Big Data-Driven Optimization Method and System for Tailings Dam Recovery Scheme

Abstract

This invention discloses a big data-driven optimization method and system for tailings dam recovery schemes, relating to the field of mineral resource recovery and intelligent optimization technology. The system's operation includes: constructing a detailed three-dimensional digital model of the dam area; conducting value zoning evaluation to delineate high-value recovery zones and low-value replacement zones; implementing surface mining and pipeline transportation of high-value zones using a floating sand mining platform; simultaneously filling low-value materials into the goaf to maintain stress balance; and dynamically regulating the hydraulic system based on intelligent monitoring data of the entire dam area to ensure the safety of the dry beach length and phreatic line. The system comprises six modules: multi-source data acquisition, digital modeling, intelligent zoning evaluation, recovery operation execution, synchronous filling and reinforcement, and dynamic monitoring and balance control. This application achieves synergistic optimization of efficient high-value tailings recovery and dam structural safety through spatial replacement and dynamic balance mechanisms.

Description

Technical Field

[0001] This invention relates to the field of mineral resource recovery and intelligent optimization technology, specifically to a method and system for optimizing tailings dam recovery schemes based on big data. Background Technology

[0002] In the process of sustainable development in modern mining, tailings ponds are not only important environmental protection facilities for handling mine waste, but also secondary resource reservoirs with enormous economic value. With the advancement of digital mine construction and the widespread application of big data analytics, using advanced data processing methods for efficient management and resource recycling of tailings ponds has become an important way to improve the overall efficiency of mines and the level of environmental governance.

[0003] Among them, big data-driven tailings dam mining optimization technology is a core means to ensure the orderly development of resources and the structural safety of the dam area. This technology aims to construct a scientific mining sequence and path planning system by comprehensively analyzing the distribution of physical properties, mechanical stability data, and spatial geometry of tailings in the dam. This system achieves a dynamic balance between maximizing the economic value of mining operations and minimizing safety risks while ensuring the overall stability of the tailings dam.

[0004] Existing tailings dam recovery technologies suffer from several drawbacks. First, the recovery logic is too rigid, generally employing a single linear approach from the tail of the dam towards the dam front. This results in excessively long recovery cycles for high-value tailings, and requires substantial investment in construction and resettlement for the initial treatment of low-value tailings. Second, there is a lack of real-time mechanical compensation capabilities for low-stress zones formed after recovery, making it difficult to effectively control the risk of instability or liquefaction of local beach surfaces when mining high-value areas. Third, existing monitoring systems primarily focus on the external morphology of the dam, lacking in-depth integration of hydraulic balance, pore water pressure, and changes in beach elevation within the dam, making precise dynamic control difficult under complex recovery conditions. Finally, maintenance and operation plans rely too heavily on historical experience, making it difficult to achieve synergistic optimization of spatial displacement and material filling based on the heterogeneity of material distribution within the dam. To address the shortcomings of traditional recovery schemes, such as rigid mining logic, lack of mechanical compensation, single monitoring dimensions, and insufficient dynamic balancing capabilities, this invention proposes a data-driven optimization method and system for tailings dam recovery schemes, which is of paramount importance. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for optimizing tailings dam recovery schemes based on big data, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, this invention provides the following technical solution: a big data-driven tailings dam recovery scheme optimization method, comprising the following steps: S1, acquiring the physical and mechanical characteristics and spatial distribution characteristics of tailings within the dam area through a multi-source heterogeneous data acquisition terminal, and constructing a three-dimensional digital fine model based on a big data processing platform; S2, conducting value zoning evaluation based on the three-dimensional digital fine model, delineating high-value recovery zones, low-value replacement zones, and safety monitoring and warning zones; S3, formulating a spatial replacement recovery strategy based on the zoning results, using a floating... The sand mining platform conducts surface mining of high-value mining areas and transfers the extracted high-value tailings to the concentrator via pipelines; S4, it executes a synchronous filling and stress reinforcement procedure for the goaf, pumping low-value materials generated from mining or drained materials from the tailings to the goaf for spatial replacement to eliminate low-stress areas and maintain stress balance within the reservoir; S5, it uses a full-area intelligent monitoring system to obtain real-time data on surface elevation, pore water pressure, and waterline position, and dynamically triggers a hydraulic balance control mechanism to ensure that the dry surface length and phreatic line burial depth are within a safe preset range.

