delay from problem discovery to implementation of regulation, and inability to achieve precise and timely closed-loop optimization.

[0009] Therefore, a method of real-time sensing, standardized judgment, self-adaptive optimization and closed-loop control is needed to solve the above problems. SUMMARY

[0010] In order to overcome the above-mentioned defects of the prior art, the present application provides an iron concentrate grinding medium intelligent regulation method based on real-time data processing to solve the problems in the background art.

[0011] To achieve the above object, the present application provides the following technical solutions: an iron concentrate grinding medium intelligent regulation and control method based on real-time data processing, comprising:

[0012] S1, collecting the particle size, composition and energy consumption related data of the iron concentrate grinding product in real time through a data acquisition module, and preprocessing the data;

[0013] S2, calculating the particle size distribution related index, chemical composition related index and energy consumption process related index of the preprocessed data through an index calculation module;

[0014] S3, generating the particle size distribution index, chemical composition index and energy consumption process index through an index generation module respectively;

[0015] S4, grading the particle size distribution index, chemical composition index and energy consumption process index respectively through an index grading module according to the generated index;

[0016] S5, regulating and controlling the composite wear-resistant medium in the iron concentrate grinding through a medium regulation and control module according to the index grading result;

[0017] S6, optimizing the mill operation parameters based on machine learning algorithm and digital twin technology through an intelligent optimization module.

[0018] An iron concentrate grinding medium intelligent regulation and control system based on real-time data processing, comprising:

[0019] The perception layer includes a laser particle size analyzer, an online element analyzer, an intelligent electric meter, a vibration sensor and a temperature sensor deployed in the mill site, for collecting the particle size, chemical composition, energy consumption and equipment operation state data of the grinding product in real time;

[0020] The edge computing layer includes an edge computing gateway deployed in the workshop site, for preprocessing the original data collected by the perception layer, including cleaning, filtering, normalization and alignment, and running the medium regulation and control module to make millisecond level real-time decision based on the index grading result;

[0021] The cloud intelligent layer includes a cloud server cluster, for running the index calculation module, index generation module, index grading module and intelligent optimization module, receiving the preprocessed data uploaded by the edge computing layer, performing index calculation, index generation, grading evaluation, and using reinforcement learning, digital twin and genetic algorithm for large-scale simulation and deep optimization, and issuing the optimization strategy to the edge computing layer for execution;

[0022] The execution layer includes a mill speed controller, a feeder and an automatic ball supplementing device, for receiving the control instructions of the edge computing layer or the cloud intelligent layer, and accurately adjusting the composite wear-resistant medium ratio and mill operation parameters.

[0023] The method realizes closed-loop optimization of improving qualified rate of particle size and reducing medium loss by collecting particle size, chemical composition and energy consumption data of iron concentrate grinding product in real time, generating three indexes through calculation and grading evaluation, and finally dynamically regulating the ratio of wear-resistant medium and mill data.

[0024] The technical effects and advantages of the present application are as follows:

[0025] 1. The present application obtains multi-dimensional data such as particle size, chemical composition, energy consumption and equipment state in real time through the data acquisition module, and processes them through professional data preprocessing process, converts traditional dispersed and lagging offline test data into continuous, clean and comparable real-time data stream, and lays a solid data foundation for accurate regulation.

[0026] 2. The present application condenses complex process characteristics into three core indexes of particle size distribution index, chemical composition index and energy consumption process index, each index is generated by weighted fusion algorithm of bottom key indicators, converts fuzzy worker experience into accurate numerical indicators, and realizes scientific and quantitative evaluation of production state.

[0027] 3. The present application introduces reinforcement learning algorithm to enable the system to learn optimal control strategy autonomously, uses digital twin technology to simulate and perform before parameter adjustment, and uses genetic algorithm to perform multi-objective optimization of medium gradation, so that the system has advanced intelligence of prediction, learning and global optimization.

[0028] 4. The present application realizes millisecond-level real-time response of index grading and medium regulation through edge computing unit, ensures the timeliness of control, relies on cloud capability for deep calculation and model training, ensures full-process automation and closed-loop of data acquisition to instruction execution, and truly realizes intelligent autonomy of production process. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 The structural diagram of the present application.

