Ore recovery efficiency improving method and system based on ore data

Abstract

The invention discloses an ore recovery efficiency improving method and system based on ore data, and relates to the technical field of data processing. An ore recovery efficiency improving system based on ore data comprises an ore data processing module and an ore recovery optimizing module. According to the method, various parameters in the recovery process can be dynamically adjusted through fusion of the geological ore recovery model and real-time data, so that the ore recovery efficiency is improved to the maximum in each stage of ore recovery; a sensor is used for collecting data in real time, and the response speed and precision of the recovery system are effectively improved based on fine control of multi-channel recovery; through the combination of the ore morphology grading channel and the ore dielectric deflection channel, grading treatment and fine separation of the ore can be effectively carried out, the treatment efficiency of concentrate and tailings is improved, and the recovery effect is further improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for improving ore recovery efficiency based on ore data. Background Technology

[0002] In the process of ore recovery, the recovery efficiency usually depends on fixed ore characteristic data and rough recovery plan, lacking dynamic adjustment and precise process control, which leads to the inability to fully improve recovery efficiency and resource utilization. Most existing ore recovery technologies are based on geological exploration data and historical mining data for preliminary recovery design, but these data are often relatively crude and fail to consider the impact of ore changes at different stages and equipment operating status on recovery efficiency.

[0003] Therefore, there is a need for an intelligent method and system for improving ore recovery efficiency based on ore data and real-time monitoring technology. This system should be able to dynamically adjust various parameters in the recovery process according to the changes in ore at different stages and the real-time operating status of the equipment, thereby improving recovery efficiency and resource utilization and overcoming the shortcomings of traditional technologies that rely on ore characteristic data and have coarse control schemes. Summary of the Invention

[0004] This invention aims to provide a method and system for improving ore recovery efficiency based on ore data. By using ore data, dynamic monitoring and refined control of the recovery process can be achieved, optimizing various parameters in the recovery process and significantly improving ore recovery efficiency and resource utilization.

[0005] A method for improving ore recovery efficiency based on ore data includes the following steps: Obtain geological exploration data, historical mining data, and beneficiation and smelting trial data of the ore to be mined; use a geological ore recovery model to calculate and obtain initial ore recovery data; the initial ore recovery data includes predicted grinding power index, predicted target mineral liberation curve, and predicted flotation recovery rate; At the feed end of ore recovery, sensors are used to collect physical data of the ore flow to be mined in real time to obtain the attribute data of the ore to be mined; based on the initial data of ore recovery, the attribute data of the ore to be mined and the real-time equipment data at the current feed end, the evolution calculation of ore recovery is performed to obtain the recommended step-by-step control parameters for ore recovery. The recommended step-by-step control parameters for ore recovery are input into the feed end of the ore recovery system for multi-channel ore recovery. The multi-channel ore recovery includes ore morphology classification channels and ore dielectric deflection channels. Based on the pre-concentrate stream and pre-tailings stream produced by the multi-channel ore recovery, targeted ore reprocessing is carried out to obtain the final optimized ore recovery product.

[0006] As a preferred embodiment of the present invention, the specific steps for calculation using a geological ore recovery model include: Based on geological exploration data, historical mining data, and beneficiation and smelting test data, an ore body attribute relationship map is constructed. In the ore body attribute relationship map, discrete sampling points of the ore to be mined are used as graph nodes, and the mineral association attributes between discrete sampling points are used as the edges of the graph nodes. An ensemble model is used to construct a geological ore recovery model. The ensemble model is constructed based on causal reasoning and reinforcement learning. In the geological ore recovery model, the ore state is set as the geological exploration data characteristics of the ore to be mined, the ore action is set as the recommended ore recovery parameters, and the ore reward is set as the actual ore recovery efficiency. The geological ore recovery model was used to process the ore body attribute relationship map to obtain the initial data for ore recovery.

[0007] As a preferred embodiment of the present invention, the specific steps for performing ore recovery evolution calculations include: An ore recovery evolution framework is constructed, with the objective function being the improvement of ore recovery efficiency within a preset recovery cycle. The initial ore recovery data and the attribute data of the ore to be mined are fused using an attention mechanism to obtain the ore attribute fluctuation pattern of the ore to be mined. The ore property fluctuation pattern and real-time equipment data are input into the ore recovery evolution framework to continuously optimize the recommended step-by-step control parameters for ore recovery over several future time steps. During the continuous optimization process, the operating boundary of the ore recovery equipment is determined based on the real-time equipment data. Within the operating boundary of the ore recovery equipment, the recommended step-by-step control parameters are adjusted according to the response speed of the recommended ore recovery step-by-step control parameters in different time steps by dividing them into different frequency levels.