[0007] Preferably, step S1 specifically includes the following steps: S11, acquiring macroscopic topographic data of the tailings dam using high-precision satellite positioning technology and UAV aerial surveying technology; S12, collecting tailings moisture content, density, particle size distribution, and shear strength parameters at different depths through a sensor array deployed inside the dam area; S13, performing multidimensional correlation between the macroscopic topographic data and microscopic physical and mechanical parameters, and generating a continuous tailings property gradient field using a spatial interpolation algorithm; S14, constructing a three-dimensional digital fine model that includes geometric topological relationships and mechanical property distribution.

[0008] Preferably, step S2 specifically includes the following steps: S21, setting a multi-dimensional value assessment index system, which includes tailings metal grade, particle size composition, mining difficulty coefficient and environmental governance cost; S22, calculating the comprehensive value score of each spatial unit using a preset weight allocation logic; S23, dividing the reservoir area into a high-value area to be mined, a low-value area for replacement, and a protected dam-related area based on the comprehensive value score and a preset threshold range.

[0009] Preferably, step S3 specifically includes the following steps: S31, planning the operating path and draft of the floating sand mining platform based on the spatial coordinate information of the high-value area; S32, driving the floating sand mining platform to the predetermined location through remote control commands, and starting the jet sand suction device or cutter suction device for operation; S33, monitoring the material flow status in the conveying pipeline through flow meters and concentration meters during the mining process, and adjusting the pumping pressure in real time.

[0010] Preferably, step S4 specifically includes the following steps: S41, real-time monitoring of the volume and geometric shape changes of the goaf formed by the mining of high-value areas; S42, calculation of the total amount and filling ratio of the required filling material according to the mechanical stability requirements of the goaf; S43, orderly filling the goaf with low-value tailings or dried tailings after dewatering treatment from the mineral processing plant through the replacement pipeline; S44, using the self-weight stress of the filling material or forced compaction to compensate for the in-situ stress field and prevent instability of the beach surface around the goaf.

[0011] Preferably, step S5 specifically includes the following steps: S51, collecting elevation data at different points using laser ranging terminals deployed on the dry beach surface to generate a real-time beach topographic map; S52, acquiring dynamic data of the seepage line at different depths using pore water pressure gauges buried inside the reservoir; S53, comparing the real-time monitoring data with a preset safety benchmark model to determine the safety redundancy under the current mining conditions; S54, when the monitoring data deviates from the preset safety range, automatically adjusting the pumping volume of the return water system or adjusting the interception direction of the temporary diversion weir to control the water level in the reservoir by changing the hydraulic gradient.

[0012] Preferably, the construction process of the three-dimensional digital fine model also includes the deep integration of historical mining data and current geological exploration data. By employing multi-scale feature extraction logic, the discrete point data obtained from borehole sampling is transformed into a continuous tensor field characterizing the heterogeneity of materials throughout the entire reservoir area. The model not only contains static spatial coordinate information but also records the evolution history of the reservoir area at different accumulation stages through time series indexing, thereby providing a complete data foundation for optimizing the mining scheme.

[0013] Preferably, the value zoning evaluation employs a heuristic search algorithm based on fuzzy comprehensive evaluation. When calculating the value score, the system comprehensively considers the market price fluctuation step of tailings, energy loss along the transportation route, and the additional disturbance to surrounding structures caused by the mining operation. Through iterative optimization, the system automatically identifies the mining sequence that generates the maximum net economic benefit while minimizing the impact on dam stability.

[0014] Preferably, the floating sand mining platform is equipped with a Beidou satellite navigation terminal and an autonomous obstacle avoidance module. During operation, the platform automatically corrects its navigation trajectory based on a real-time generated beach topographic map and uses a multi-beam sonar system to detect the geometry of the underwater mining interface in real time, ensuring that the mining depth does not exceed the preset safe mining depth and preventing excessive local slope ratios due to over-mining.

[0015] Preferably, the core logic of the spatial replacement mining strategy lies in the dynamic balance of materials. While mining high-value tailings, the system precisely controls the input rate of the backfill material according to the law of conservation of mass and the principle of volume compensation. The selection of backfill material is based on its mechanical compatibility with the in-situ tailings, prioritizing low-value materials that have undergone consolidation treatment and have a specific gradient to form a support structure with structural strength. This replacement process is not a simple stockpiling, but a controlled backfilling based on stress field reconstruction.