[0030] Figure 2 The flowchart of the present application. DETAILED DESCRIPTION

[0031] The technical solutions in the present application will be described clearly and completely in combination with the drawings in the present application, and the forms of each structure described in the following embodiments are only examples, and the automatic unloading device of the rotary furnace with self-cooling function involved in the present application is not limited to each structure described in the following embodiments, all other embodiments obtained by those skilled in the art without making creative efforts belong to the scope of protection of the present application.

[0032] Refer toFigure 1 The application provides an intelligent control method for iron concentrate grinding medium based on real-time data processing, which comprises a perception layer, an edge computing layer, a cloud intelligent layer and an execution layer.

[0033] With reference to Figure 2 The specific implementation steps of the application comprise the following steps:

[0034] S1, collecting the particle size, composition and energy consumption related data of the iron concentrate grinding product in real time through a data acquisition module, and pre-processing the data.

[0035] It should be specifically pointed out that the particle size, composition and energy consumption related data of the product specifically include:

[0036] medium density and median particle size, medium size and fine powder ratio, sphericity and particle size concentration, total iron content, impurity content, and ferrous content; unit energy consumption, filling rate and grinding efficiency.

[0037] It should be specifically pointed out that the data pre-processing specifically includes:

[0038] Data cleaning and outlier processing: a dynamic threshold method based on the Laplace criterion, i.e. 3σ criterion, is used to identify and process outliers. For any parameter sequence, the mean and standard deviation are calculated, and the data points whose absolute value difference from the mean value exceeds 3σ are judged as gross errors and are removed. After removal, the linear interpolation method is used to fill the missing position with the adjacent effective data points to ensure the continuity of the data sequence.

[0039] Data smoothing filtering: in order to suppress high-frequency noise and improve data stability, a first-order lag filter method, i.e. exponential weighted moving average, is used to smooth the cleaned data. The filtered data value is the weighted calculation of the current sampling value and the previous filtering output value. The filtering coefficient of the current sampling value is in the range of 0.1 to -0.3, which can effectively smooth the fluctuations while retaining the true trend of the data.

[0040] Data normalization: due to the huge difference in the dimension and numerical range of each feature parameter, normalization processing is needed to eliminate the dimension effect, make it comparable and facilitate fusion calculation. The Min-Max normalization method is used to linearly map the original data to the [0, 1] interval.

[0041] Data alignment and synchronization: since the data output frequencies of the particle size analyzer, the online chemical composition analyzer and the electric energy meter are different, data alignment is needed. The system takes the data source with the highest frequency as the reference. For data with lower frequency, the latest value is kept unchanged within its effective period until new updated data is received. In this way, a complete multi-parameter data vector is constructed under the unified timestamp, ensuring the synchronization and accuracy of subsequent fusion calculation.

[0042] S2, the pre-processed data is subjected to a granularity distribution related index, a chemical composition related index and an energy consumption process related index calculation through an index calculation module.

[0043] The granularity distribution related index includes a median particle size coefficient, a fine powder proportion coefficient, a sphericity coefficient and a concentration coefficient.

[0044] The median particle size coefficient is calculated as follows:

[0045]

[0046] Where M is the median particle size coefficient, quantifying the center position of the particle group, reflecting the overall fineness.

[0047] d 50 is the particle size of 50% of the particles in the particle group, representing the typical size of the particle group, d 50 The smaller the d r is an industrial reference value, realizing data dimensionless, realizing cross-batch comparability.

[0048] The index ω1 is negative, with a value range of -0.5 to -0.6, making it inversely proportional to the particle size distribution. The inverse proportional design reflects that the coarser the particle size, the worse the comprehensive performance, and the coarse particles will reduce the particle contact points and weaken the metallurgical reaction efficiency.

[0049] The fine powder proportion coefficient is calculated as follows:

[0050]

[0051] Where F is the fine powder proportion coefficient, strengthening the contribution of the target fine particle proportion, directly affecting the specific surface area and reaction efficiency.

[0052] F f is the target fine particle proportion, i.e. the proportion of particles smaller than a certain particle size, divided by 100 to convert to decimal form; ω2>1, with a value range of 1.2-1.3, significantly increasing the fine powder weight, strengthening the dominant role of fine powder, and high fine powder proportion improving reaction activity, such as the reducibility of pellet.