[0008] As a preferred embodiment of the present invention, the specific steps for multi-channel ore recovery include: For the ore morphology classification channel, the ore to be quarried is pre-sorted based on the recommended step-by-step control parameters for ore recovery, resulting in several ore morphology classification processing streams; the ore morphology classification processing model is used to monitor all ore morphology classification processing streams to obtain the pre-concentrate stream and the pre-tailings stream. For the ore dielectric deflection channel, the pre-concentrate stream output from the ore morphology classification channel is received, and the dielectric response characteristics of the pre-concentrate stream are evaluated to obtain the dielectric characteristic trajectory of the ore surface and the real-time dielectric response characteristic evaluation results. When a deviation is detected between the dielectric characteristic trajectory of the ore surface and the optimal separation target trajectory, the edge controller is used for correction. Based on the final position of the dielectric characteristic trajectory of the ore surface and the final real-time dielectric response characteristic evaluation results, the input pre-concentrate stream is further segmented to obtain a new pre-concentrate stream, and the remainder is assigned to the pre-tailings stream.

[0009] As a preferred embodiment of the present invention, the specific steps for targeted ore reprocessing based on the pre-concentrate stream and pre-tailings stream produced by multi-channel ore recovery include: Based on the monitoring of the target ore recovery content in the pre-concentrate stream and pre-tailings stream, real-time target ore recovery results and real-time non-target ore recovery results are obtained; The recovery value of real-time non-target ore recovery results is determined to obtain the recovery value assessment result; the pre-concentrate stream and pre-tailings stream are reprocessed according to the recovery value assessment result to obtain the ore recovery reprocessing path; and the optimized ore recovery product is obtained according to the ore recovery reprocessing path. Repeat the specific steps of ore reprocessing until the real-time non-target ore recovery results no longer meet the reprocessing requirements, thus completing the ore recovery process for the target ore.

[0010] As a preferred technical solution of the present invention, the ore recovery evolution framework is constructed based on the MPC framework.

[0011] A system for improving ore recovery efficiency based on ore data includes: The ore data processing module includes a data processing unit; the data processing unit is used to acquire geological exploration data, historical mining data, and beneficiation and smelting trial data of the ore to be mined; it uses a geological ore recovery model to perform calculations to obtain initial ore recovery data; the initial ore recovery data includes predicted grinding power index, predicted target mineral liberation curve, and predicted flotation recovery rate; The ore recovery optimization module includes a step-by-step recovery unit and an optimized recovery unit. The step-by-step recovery unit uses sensors at the ore feed end to collect physical data of the ore stream to be mined in real time, obtaining the attribute data of the ore. Based on the initial ore recovery data, the attribute data of the ore to be mined, and the real-time equipment data at the current feed end, it performs ore recovery evolution calculations to obtain recommended step-by-step ore recovery control parameters. The optimized recovery unit is used to input the recommended step-by-step ore recovery control parameters to the ore feed end for multi-channel ore recovery. The multi-channel ore recovery includes ore morphology classification channels and ore dielectric deflection channels. Based on the pre-concentrate stream and pre-tailings stream produced by the multi-channel ore recovery, targeted ore reprocessing is performed to obtain the final optimized ore recovery product.

[0012] The present invention has the following advantages: 1. This invention, through the fusion of geological ore recovery models and real-time data, can dynamically adjust various parameters in the recovery process, thereby maximizing ore recovery efficiency at each stage of ore recovery; by utilizing sensors to collect data in real time and based on multi-channel recovery-based refined control, the response speed and accuracy of the recovery system are effectively improved; through the combination of ore morphology classification channels and ore dielectric deflection channels, ore classification and fine separation can be effectively carried out, improving the processing efficiency of concentrate and tailings, and further enhancing the recovery effect.

[0013] 2. This invention improves the economic benefits of the recovered products and the final quality of the ore by assessing the recovery value of non-target ores in real time and carrying out targeted ore reprocessing. This method ensures minimal waste and maximizes resource utilization during the ore recovery process; by using the MPC framework for ore recovery evolution calculations, ore recovery efficiency can be flexibly optimized within a preset recovery cycle; and by using real-time equipment data feedback, control parameters can be adjusted in a timely manner to ensure continuous optimization of the ore recovery process. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of a system for improving ore recovery efficiency based on ore data, used in an embodiment of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this invention.