[0016] Preferably, the intelligent monitoring system for the entire reservoir area adopts a combination of distributed fiber optic sensing technology and wireless sensor networks. By deploying highly sensitive monitoring nodes between the dam body and the mining area, it achieves all-weather capture of minute displacement and vibration signals. The monitoring data is processed by an edge computing gateway and then uploaded to the central cloud for trend prediction. The system uses a long short-term memory neural network algorithm to predict the evolution trend of the seepage line, and when it is predicted that the water level may exceed the warning level within a preset time, it issues emergency control commands in advance.

[0017] Preferably, the hydraulic balance control mechanism is achieved by adjusting the effective volume and flood discharge capacity within the reservoir. Under conditions of localized water level fluctuations caused by mining, the system accurately calculates the hydraulic connections between the mining area, the filling area, and the clarifier. By adjusting the height and angle of the temporary diversion weir at the millisecond level, the flow direction can be effectively guided, preventing scouring of the unconsolidated filling area while ensuring that the dry beach length always meets the specific proportion requirements of the flood control standard.

[0018] Preferably, the system also includes a decision support terminal equipped with an interactive visual interface. Operators can view a digital twin of the storage area in real time through this interface, including color heatmaps of material grades, cloud maps of stress distribution, and early warning indicators of safety risks. The system supports simulation of different mining schemes, and by adjusting simulation parameters, compares the economic benefits and safety factors under different spatial replacement paths, thereby selecting the optimal mining execution strategy.

[0019] The big data-driven tailings dam recovery scheme optimization system, used to implement the above-mentioned methods, includes: a multi-source data acquisition module for acquiring topographic, material physical and mechanical parameters, and spatial distribution data of the tailings dam; a digital modeling module connected to the multi-source data acquisition module for constructing a dynamically updated three-dimensional digital fine model; an intelligent zoning assessment module for performing value gradient analysis and safety risk assessment based on the model to determine the recovery zone and replacement zone; a recovery operation execution module, including a remotely controlled floating sand mining platform and pipeline transportation subsystem, for performing the extraction and transportation of high-value tailings; a synchronous filling and reinforcement module configured to pump low-value materials to the vacancies generated by recovery, achieving spatial replacement and mechanical compensation; and a dynamic monitoring and balance control module for monitoring key safety parameters of the dam area and providing feedback to adjust the hydraulic system to maintain the overall dynamic balance of the dam area.

[0020] This invention provides a method and system for optimizing tailings dam recovery schemes based on big data, which has the following beneficial effects: (1) During system operation, this invention breaks through the single and rigid linear mining mode in traditional tailings dam mining schemes. By introducing big data-driven spatial replacement and dynamic balance logic, it achieves a dual improvement in resource recovery efficiency and structural safety performance. First, this invention utilizes a three-dimensional digital fine model constructed by multi-source data fusion, which can accurately identify the heterogeneity of the value distribution of materials in the dam. This allows mining operations to no longer blindly follow the order from the tail of the dam to the front of the dam, but can prioritize the development of high-value areas. This greatly shortens the capital recovery cycle of enterprises, avoids early ineffective investment in the treatment of low-value tailings, and fundamentally solves the hidden dangers of illegal mining driven by economic interests. The spatial replacement strategy proposed in this invention organically combines mining and treatment through synchronous filling and stress reinforcement of the goaf. This approach changes the passive mode of relying solely on dam reinforcement to ensure safety in the past, and instead adopts an active strategy to address the source of risk. By using low-value materials to fill the space left after the removal of high-value areas in a timely manner, the low-stress zone formed by the goaf can be effectively eliminated, maintaining the continuity and stability of the stress field inside the dam. This physical-level replacement of equal volume or mass prevents liquefaction and collapse of the surrounding beach surface from a mechanical perspective, significantly improving the inherent safety level during the mining process.

[0021] (2) The system described in this invention adopts a modular and intelligent design concept, with seamless information flow between modules via high-speed data links. The closed-loop system formed by digital modeling, intelligent zoning, operation execution, and dynamic monitoring allows the mining plan to be optimized and adjusted in real time according to changes in the actual mining conditions. This high degree of flexibility and adaptability enables this invention to be widely applied to tailings dam mining projects of different types and scales, and has strong promotional value and social benefits. By adopting a specific spatial interpolation algorithm and multi-scale feature extraction logic, this invention significantly improves the accuracy of tailings distribution prediction. Compared with traditional exploration methods, this invention can obtain higher precision attribute gradient fields with a smaller data sample size, reducing data acquisition costs, and providing a more reliable scientific basis for mining decisions.