[0053] The sphericity coefficient is calculated as follows:

[0054]

[0055] Where Q is the sphericity coefficient, coupling the particle morphology characteristics, and high sphericity improving flowability and bulk density.

[0056] S is the degree to which the particle approaches an ideal sphere, ranging from 0 to 1, with S=1 indicating a perfect sphere; high sphericity improves flowability and reduces energy consumption; Smax Maximum sphericity, dimensionless; ω3 > 0, the value range is between 0.8-0.9, which reflects the indirect optimization of high sphericity on the dispersibility and bulk density of particles.

[0057] The concentration coefficient calculation is specifically:

[0058]

[0059] Where C is the concentration coefficient, which amplifies the impact of particle size distribution concentration. Narrow distribution, that is, n is large, reduces over-grinding and improves process stability.

[0060] n is the particle size distribution concentration data, which represents the particle size distribution range. The larger n is, the narrower the particle size distribution, which determines the particle size uniformity; n r is an industrial experience value, which realizes data dimensionless; ω4 > 1, the value range is between 1.5-1.6, which emphasizes the value of narrow distribution in reducing segregation and improving mixing uniformity, and amplifies the impact of narrow distribution on process stability.

[0061] The chemical composition-related indicators include total iron coefficient, impurity coefficient and reducibility coefficient.

[0062] The total iron coefficient calculation is specifically:

[0063]

[0064] Where T is the total iron coefficient, which quantifies the core value of the ore. The dimensionless effect is eliminated by normalization, and the convex function characteristic makes the premium of high-iron ore more significant. The total iron coefficient directly affects the smelting efficiency and pig iron yield, which accounts for more than 60% of the cost of molten iron, and is the core basis for resource pricing.

[0065] TFe is the total iron content, which reflects the total amount of iron elements in the iron concentrate; TFe max is the theoretical maximum iron content, which corresponds to the pure limit in the iron concentrate; α is the weight index, α > 1, the value range is between 1.2-1.3, which strengthens the contribution of high-iron content.

[0066] The impurity coefficient calculation is specifically:

[0067]

[0068] Where Z is the impurity coefficient, and the total amount of impurities is mapped to the [0, 1] interval through linear normalization. The concave function characteristic makes the punishment in the medium and low impurity area more gentle. Reducing impurities can reduce the amount of blast furnace slag, significantly reduce energy consumption and carbon emissions.

[0069] ∑impurities is the total of main impurities, which includes SiO2 / Al2O3 that increases slag, and elements P and S that cause cold brittleness and hot brittleness.

[0070] K is the impurity tolerance coefficient, which determines the penalty threshold. When ∑impurity≥K, this term is zero.

[0071] β is the penalty index, β<1, and the value range is between 0.7-0.8, which alleviates the excessive punishment in the low impurity area.

[0072] The reduction coefficient is calculated as follows:

[0073]

[0074] Where Y is the reduction coefficient, the exponential function amplifies the benefits of high ferrous ratio, and high ferrous ratio shortens the reduction time, which can reduce the coke ratio and fuel consumption, and help low-carbon smelting.

[0075] Fe 2+ TFe is the ferrous ratio, with a value range of 0-1, representing the proportion of easily reduced iron.

[0076] γ is the gain coefficient, γ>0, and the value range is between 0.5-0.6, which controls the incentive intensity.

[0077] The energy consumption process-related indicators include energy consumption coefficient, grinding efficiency and medium filling rate.

[0078] The energy consumption coefficient is calculated as follows:

[0079]

[0080] Where E is the energy consumption coefficient, which directly quantifies the gap between actual energy consumption and ideal target, reflecting the energy conversion efficiency of the equipment.

[0081] E j is the baseline energy consumption, which uses the industry advanced value or the theoretical lower limit value of the equipment, representing the optimal energy consumption target under technically feasible conditions.

[0082] E s is the actual energy consumption, which is the comprehensive power consumption per unit mass of material processed in the statistical period, covering the main motor, auxiliary machine and transmission loss.