[0016] Example 1: A method for improving ore recovery efficiency based on ore data, comprising the following steps: Obtain geological exploration data, historical mining data, and beneficiation and smelting trial data of the ore to be mined; use a geological ore recovery model to calculate and obtain initial ore recovery data; the initial ore recovery data includes predicted grinding power index, predicted target mineral liberation curve, and predicted flotation recovery rate; Geological exploration data, historical mining data, and beneficiation and smelting test data represent the geological characteristics, mining history, and experimental results of the ore beneficiation process of the mining area, respectively. Geological exploration data provides information on the composition, distribution, and structure of the ore body, serving as an important basis for ore recovery design. Historical mining data reflects the past mining conditions of the mining area, including mining depth, ore grade, and other relevant mining activities, which helps to determine the ore recovery potential. Beneficiation and smelting test data refers to the data on mineral separation and recovery efficiency obtained through a series of experiments during the ore beneficiation process. These experimental data provide a scientific basis for optimizing the ore recovery process.

[0017] Specifically, the predicted grinding power index indicates the energy consumption of ore during the grinding process and reflects the grinding efficiency; the predicted target mineral liberation curve shows the trend of the degree of liberation of the target mineral over time at different grinding particle sizes, and is an important parameter for evaluating the mineral separation effect in the beneficiation process; the predicted flotation recovery rate reflects the amount of minerals that can be effectively recovered from the ore after the flotation process, and directly affects the ore recovery efficiency.

[0018] The specific steps for calculations using a geological ore recovery model include: Based on geological exploration data, historical mining data, and beneficiation and smelting test data, an ore body attribute relationship map is constructed. In the ore body attribute relationship map, discrete sampling points of the ore to be mined are used as graph nodes, and the mineral association attributes between discrete sampling points are used as the edges of the graph nodes. In constructing the ore body attribute relationship map, various geological information obtained from the ore body, including exploration data such as borehole core analysis, geochemical testing, mineral composition identification, geological structural characteristics, and rock mechanical parameters; historical data such as actual mining area, output, energy consumption, and blasting fracturing degree obtained during past mining processes; and beneficiation and smelting test data such as grinding tests, flotation tests, and mineral liberation degree analysis conducted in the laboratory, are deeply integrated and correlated. In the map, each discrete sampling point of the ore to be mined or a representative geological unit will be abstracted as a graph node, and the mineral coexistence, associated occurrence, spatial adjacency relationships, or similarities and correlations in mineral characteristics between these sampling points will be abstracted as edges connecting these graph nodes. An ensemble model is used to construct a geological ore recovery model. The ensemble model is constructed based on causal reasoning and reinforcement learning. In the geological ore recovery model, the ore state is set as the geological exploration data characteristics of the ore to be mined, the ore action is set as the recommended ore recovery parameters, and the ore reward is set as the actual ore recovery efficiency. The geological ore recovery model is designed based on the principles of causal reasoning and reinforcement learning. It aims to deeply explore the intrinsic relationship between geological characteristics and mineral processing performance, and to learn how to make optimal process decisions based on geological conditions. In the model's design, the state of the ore is defined as the detailed geological exploration data characteristics of the area where the ore is located. This includes key attributes affecting its selectivity, such as mineral composition, grade, hardness, and structure. The model's actions are defined as recommended ore recovery parameters, such as the settings of ore recovery equipment, which directly guide subsequent mineral processing. The model's reward mechanism is set as the actual ore recovery efficiency or economic benefit. Through continuous learning and optimization, the model seeks to maximize this reward through a state-action mapping, thereby establishing a causal relationship from geological characteristics to the optimal mineral processing strategy.

[0019] It is important to note that in the initial stages of building a geological ore recovery model, a causal graph of ore recovery needs to be constructed using expert knowledge, historical data analysis, and domain expertise. This graph should clearly represent the causal relationship between specific geological characteristics of the ore (such as mineral type, content, grain size, and ore hardness) and key process parameters in the beneficiation process (such as grinding fineness, flotation reagent formulation, and stirring intensity) and the final beneficiation recovery efficiency, rather than a simple statistical correlation. For example, historical data can be used to conclude that "higher quartz content leads to higher ore hardness, which in turn requires greater grinding work and may affect the subsequent flotation reagent effect." The causal graph will serve as the basic knowledge base of the model, guiding the exploration direction and strategy optimization of reinforcement learning.

[0020] In the aforementioned reinforcement learning model, the state is defined as the comprehensive geological exploration data characteristics of the current ore block to be mined. This includes a series of detailed information such as the various grades of the ore, mineralogical characteristics, physical and mechanical properties, and its geological structural environment, depicting the essential attributes and beneficiation potential of the ore. The action space of the reinforcement learning model corresponds to the ore recovery parameters that the geological model needs to recommend, which are process suggestions that the model can directly operate and adjust. The reward signal is the core driving force for the reinforcement learning algorithm to learn and optimize. It comprehensively considers the actual ore recovery efficiency and beneficiation product grade associated with the geological area in historical mining data, and may also include economic benefit indicators such as energy consumption, water consumption, and subsequent processing costs, to form a quantitative feedback that comprehensively reflects the beneficiation performance and economic value.