[0022] (3) By using an optimization algorithm with multiple loss terms for zoning evaluation, the optimal balance between maximizing economic benefits and minimizing safety impacts can be automatically found. This multi-objective optimization capability makes the mining scheme not only technically feasible but also economically competitive. Through the combination of distributed optical fiber sensing technology and long short-term memory neural network algorithm, this invention realizes the transformation from passive discovery to proactive early warning of safety risks in the reservoir area. By predicting the evolution trend of the seepage line in advance, managers can take control measures in advance, thereby greatly reducing the probability of sudden disasters and ensuring the continuity of mine production.

[0023] (4) This invention provides a systematic solution for tailings dam resource recovery that ensures both economic benefits and strict adherence to safety regulations through the deep integration of spatial substitution and dynamic equilibrium, achieving a unity of economic, social, and environmental benefits. The application of this invention will strongly promote the advancement of secondary resource utilization technology in mines and provide key technical support for the construction of digital and intelligent mines.

[0024] (5) This invention achieves real-time perception and precise intervention of the reservoir's safety status through a full-area intelligent monitoring and hydraulic balance control mechanism. By shifting the monitoring focus from the dam's external surface to the reservoir's internal dynamics, it can detect potential danger signals such as water level fluctuations and abnormal pore water pressure caused by mining operations earlier. By dynamically adjusting the return water system and diversion device, it ensures that the dry beach length and phreatic line burial depth are always in an ideal state. This dynamic control based on real-time feedback effectively solves the contradiction between mining operations and the reservoir's hydraulic dynamic balance, ensuring the reservoir's flood control capacity and anti-sliding stability under complex mining conditions. This invention utilizes a floating sand mining platform and pipeline transportation system to achieve remotely controlled beach mining, significantly reducing the safety risks for workers in complex reservoir environments. Through high-precision positioning and obstacle avoidance algorithms, it ensures the accuracy of the mining trajectory, avoiding accidental damage to the reservoir bottom seepage prevention layer or dam structure by mechanical equipment. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the overall technical solution architecture of the tailings dam recovery optimization method based on big data proposed in this invention. Figure 2 This is a schematic diagram of the core principle framework of the mining optimization based on spatial substitution and stress compensation in this invention; Figure 3 This is a logical flowchart of the intelligent monitoring system and hydraulic balance control mechanism for the entire reservoir area in this invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1 This invention provides a method and system for optimizing tailings dam recovery schemes based on big data. Please refer to [link / reference]. Figure 1 This embodiment details the specific execution process and underlying engineering logic of a tailings dam recovery optimization method based on big data. The application scenario of this embodiment is an open-pit tailings dam that is in its closure period and contains a large amount of high-grade gold ore tailings due to limitations in early mineral processing technology.

[0028] First, step S1 is executed, where the physical and mechanical characteristics and spatial distribution features of tailings within the tailings reservoir are acquired using a multi-source heterogeneous data acquisition terminal. A detailed three-dimensional digital model is then constructed based on a big data processing platform. In practice, multi-source data acquisition is a multi-dimensional process of capturing physical quantities. In step S11, technicians utilize high-precision satellite positioning technology with centimeter-level positioning accuracy and UAV aerial surveying technology equipped with multispectral cameras to conduct a comprehensive scan of the tailings reservoir. The UAV flies in a cross-grid pattern along preset waypoints to acquire macroscopic topographic data of the tailings reservoir. This data includes not only the latitude and longitude coordinates and altitude of the surface but also, through multispectral images, reflects the oxidation degree and water content characteristics of the tailings surface layer in different areas.