[0083] υ1 is the exponential weight, which strengthens the marginal increasing characteristics of energy saving benefits. The closer the actual energy consumption is to the baseline value, the greater the energy saving will be to the energy consumption process index, and the value range is between 0.7-0.9.

[0084] The grinding efficiency is calculated as follows:

[0085]

[0086] Where X is the grinding efficiency, which represents the ability of converting unit electric energy into effective crushing power, and reveals the gap between the actual crushing process and the theoretical limit.

[0087] η is the actual grinding efficiency, defined as the amount of qualified minerals processed per unit of electrical energy. To ensure data accuracy, invalid energy consumption such as downtime and idling needs to be excluded.

[0088] η o is the optimal efficiency threshold, taking 80%-90% of the theoretical maximum efficiency of the equipment. If the actual grinding efficiency exceeds this value, it is fixed at 1 to avoid over-optimization.

[0089] υ2 is the exponential weight, highlighting the dominant influence of efficiency on coefficient economy, with a value range of 1.5-1.8.

[0090] The medium filling rate is calculated as follows:

[0091]

[0092] where J is the medium filling rate, evaluating the optimization degree of grinding medium load on crushing dynamics and energy consumption.

[0093] φ is the actual medium filling rate, the volume of wear-resistant medium in the mill cylinder accounting for the percentage of effective volume.

[0094] φ o is the optimal filling rate, dynamically set according to the type of medium, determined by the medium density and crushing mechanical properties.

[0095] υ3 is the exponential weight, using the absolute value deviation form, equally punishing high impact energy consumption and insufficient grinding, reducing the impact of non-critical items, with a value range of 0.5-0.6.

[0096] S3, the calculated relevant indicators are generated by the index generation module to generate particle size distribution index, chemical composition index and energy consumption process index.

[0097] It needs to be specifically pointed out that the particle size distribution index is specifically:

[0098] L = M * F * Q * C;

[0099] where L is the particle size distribution index, coupling the four key dimensions of particle size distribution, particle size concentration, fine powder content, particle morphology and distribution width, providing a single quantitative index to replace fragmented data, evaluating the comprehensive performance of powder particle size distribution, which is of great significance in industrial production quality control, process optimization and product performance prediction.

[0100] M is the median particle size coefficient, suppressing the negative impact of coarse particles. The larger M means that the particles are overall coarsened, reducing the flowability and reactivity of the powder, and reducing the particle size distribution index.

[0101] F is a fine powder proportionality coefficient, reflecting the contribution of maximizing reaction activity, strengthening the proportion of target fine particles, directly affecting the specific surface area and reaction efficiency, and increasing the proportion of fine powder, the activity is improved, and the particle size distribution index is larger.

[0102] Q is a sphericity coefficient, coupled with particle morphology characteristics, balancing flowability and reaction efficiency, high sphericity improves flowability and bulk density, and the larger the sphericity, the larger the particle size distribution index.

[0103] C is a concentration coefficient, which amplifies the influence of particle size distribution concentration, ensures particle size stability and energy economy, reduces over-grinding with narrow distribution, and improves process stability. The larger the concentration, the narrower the distribution, and the larger the particle size distribution index.

[0104] It should be specifically pointed out that the chemical composition index is specifically:

[0105] H=T*Z*Y;

[0106] Where H is the chemical composition index, which is the core index for evaluating the chemical quality of iron concentrate. It integrates the dispersed chemical composition indexes of main content, impurities and element forms into a single value, enabling horizontal comparison of quality between different batches or sources.

[0107] T is the total iron coefficient, which is the main grade item, highlighting the core position of total iron content, reflecting the theoretical iron potential of the ore, and serving as the main index for measuring the economic value of iron ore, which dominates the overall level of the chemical composition index.

[0108] Z is the impurity coefficient, which is the impurity penalty item, suppressing the negative impact of harmful impurities SiO2, Al2O3, P and S on purity. These impurities increase the smelting slag volume and cause steel brittleness.

[0109] Y is the reducibility coefficient, which is the reducibility incentive item, rewarding easy-to-reduce iron forms. High ferrous iron can reduce smelting energy consumption and improve reaction efficiency.

[0110] The main grade item dominates the basic value, the impurity penalty item restricts the smelting cost, and the reducibility incentive item improves the reaction efficiency. The three are weighted and coupled to realize purity quantification.