[0021] The geological ore recovery model was used to process the ore body attribute relationship map to obtain the initial data for ore recovery.

[0022] Using the established geological ore recovery model, the previously created orebody attribute relationship map is processed. The model performs inference and calculation based on the geological state information of each discrete sampling point of the ore to be mined in the map, combined with the causal association rules obtained through reinforcement learning. For each node in the map and its associated information, the model predicts key beneficiation performance indicators such as the grinding work index, the target mineral liberation curve, and the flotation recovery rate. These predictions, calculated by the model, constitute the initial data for ore recovery. They provide important macro-level guidance and preliminary planning basis for subsequent real-time ore recovery evolution calculations, ensuring that the entire ore recovery process can be optimized from the geological source, thereby significantly improving overall recovery efficiency.

[0023] At the feed end of ore recovery, sensors are used to collect physical data of the ore flow to be mined in real time to obtain the attribute data of the ore to be mined; based on the initial data of ore recovery, the attribute data of the ore to be mined and the real-time equipment data at the current feed end, the evolution calculation of ore recovery is performed to obtain the recommended step-by-step control parameters for ore recovery. The specific steps for performing ore recovery evolution calculations include: An ore recovery evolution framework is constructed, with the objective function being the improvement of ore recovery efficiency within a preset recovery cycle. Initial ore recovery data and the attribute data of the ore to be mined are fused using an attention mechanism to obtain the ore attribute fluctuation pattern. This framework is used to integrate information, predict the future, and optimize decisions. Its fundamental goal is to maximize ore recovery efficiency, while potentially considering factors such as grade, energy consumption, and water consumption to achieve optimal overall benefits. The ore recovery evolution framework is built upon the MPC framework. The ore property fluctuation pattern and real-time equipment data are input into the ore recovery evolution framework to continuously optimize the recommended step-by-step control parameters for ore recovery over several future time steps. During the continuous optimization process, the operating boundary of the ore recovery equipment is determined based on the real-time equipment data. Within the operating boundary of the ore recovery equipment, the recommended step-by-step control parameters are adjusted according to the response speed of the recommended ore recovery step-by-step control parameters in different time steps by dividing them into different frequency levels.

[0024] The ore attribute data includes physical properties such as particle size, mineral composition, moisture content, and density. Based on sensor data, the state of the ore at the feed end can be monitored in real time, providing accurate input data for the subsequent recovery process. Ore recovery evolution calculations are performed based on initial ore recovery data, ore attribute data, and real-time equipment data at the feed end. Initial ore recovery data provides preliminary prediction information for the recovery process, while ore attribute data provides a dynamic adjustment basis for the recovery process based on real-time collected ore characteristics. Real-time equipment data reflects the current operating status of the equipment. The combination of these three ensures optimal control at each stage of the recovery process. Simultaneously, in the MPC framework, initial ore recovery data and ore attribute data are fused using an attention mechanism to obtain the ore attribute fluctuation pattern. Through the attention mechanism, the framework can effectively focus on key features of ore attribute changes at different time points, helping the model better understand the impact of ore characteristics on recovery efficiency.

[0025] Ore property fluctuation patterns and real-time equipment data are input into the ore recovery evolution framework for rolling optimization, calculating recommended step-by-step control parameters for ore recovery over several future time steps. During rolling optimization, the framework uses real-time equipment data as a benchmark to determine the operating boundaries of the ore recovery equipment, ensuring that the equipment operates safely and efficiently. Based on the recovery process requirements and equipment status at different time steps, the framework dynamically adjusts the recommended step-by-step control parameters for ore recovery and further optimizes and adjusts these parameters by dividing them into different frequency levels. Through this rolling optimization process, ore recovery efficiency can be maximized while ensuring stable equipment operation.

[0026] Specifically, the steps for adjusting different frequency levels based on the response speed of the recommended step-by-step control parameters in the ore recovery evolution framework are as follows: First, a preliminary classification and response characteristic analysis of the recommended step-by-step control parameter set for ore recovery is performed. Among the recommended control parameters output by the ore recovery evolution framework, there are various types of parameters, such as mill speed and feed rate affecting grinding efficiency; reagent dosage, aeration rate, and stirring intensity affecting flotation; and electric field strength and fluid velocity affecting separation. For these parameters, their physical response speed and time lag need to be pre-assessed through system identification, historical operating data analysis, and domain expert knowledge. For example, adjustments to electric field strength or reagent dosage may have a significant impact on separation or flotation within seconds, while adjustments to mill speed may take tens of seconds or even minutes to stabilize and show an overall impact on particle size distribution. The purpose of classification is to identify which parameters are fast-response, medium-response, and slow-response, laying the foundation for subsequent frequency level division.