[0029] Step S12 is executed synchronously, collecting deep data through a sensor array deployed within the reservoir area. Sensor groups are installed on key monitoring profiles within the reservoir area using engineering drilling to collect tailings moisture content, density, particle size distribution, and shear strength parameters at different depths. Each sensor node has an independent code, enabling the conversion of collected analog signals into digital signals, which are then transmitted to a data concentrator via wired or wireless communication links. In step S13, the big data processing platform performs multi-dimensional correlation between the acquired macroscopic topographic data and microscopic physical and mechanical parameters. Using a spatial interpolation algorithm, the system performs surface fitting and voxel filling on discrete sampling points in a three-dimensional coordinate system. This interpolation logic is not a simple linear averaging but considers the gravity sorting law during tailings deposition, thereby generating a continuous tailings property gradient field. Finally, in step S14, a three-dimensional digital fine model containing geometric topological relationships and mechanical property distributions is constructed. This model divides the entire reservoir area into millions of tiny computational units, each carrying key attributes such as coordinates, grade, density, cohesion, and internal friction angle.

[0030] Next, step S2 is executed, where a value zoning evaluation is performed based on a three-dimensional digital fine model, delineating high-value mining areas, low-value replacement areas, and safety monitoring and warning areas. In step S21, the system establishes a multi-dimensional value assessment index system. The core parameters of this system include the tailings metal grade, i.e., the number of grams of target minerals contained in each ton of tailings. In addition, a particle size composition parameter is introduced, as particle size directly affects the grinding cost of subsequent mineral processing. The mining difficulty coefficient is quantified based on the unit depth, distance from the dam body, and thickness of the overlying soil layer. Environmental remediation costs include funds reserved for vegetation restoration after mining.

[0031] In step S22, the comprehensive value score of each spatial unit is calculated using a preset weight allocation logic. The system dynamically adjusts the weight of the grade index by comparing it with the current market price step of the target mineral. If the market price is in an upward cycle, the weight of the grade will be increased accordingly. In step S23, based on the comprehensive value score and a preset threshold range, the reservoir area is divided into a high-value area to be mined, a low-value area for replacement, and a protected dam-related area. The high-value area is usually located near the early discharge outlet, where the tailings particles are coarser and the metal grade is higher. The protected dam-related area is located within a certain horizontal distance upstream of the dam. The stability of this area is directly related to the structural safety of the reservoir area, and direct mining is prohibited in principle.

[0032] The core extraction execution phase, step S3, then begins, where a spatial replacement extraction strategy is formulated based on the zoning results. In step S31, the operating path and draft of the floating sand mining platform are planned based on the spatial coordinates of the high-value area. The floating sand mining platform is an integrated electromechanical system combining BeiDou positioning, power propulsion, and a sand mining mechanism. In step S32, operators remotely control the platform to reach the designated location. The platform activates its jet suction device, using a high-speed jet generated by a high-pressure water pump to agitate the underwater tailings, creating a slurry of a certain concentration. The slurry is then mechanically sucked into a delivery pump using a sluice suction device. During the delivery process in step S33, electromagnetic flowmeters and nuclear density meters installed on the pipeline continuously monitor the material flow. The system adjusts the pumping pressure in real time based on the measured flow rate and concentration to prevent pipeline blockage due to excessively low flow rate or damage to joints due to excessively high pressure.

[0033] To ensure mechanical stability during the mining process, step S4, namely the synchronous filling and stress reinforcement procedure for the goaf, is executed. Step S41 involves real-time monitoring of the goaf volume formed by mining in the high-value area. Since mining disrupts the original stress balance, step S42 calculates the required total amount and proportion of filling material based on the mechanical stability requirements of the goaf. This invention employs spatial substitution logic. In step S43, low-value tailings discharged from the concentrator, from which high-value components have been extracted, are systematically filled into the goaf through substitution pipes. To enhance the strength of the filling material, a certain amount of solidifying agent can be added to the tailings in proportion. Step S44 utilizes the self-weight stress of the filling material and subsequent consolidation and compaction methods to effectively compensate for the in-situ stress field. This equal-volume or equal-mass substitution ensures that the effective normal stress around the goaf does not drastically decrease due to mining, thereby preventing slippage or liquefaction of the surrounding beach surface towards the mining center.

[0034] Finally, dynamic balance is maintained through the intelligent monitoring system covering the entire reservoir area in step S5. Step S51 uses laser ranging terminals deployed on the dry beach surface to collect elevation data at different points, generating a real-time beach topographic map. Step S52 uses pore water pressure gauges embedded inside the reservoir body to obtain the dynamic position of the phreatic line. Step S53 compares the real-time monitoring data with a preset safety benchmark model. If it is found that the waterline advances too quickly towards the dam body due to localized mining, or if the phreatic line depth is found to be less than the safety threshold, step S54 automatically triggers the hydraulic balance control mechanism. The system increases the pumping volume by adjusting the frequency converter of the return water pumping station, or adjusts the interception direction of the temporary diversion weir through the actuator, changing the movement path of the water flow on the beach, thereby increasing the effective length of the dry beach, lowering the phreatic line, and ensuring that the safety redundancy of the reservoir area remains within the standard range throughout the entire mining process.