[0111] It should be specifically pointed out that the energy consumption process index is specifically:

[0112] N=E*X*J;

[0113] Where N is the energy consumption process index, which quantitatively evaluates the comprehensive energy efficiency of the grinding system, realizes precise diagnosis of the optimization direction of multi-dimensional energy consumption process, and provides data support for equipment parameter adjustment, medium selection and energy saving reconstruction.

[0114] E is the energy consumption coefficient, which is a direct reflection of the energy consumption intensity. When the energy consumption coefficient is less than 1, the energy consumption process index decreases significantly. In the low load working condition, the weight is increased to focus on energy saving priority.

[0115] X is the grinding efficiency, which is a key indicator of energy conversion efficiency. In the high load working condition, the weight is increased to ensure the priority of processing capacity. When the efficiency is low, it becomes the main restriction of the energy consumption process index.

[0116] J is the medium filling rate, which is a regulator of running stability. Optimizing the filling rate can reduce invalid energy consumption, indirectly improve the scores of energy consumption coefficient and grinding efficiency coefficient, and thus improve the energy consumption process index.

[0117] S4, grading the particle size distribution index, the chemical composition index and the energy consumption process index respectively through the index grading module according to the generated index.

[0118] It should be specifically pointed out that the grading of the particle size distribution index is as follows:

[0119] When L≥x1, it is excellent, and the value range of x1 is between 3.5-4;

[0120] When x2≤L<x1, it is good, and the value range of x2 is between 2.8-3;

[0121] When x3≤L<x2, it is qualified, and the value range of x3 is between 2.0-2.1;

[0122] When L<x3, it is poor.

[0123] It should be specifically pointed out that the grading of the chemical composition index is as follows:

[0124] When H≥y1, it is special, and the value range of y1 is between 0.85-0.9;

[0125] When y2≤H<y1, it is excellent, and the value range of y2 is between 0.75-0.8;

[0126] When y3≤H<y2, it is industrial grade, and the value range of y3 is between 0.6-0.65;

[0127] When H<y3, it is rough refining grade.

[0128] It should be specifically pointed out that the grading of the energy consumption process index is as follows:

[0129] When E≥z1, it is excellent, and the value range of z1 is between 0.85-0.9;

[0130] When z2≤E<z1, it is optimized, and the value range of z2 is between 0.6-0.7;

[0131] E < z2, the transformation level.

[0132] S5, according to the index classification results, the composite wear-resistant medium in the iron concentrate grinding is regulated by the medium regulation module.

[0133] Need to be specific, the execution of the regulation of the composite wear-resistant medium in the iron concentrate grinding depends on a real-time decision unit based on edge computing, which is deployed in the mill site, receives the results from the index classification module, and generates regulation instructions in milliseconds by combining the preset expert rule base, including:

[0134] According to the particle size distribution index classification results, the composite wear-resistant medium in the iron concentrate grinding is regulated:

[0135] When the particle size distribution index classification is excellent, the medium material selects high-chromium cast ball, the wear resistance is 2 times that of ordinary low-chromium ball, and the crushing rate is low; the grading strategy is single particle size fine matching, the proportion of small balls with a diameter of 3-5mm is ≥80%, the excessive crushing is reduced, the filling rate is controlled between 32%-35%, the low filling is maintained to maintain high kinetic energy, and the micro-impact grinding mode is mainly adopted to protect the sphericity of fine particles.

[0136] When the particle size distribution index classification is good, the medium material selects medium-chromium alloy ball, which is matched with low-proportion zirconia medium; the grading strategy is three-level matching optimization, selecting balls with diameters of 8mm, 12mm and 15mm, with a ratio of 3:4:3 balanced impact and grinding; the filling rate is controlled between 36%-38%, the contact area of grinding is improved, and the impact and grinding are adopted in a way, and the dissociation of coarse particles is strengthened.

[0137] When the particle size distribution index classification is qualified, the medium material selects low-carbon alloy steel ball, giving priority to toughness, and the outer layer selects tungsten carbide plating to improve wear resistance; the grading strategy is double-peak distribution, selecting large balls with a diameter of 20mm and small balls with a diameter of 6mm, with a ratio of 6:4, and the large balls crush coarse particles and the small balls fill gaps; the filling rate is controlled between 40%-42%, the shear force is enhanced by high filling rate, and the way dominated by high-intensity impact is adopted to quickly reduce d 50 .