[0027] Based on the analysis of the parameter response characteristics, different frequency levels are defined, and a specific control cycle and adjustment granularity are assigned to each level. Typically, this can be divided into at least three or more levels. For example: a high-frequency adjustment level, primarily for parameters with fast response speeds, sensitive to instantaneous disturbances, and with immediate effects, such as local electric field strength and jet velocity in sorting equipment; its adjustment cycle may be on the order of milliseconds to seconds, with fine adjustment granularity; a medium-frequency adjustment level, for parameters with moderate response speeds, significant impacts on local processes but requiring a certain amount of time to take effect, such as flotation reagent dosage, aeration rate, and stirring intensity; its adjustment cycle may be on the order of several seconds to tens of seconds, with relatively moderate adjustment granularity; and a low-frequency adjustment level, for parameters with slow response speeds, far-reaching impacts on the overall process, and high adjustment costs, such as mill speed, total feed rate, and slurry concentration; its adjustment cycle may be tens of seconds to several minutes or even longer, with larger adjustment granularity and usually more closely aligned with macroscopic optimization objectives. This stratified approach ensures that parameters of different properties are adjusted within the most suitable cycle, avoiding resource waste.

[0028] During actual parameter adjustment, the MPC framework will adjust the recommended step-by-step control parameters for ore recovery according to the defined frequency levels, employing multi-rate or cascaded control strategies. Specifically, parameters at the high-frequency adjustment level will be fine-tuned in real-time at extremely high frequencies by fast-response local controllers (such as PLCs or edge controllers) to quickly respond to instantaneous fluctuations and disturbances, ensuring rapid process stabilization. Parameters at the medium- and low-frequency adjustment levels will be handled by the MPC framework as the upper-level controller. Over a longer rolling optimization cycle, the framework will provide strategic setpoints or trends for these parameters based on future predictions and long-term goals. These setpoints will then be passed to lower-level controllers (such as DCS or field controllers) for execution and local feedback control, ensuring operation under the macro-level guidance of the MPC. By fully leveraging the global optimization and predictive capabilities of the MPC, while utilizing fast controllers for precise and timely management of local details, the framework effectively addresses uncertainties and dynamics across various time scales during ore recovery, maximizing recovery efficiency and system stability.

[0029] The recommended step-by-step control parameters for ore recovery are input into the feed end of the ore recovery system for multi-channel ore recovery. The multi-channel ore recovery system includes ore morphology classification channels and ore dielectric deflection channels. Based on the pre-concentrate stream and pre-tailings stream produced by the multi-channel ore recovery system, targeted ore reprocessing is carried out to obtain the final optimized ore recovery product. Throughout the ore recovery process, recommended step-by-step control parameters for ore recovery are input to the feed end of the ore recovery system to initiate and guide the multi-channel ore recovery process. The ore morphology classification channel is a physical separation method that utilizes the differences in ore particles' shape, size, density, and surface roughness in a fluid or gas flow field, resulting in different stress states and motion trajectories. Its function is to pre-separate gangue particles that clearly lack the target mineral or have extremely low grades, as well as target mineral particles with high enrichment and good liberation, through large-scale, coarse but efficient pre-separation before the ore enters the finer and more expensive beneficiation process. This effectively discards a large amount of tailings, reducing the feed volume and processing load of subsequent processes, thereby significantly reducing the operating costs of high-energy-consuming and high-cost stages such as grinding, flotation, or dielectric separation. It also provides a more uniform and high-quality pre-concentrate feed for the subsequent dielectric deflection channel, thus improving the overall efficiency of the entire beneficiation system.

[0030] Dielectric deflection channels for ores are a physical separation method that utilizes the differences in dielectric properties such as dielectric constant and conductivity of mineral particles in a non-uniform alternating electric field to produce different deflection trajectories. Specifically, it enables more precise fine separation of pre-concentrate streams after initial enrichment through morphology-classifying channels. Since different minerals (even those similar in morphology and density) often possess unique dielectric properties, dielectric deflection channels can further separate target minerals at the microscale, achieving a final improvement in grade and deep removal of impurities. Especially for fine-grained or complexly embedded minerals that are difficult to separate using traditional flotation or gravity separation methods, dielectric deflection technology provides an efficient, environmentally friendly, and precise alternative, significantly improving the grade and recovery rate of the final concentrate, and is a key link in the production of high-value-added mineral products.