[0138] When the particle size distribution index classification is poor, the medium material selects tungsten alloy cylindrical medium to improve unit impact energy; the grading strategy is extreme coarse particle matching, selecting 70% of large balls with a diameter of 25mm-30mm; the filling rate is controlled between 28%-30%, the impact space is guaranteed by low filling, and impact crushing is given priority to, and the coarse crushing cycle is shortened.

[0139] According to the chemical composition index classification results, the composite wear-resistant medium in the iron concentrate grinding is regulated:

[0140] When the chemical composition index is classified as special, the composite wear-resistant medium selects zirconia ceramic balls with a particle size of ≤1 mm, the filling rate is controlled between 25% and 30%, and the rotating speed is reduced to reduce mechanical iron pollution.

[0141] When the chemical composition index is classified as excellent, the composite wear-resistant medium selects high chromium cast iron balls, optimizes the ball size ratio, the proportion of large balls with a diameter of 30 mm is 40%, and the proportion of micro balls with a diameter of 5 mm is 60%, to compensate for the loss of hardness.

[0142] When the chemical composition index is classified as industrial grade, the composite wear-resistant medium selects forged steel balls, the ball filling rate is increased to more than 35%, and the ball replacement period is shortened to 30% to cope with high impurity wear.

[0143] When the chemical composition index is classified as rough refining grade, the composite wear-resistant medium selects white cast iron balls and steel forging composite medium, and the pre-milling process, i.e. high-pressure roller milling, is enabled, and the medium hardness is designed with a gradient.

[0144] According to the classification results of energy consumption process index, the composite wear-resistant medium in iron concentrate grinding is regulated:

[0145] When the energy consumption process index is classified as excellent, the medium options for the main wear-resistant layer are high chromium cast iron and nano alumina ceramic composite structure, and the interface strengthening is realized through gradient metallurgy combination technology; the reinforcing phase uses carbon fiber surface modification to improve the interface bonding force, and the filling rate is strictly controlled.

[0146] In terms of process, vertical mill continuous grinding is adopted to control the rotating speed and slurry concentration to reduce over-grinding and micro powder.

[0147] When the energy consumption process index is classified as optimized, the gradation is corrected, steel balls of different diameters are used, and the optimal diameter and ratio are calculated using the law, among which ceramic balls are used in the middle ring area to reduce friction energy consumption.

[0148] Rare earth is added to refine grains and improve hardness, and automatic ball replacement is realized when the filling rate deviates from the threshold value.

[0149] When the energy consumption process index is classified as improved, the medium system is restructured, the centrifugal casting bimetal tube is replaced as the base body to improve the resistance to erosion, and nano eutectic powder is added for lubrication improvement.

[0150] The linkage laser particle size instrument automatically adjusts the mill rotating speed and feeding amount.

[0151] Need to explain is that each shift scanning energy consumption process index three indicators, if found energy consumption coefficient is abnormal, automatically trigger the magnetic lining replacement, reduce invalid friction, reduce energy consumption; if found efficiency item recession, automatically activate the ceramic ball ratio up, ensure the grinding particle size compliance rate; if found filling rate fluctuation, automatically start gravity sensor ball mechanism, and monitor the wear.

[0152] S6, through the intelligent optimization module, based on machine learning algorithm and digital twin technology, optimize the mill operation parameters.

[0153] Need to be specific is that the intelligent optimization module specifically includes:

[0154] Based on reinforcement learning parameter adaptive optimization: the mill system is modeled as a Markov decision process MDP, wherein the state is the real-time collected and pre-processed particle size index, chemical index, energy consumption index, equipment vibration and temperature; action is the adjustable parameter adjustment amount including mill speed, ore supply and water supply; reward is the weighted function of the improvement of particle size qualified rate and the reduction of medium loss rate; using deep deterministic policy gradient DDPG or proximal policy optimization PPO algorithm, let the agent through the continuous interaction with the environment, every 30 seconds for a decision cycle, learn the optimal control strategy autonomously, realize the dynamic fine adjustment of parameters.

[0155] Based on digital twin simulation prediction and decision rehearsal: a high-fidelity digital twin model of the mill system is established, which integrates crushing mechanics, fluid dynamics and wear model, and can simulate and predict the grinding effect, energy consumption trend and medium wear condition in the future according to the current operating parameters and ore characteristics hardness and composition. Before actual parameter adjustment, multiple parameter combinations can be preformed in the digital twin, the effects of different strategies can be quickly evaluated, and the optimal solution can be selected to avoid production fluctuations and realize forward-looking control.

[0156] Based on genetic algorithm of grading multi-objective optimization: for the size ratio problem of composite wear-resistant medium, it is constructed as a multi-objective optimization problem, the objective function includes maximizing particle size qualified rate, minimizing energy consumption index and minimizing medium wear rate, and non-dominated sorting genetic algorithm NSGA-II with elitist strategy is used to solve it, which can efficiently search the Pareto optimal solution set in the huge solution space, and provide multiple excellent grading scheme selection for the operator.

[0157] Those skilled in the art can clearly understand the implementation of the embodiments of the present application by the description of the above embodiments, which can be realized by software or software combined with necessary general hardware platform, and of course can also be realized by hardware function; based on such understanding, the technical solutions of the present application can be embodied in the form of a software product in essence or the part of the present application which contributes to the prior art, which is stored in a storage medium and includes a plurality of instructions to make a computer device, such as but not limited to a personal computer, a server, or a network device, etc., execute all or part of the steps of the method described in any embodiment of the present application.

[0158] The above describes the exemplary embodiments of the present application, and it should be understood that the above exemplary embodiments are not restrictive but illustrative, and the protection scope of the present application is not limited thereto; it should be understood that those skilled in the art can modify and change the embodiments of the present application without departing from the spirit and scope of the present application, and these modifications and changes should be within the protection scope of the present application.

Claims

1. A method for intelligent control of iron concentrate grinding media based on real-time data processing, characterized in that, Specifically, the following steps are included: S1. Collect data on particle size, composition and energy consumption of iron concentrate grinding products in real time through the data acquisition module, and preprocess the data. S2. The preprocessed data is used to calculate particle size distribution-related indicators, chemical composition-related indicators, and energy consumption process-related indicators through the indicator calculation module. S3. Generate particle size distribution index, chemical composition index and energy consumption process index respectively through the index generation module using the calculated relevant indicators; S4. Based on the generated indices, the particle size distribution index, chemical composition index, and energy consumption process index are classified separately through the index classification module; S5. Based on the index classification results, the composite wear-resistant media in iron concentrate grinding is controlled through the media control module. S6. Through the intelligent optimization module, based on machine learning algorithms and digital twin technology, the operating parameters of the mill are optimized.

2. The intelligent control method for iron concentrate grinding media based on real-time data processing according to claim 1, characterized in that: The particle size distribution related indicators include median particle size coefficient, fine powder ratio coefficient, sphericity coefficient, and concentration coefficient; The median particle size coefficient is calculated as follows: Where M is the median particle size coefficient, d 50 The particle size representing the 50% of particles in a particle group smaller than this value is d, which characterizes the typical size of the particle group. r This is an industrial reference value, and ω1 is the weighting index. The calculation of the fine powder ratio coefficient is as follows: Where F is the fine powder ratio coefficient, F f The target fine particle percentage, i.e., the percentage of particles smaller than a specific particle size, is divided by 100 to convert to decimal form; ω2 is the weighting index; The sphericity coefficient is calculated as follows: Where Q is the sphericity coefficient, and S is the degree to which the particle approaches an ideal sphere, ranging from 0 to 1, with S = 1 indicating a perfect sphere; S max Maximum sphericity; ω3 is the weighting exponent; The concentration factor is calculated as follows: Where C is the concentration coefficient, and n is the particle size distribution concentration data, characterizing the particle size distribution range. r ω4 is an industrial experience value, and ω4 is a weighting index.