[0031] The specific steps for multi-channel ore recovery include: For the ore morphology classification channel, the ore to be quarried is pre-sorted based on recommended ore recovery step-by-step control parameters to obtain several ore morphology classification processing streams. The ore morphology classification processing model is used to monitor all ore morphology classification processing streams to obtain pre-concentrate streams and pre-tailings streams. The ore morphology classification processing model combines image recognition and machine learning technology to monitor and analyze all these separated ore morphology classification processing streams in real time, thereby accurately identifying the pre-concentrate stream rich in target minerals and the pre-tailings stream mainly containing gangue or other non-target minerals, laying the foundation for subsequent more refined sorting. For the ore dielectric deflection channel, the pre-concentrate stream output from the ore morphology classification channel is received. The dielectric response characteristics of the pre-concentrate stream are evaluated to obtain the dielectric characteristic trajectory of the ore surface and the real-time dielectric response characteristic evaluation results. When a deviation is detected between the dielectric characteristic trajectory of the ore surface and the optimal separation target trajectory, an edge controller is used for correction. Based on the final position of the dielectric characteristic trajectory of the ore surface and the final real-time dielectric response characteristic evaluation results, the input pre-concentrate stream is further segmented to obtain a new pre-concentrate stream, and the remainder is assigned to the pre-tailings stream. The dielectric response characteristics of each ore particle in the input pre-fines stream are evaluated. High-precision sensors and electromagnetic fields are used to measure the dielectric constant, dielectric loss, and other properties of the ore in a specific electric field in real time, and to track the dielectric trajectory of the ore surface in the electric field, simultaneously obtaining real-time dielectric response characteristic evaluation results. The deviation between the actual motion trajectory of these ore particles and the preset optimal separation target trajectory based on the optimal separation strategy is continuously monitored. Once a deviation is detected, the edge controller will perform high-frequency adjustments to parameters such as local electric field strength, frequency, or slurry flow rate. Adaptive corrections are used to ensure that particles can move precisely along the target trajectory, thereby correcting deviations caused by fluctuations in ore properties or environmental disturbances. Based on the final position of the ore particles' surface dielectric properties trajectory in the electric field and the final real-time dielectric response characteristic evaluation results, the input pre-concentrate stream is further segmented. Particles with ideal dielectric properties and successfully deflected to the target area are collected as a new, higher-purity pre-concentrate stream, while the remaining particles that fail to deflect effectively or whose dielectric properties do not match are assigned to the pre-tailings stream, thereby achieving deeper and finer separation and enrichment of minerals.

[0032] Specifically, the optimal separation target trajectory refers to the theoretical motion path of the target mineral particles with specific dielectric response characteristics, pre-set under ideal operating conditions, in the ore dielectric deflection channel, which maximizes separation efficiency and recovery rate. It represents how, through the most effective combination of electric field configuration, fluid dynamics conditions, and equipment parameters, the target mineral particles and gangue particles can be deflected along different paths to the greatest extent and ultimately effectively separated. The steps to determine the optimal separation target trajectory are as follows: in a controlled laboratory environment, using representative samples of the ore to be separated, under different electric field strengths, frequencies, fluid velocities, and slurry concentrations, precise measurements are taken... The dielectric response characteristics (such as dielectric constant and conductivity) of the target mineral particles and gangue particles are determined, and their actual trajectories in the electric field are recorded using high-speed imaging and particle tracking technology. The experimental data will serve as the basis for constructing a theoretical model. Through iterative optimization algorithms, such as genetic algorithms or particle swarm optimization, with separation efficiency, concentrate grade, and recovery rate as objective functions, a search is conducted in the parameter space of the model to determine a set of optimal electric field parameters and fluid conditions that maximize the spatial separation of the target mineral particles and gangue particles and achieve the best trajectory convergence. The corresponding target mineral particle motion path under the parameter combination is then determined as the optimal separation target trajectory.

[0033] Specific steps for targeted ore reprocessing based on the pre-concentrate stream and pre-tailings stream produced by multi-channel ore recovery include: Based on the monitoring of the target ore recovery content in the pre-concentrate stream and pre-tailings stream, real-time target ore recovery results and real-time non-target ore recovery results are obtained; for example, X-ray fluorescence spectrometry or laser-induced breakdown spectrometry is used to perform real-time and continuous mineral composition scanning on these two streams; by measuring the precise content of the target mineral (i.e. the main mineral to be recovered) in each stream, the real-time target ore recovery results are obtained.

[0034] The system determines the recovery value of real-time non-target ore recovery results to obtain a recovery value assessment result. Based on the recovery value assessment result, the pre-concentrate stream and pre-tailings stream are reprocessed to obtain the ore recovery and reprocessing path. Based on the ore recovery and reprocessing path, the optimized ore recovery product is obtained. The recovery value determination can be achieved by building an expert system. When receiving real-time non-target ore recovery results, semantic analysis and association reasoning techniques are used to search for potential value information related to the current non-target mineral in the knowledge graph. For example, if a certain amount of quartz is identified in the tailings, the system will retrieve the market price and quality requirements of quartz in glass, ceramics, building materials and other fields, and combine the purity, particle size and other characteristics of the mineral itself to assess its potential value as an industrial raw material.