3. The intelligent control method for iron concentrate grinding media based on real-time data processing according to claim 1, characterized in that: The relevant indicators of the chemical composition include the total iron coefficient, the impurity coefficient, and the reducing power coefficient; The calculation of the total iron coefficient is as follows: Where T is the total iron coefficient, quantifying the core value of the ore, and TFe is the total iron content, reflecting the total amount of iron in the iron concentrate; TFe max α represents the theoretical maximum iron content, corresponding to the purity limit in iron concentrate; α is the weighting index. The impurity coefficient is calculated as follows: Where Z is the impurity coefficient, ∑impurities is the sum of major impurities, covering elements that increase slag content SiO2 / Al2O3, cause cold brittleness P and hot brittleness S, K is the impurity tolerance coefficient, and β is the penalty index. The calculation of the reducing factor is as follows: Where Y is the reducing coefficient, Fe 2+ / TFe is the ferriferroroxy ratio, which ranges from 0 to 1, and γ is the gain coefficient.

4. The intelligent control method for iron concentrate grinding media based on real-time data processing according to claim 1, characterized in that: The energy consumption process-related indicators include energy consumption coefficient, grinding efficiency, and media filling rate. The energy consumption coefficient is calculated as follows: Where E is the energy consumption coefficient, E j As a baseline energy consumption, E s υ1 represents the actual energy consumption, and υ1 represents the exponential weight. The grinding efficiency is calculated as follows: Where X is the grinding efficiency, η is the actual grinding efficiency, and η o υ2 is the optimal efficiency threshold, and υ2 is the exponential weight. The specific calculation of the medium filling rate is as follows: Where J is the medium filling rate, φ is the actual medium filling rate, and φ o υ3 represents the optimal fill rate and is the exponential weight.

5. The intelligent control method for iron concentrate grinding media based on real-time data processing according to claim 1, characterized in that: The particle size distribution index, chemical composition index, and energy consumption process index are specifically as follows: L = M * F * Q * C; Where L is the particle size distribution index, M is the median particle size coefficient, F is the fine powder proportion coefficient, Q is the sphericity coefficient, and C is the concentration coefficient; H = T * Z * Y; Where H is the chemical composition index, T is the total iron coefficient, which is the main grade item, Z is the impurity coefficient, which is the impurity penalty item, and Y is the reducing power coefficient, which is the reducing power incentive item. N = E * X * J; Where N is the energy consumption process index, E is the energy consumption coefficient, X is the grinding efficiency, and J is the media filling rate.

6. The intelligent control method for iron concentrate grinding media based on real-time data processing according to claim 1, characterized in that: The specific steps for classifying the particle size distribution index, chemical composition index, and energy consumption process index are as follows: When L≥x1, it is excellent; when x2≤L<x1, it is good; when x3≤L<x2, it is acceptable; when L<x3, it is poor. When H≥y1, it is a special grade; when y2≤H<y1, it is a superior grade; when y3≤H<y2, it is an industrial grade; when H<y3, it is a crude grade. When E≥z1, it is rated as excellent; when z2≤E<z1, it is rated as optimized; when E<z2, it is rated as modified.

7. The intelligent control system for iron concentrate grinding media based on real-time data processing according to claim 1, characterized in that, Specifically, it includes: Sensing layer: This includes laser particle size analyzers, online elemental analyzers, smart meters, vibration sensors, and temperature sensors deployed at the mill site, used to collect data on the particle size, chemical composition, energy consumption, and equipment operating status of the grinding products in real time. Edge computing layer: includes an edge computing gateway deployed on the workshop site, used to preprocess the raw data collected by the perception layer by cleaning, filtering, normalizing and aligning, and to run the media control module to make millisecond-level real-time decisions based on the exponential classification results; The cloud-based intelligent layer includes a cloud server cluster, which is used to run the indicator calculation module, index generation module, index classification module and intelligent optimization module. By receiving preprocessed data uploaded from the edge computing layer, it performs indicator calculation, index generation and classification evaluation, and uses reinforcement learning, digital twin and genetic algorithm to perform large-scale simulation and deep optimization, and distributes the optimization strategy to the edge computing layer for execution. The execution layer includes a mill speed controller, a feeder, and an automatic ball replenishing device. It is used to receive control commands from the edge computing layer or the cloud-based intelligent layer and to precisely adjust the ratio of composite wear-resistant media and the mill operating parameters.

Images