[0035] Based on the content of the target ore, the recovery value of non-target ores, and their specific mineralogical characteristics (such as grain size and mineral phase composition) in each stream, the optimal reprocessing process route and equipment combination are selected. For example, a pre-tailings stream containing a small amount of target minerals but associated with high-value non-target minerals may be directed to a dedicated associated element enrichment unit; a pre-concentrate stream with acceptable grade but insufficient liberation may require ultrafine grinding combined with specific flotation. The resulting ore recovery and reprocessing route indicates the specific beneficiation process, technology, and corresponding process parameters for each stream. Following this route, the corresponding beneficiation operations are performed, ultimately yielding optimized ore recovery products, which may include not only high-purity target concentrates but also valuable associated concentrates or other usable byproducts.

[0036] The specific steps of ore reprocessing are repeated until the real-time non-target ore recovery results no longer meet the reprocessing requirements, at which point the ore recovery process for the ore to be mined is complete. The entire ore reprocessing process is not completed in one go, but is continuously optimized by repeatedly executing the specific steps of ore reprocessing described above. The purpose of this iterative cycle is to continuously extract all potential valuable components from the current logistics, until the content of valuable components in the real-time non-target ore recovery results is extremely low and does not meet the preset minimum requirements for economic or technical reprocessing. Only then is the current ore recovery process considered to have achieved its maximum economic benefits and resource utilization efficiency, and the entire ore recovery process for the ore to be mined is completed.

[0037] Example 2: A system for improving ore recovery efficiency based on ore data, see [link / reference]. Figure 1 As shown, it includes: The ore data processing module includes a data processing unit; the data processing unit is used to acquire geological exploration data, historical mining data, and beneficiation and smelting trial data of the ore to be mined; it uses a geological ore recovery model to perform calculations to obtain initial ore recovery data; the initial ore recovery data includes predicted grinding power index, predicted target mineral liberation curve, and predicted flotation recovery rate; The ore recovery optimization module includes a step-by-step recovery unit and an optimized recovery unit. The step-by-step recovery unit uses sensors at the ore feed end to collect physical data of the ore stream to be mined in real time, obtaining the attribute data of the ore. Based on the initial ore recovery data, the attribute data of the ore to be mined, and the real-time equipment data at the current feed end, it performs ore recovery evolution calculations to obtain recommended step-by-step ore recovery control parameters. The optimized recovery unit is used to input the recommended step-by-step ore recovery control parameters to the ore feed end for multi-channel ore recovery. The multi-channel ore recovery includes ore morphology classification channels and ore dielectric deflection channels. Based on the pre-concentrate stream and pre-tailings stream produced by the multi-channel ore recovery, targeted ore reprocessing is performed to obtain the final optimized ore recovery product.

[0038] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Parts not described in detail in this specification are prior art known to those skilled in the art.

Claims

1. A method for improving ore recovery efficiency based on ore data, characterized in that, Includes the following steps: Obtain geological exploration data, historical mining data, and beneficiation and smelting trial data of the ore to be mined; use a geological ore recovery model to calculate and obtain initial ore recovery data; the initial ore recovery data includes predicted grinding power index, predicted target mineral liberation curve, and predicted flotation recovery rate; At the feed end of ore recovery, sensors are used to collect physical data of the ore flow to be mined in real time, and obtain the attribute data of the ore to be mined. Based on the initial data of ore recovery, the attribute data of the ore to be mined, and the real-time equipment data at the current feed end, the evolution calculation of ore recovery is performed to obtain the recommended step-by-step control parameters for ore recovery. The recommended step-by-step control parameters for ore recovery are input into the feed end of the ore recovery system for multi-channel ore recovery. The multi-channel ore recovery includes ore morphology classification channels and ore dielectric deflection channels. Based on the pre-concentrate stream and pre-tailings stream produced by the multi-channel ore recovery, targeted ore reprocessing is carried out to obtain the final optimized ore recovery product.

2. The method for improving ore recovery efficiency based on ore data according to claim 1, characterized in that, The specific steps for calculations using a geological ore recovery model include: Based on geological exploration data, historical mining data, and beneficiation and smelting test data, an ore body attribute relationship map is constructed. In the ore body attribute relationship map, discrete sampling points of the ore to be mined are used as graph nodes, and the mineral association attributes between discrete sampling points are used as the edges of the graph nodes. An ensemble model is used to construct a geological ore recovery model. The ensemble model is constructed based on causal reasoning and reinforcement learning. In the geological ore recovery model, the ore state is set as the geological exploration data characteristics of the ore to be mined, the ore action is set as the recommended ore recovery parameters, and the ore reward is set as the actual ore recovery efficiency. The geological ore recovery model was used to process the ore body attribute relationship map to obtain the initial data for ore recovery.

3. The method for improving ore recovery efficiency based on ore data according to claim 2, characterized in that, The specific steps for performing ore recovery evolution calculations include: An ore recovery evolution framework is constructed, with the objective function being the improvement of ore recovery efficiency within a preset recovery cycle. The initial ore recovery data and the attribute data of the ore to be mined are fused using an attention mechanism to obtain the ore attribute fluctuation pattern of the ore to be mined. The ore property fluctuation pattern and real-time equipment data are input into the ore recovery evolution framework to continuously optimize the recommended step-by-step control parameters for ore recovery over several future time steps. During the continuous optimization process, the operating boundary of the ore recovery equipment is determined based on the real-time equipment data. Within the operating boundary of the ore recovery equipment, the recommended step-by-step control parameters are adjusted according to the response speed of the recommended ore recovery step-by-step control parameters in different time steps by dividing them into different frequency levels.

4. The method for improving ore recovery efficiency based on ore data according to claim 3, characterized in that, The specific steps for multi-channel ore recovery include: For the ore morphology classification channel, the ore to be quarried is pre-sorted based on the recommended step-by-step control parameters for ore recovery, resulting in several ore morphology classification processing streams; the ore morphology classification processing model is used to monitor all ore morphology classification processing streams to obtain the pre-concentrate stream and the pre-tailings stream. For the ore dielectric deflection channel, the pre-concentrate stream output from the ore morphology classification channel is received, and the dielectric response characteristics of the pre-concentrate stream are evaluated to obtain the dielectric characteristic trajectory of the ore surface and the real-time dielectric response characteristic evaluation results. When a deviation is detected between the dielectric characteristic trajectory of the ore surface and the optimal separation target trajectory, the edge controller is used for correction. Based on the final position of the dielectric characteristic trajectory of the ore surface and the final real-time dielectric response characteristic evaluation results, the input pre-concentrate stream is further segmented to obtain a new pre-concentrate stream, and the remainder is assigned to the pre-tailings stream.

5. The method for improving ore recovery efficiency based on ore data according to claim 4, characterized in that, Specific steps for targeted ore reprocessing based on the pre-concentrate stream and pre-tailings stream produced by multi-channel ore recovery include: Based on the monitoring of the target ore recovery content in the pre-concentrate stream and pre-tailings stream, real-time target ore recovery results and real-time non-target ore recovery results are obtained; The recovery value of real-time non-target ore recovery results is determined to obtain the recovery value assessment result; the pre-concentrate stream and pre-tailings stream are reprocessed according to the recovery value assessment result to obtain the ore recovery reprocessing path; and the optimized ore recovery product is obtained according to the ore recovery reprocessing path. Repeat the specific steps of ore reprocessing until the real-time non-target ore recovery results no longer meet the reprocessing requirements, thus completing the ore recovery process for the target ore.

6. The method for improving ore recovery efficiency based on ore data according to claim 5, characterized in that, The ore recycling evolution framework is built on the MPC framework.

7. A system for improving ore recovery efficiency based on ore data, characterized in that, The system employs a method for improving ore recovery efficiency based on ore data as described in any one of claims 1-6, comprising: The ore data processing module includes a data processing unit; the data processing unit is used to acquire geological exploration data, historical mining data, and beneficiation and smelting trial data of the ore to be mined; it uses a geological ore recovery model to perform calculations to obtain initial ore recovery data; the initial ore recovery data includes predicted grinding power index, predicted target mineral liberation curve, and predicted flotation recovery rate; The ore recovery optimization module includes a step-by-step recovery unit and an optimized recovery unit. The step-by-step recovery unit uses sensors at the ore feed end to collect physical data of the ore stream to be mined in real time, obtaining the attribute data of the ore. Based on the initial ore recovery data, the attribute data of the ore to be mined, and the real-time equipment data at the current feed end, it performs ore recovery evolution calculations to obtain recommended step-by-step ore recovery control parameters. The optimized recovery unit is used to input the recommended step-by-step ore recovery control parameters to the ore feed end for multi-channel ore recovery. The multi-channel ore recovery includes ore morphology classification channels and ore dielectric deflection channels. Based on the pre-concentrate stream and pre-tailings stream produced by the multi-channel ore recovery, targeted ore reprocessing is performed to obtain the final optimized ore recovery product.

Images