[0035] Example 2 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, this embodiment focuses on illustrating the system architecture and modular collaborative logic of the present invention when dealing with tailings ponds with complex sedimentary structures and high water content. Such tailings ponds often exhibit severe layered heterogeneity, meaning that sedimentary layers from different periods show significant differences in physical strength and permeability.

[0036] The system's multi-source data acquisition module is activated first. In addition to the drones and conventional sensors mentioned in Example 1, this example also introduces distributed fiber optic sensing technology. Sensing fibers are laid laterally within the dam body surrounding the mining impact zone. By sensing the frequency drift of light waves propagating in the fibers, continuous monitoring of micron-level deformation and temperature changes is achieved. This monitoring method can capture localized stress concentration phenomena that traditional point sensors cannot detect.

[0037] After receiving this high-frequency data, the digital modeling module employs multi-scale feature extraction logic. It uses discrete point data from borehole sampling as hard constraints and wave velocity cloud maps obtained from geophysical surveys as soft constraints, constructing a continuous tensor field characterizing the heterogeneity of materials across the entire reservoir area through a deep fusion algorithm. In this tensor field, each spatial coordinate not only corresponds to a density value but is also associated with a response function related to the strain rate. This means the model can not only describe "what it is now" but also predict "what it will become under specific perturbations" through simulation.

[0038] In this embodiment, the intelligent zoning assessment module employs a heuristic search algorithm based on fuzzy comprehensive evaluation. When calculating the value score, it considers not only static ore grade but also introduces a dynamic risk reduction factor. For example, if a high-value block is located in a sensitive area affected by the reservoir's water level, its assessment score will be automatically reduced based on the predicted water inflow and obstacle removal costs. Through thousands of iterations, the system automatically identifies the mining sequence that generates the maximum net economic benefit while minimizing the impact on dam stability. This sequence is not static; it is dynamically reconstructed based on the progress of mining and feedback from real-time monitoring data.

[0039] In this embodiment, the floating sand mining platform in the mining operation execution module is equipped with a multibeam sonar system. During underwater operations, the sonar system detects the geometry of the mining interface in real time using a circular scanning method. This morphological data is transmitted back to the digital modeling module in real time to update the geometric parameters of the goaf in the 3D model. If the sonar detects that the mining slope exceeds the natural angle of repose of the tailings, the operation execution module will automatically limit the drilling depth of the sand mining head and issue a risk warning to the central control terminal.

[0040] In this embodiment, the synchronous filling and reinforcement module implements a more complex material distribution strategy. Considering the large amount of dried tailings at the tail end of the reservoir, the system schedules a belt conveyor to mix the dried tailings with the wet tailings discharged from the concentrator in a mixing chamber. By adjusting the speed of the agitator and the mixing ratio, a high-concentration filling slurry with specific rheological properties is produced. This slurry is injected into the bottom of the goaf through a high-pressure pumping system, using its good fluidity to fill every void, and forming a support structure with a certain self-stabilizing ability after consolidation.

[0041] The dynamic monitoring and balance control module is responsible for the overall safety closed loop. The system uses a long short-term memory neural network algorithm to predict the evolution trend of the infiltration line. By learning the nonlinear relationship between historical rainfall, water level changes, and the depth of the infiltration line, this algorithm can predict the potential risk status of the reservoir area within the next 48 hours. Once it is predicted that the water level will exceed the warning value, the control mechanism will open the backup flood discharge channel in advance or adjust the opening of the return water valve, realizing the transformation from "passive emergency response" to "active prevention".

[0042] Example 3 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, this embodiment further elaborates on the interaction logic of the decision support terminal in this invention and the adaptive optimization process under extreme environmental conditions. This embodiment assumes that continuous heavy rainfall occurred during the tailings dam mining process, posing a severe challenge to the hydraulic balance of the dam area.

[0043] The decision support terminal is equipped with an interactive visual interface, providing managers with a digital twin mirror. On this interface, the material quality of the entire warehouse area is presented in the form of a color heatmap, with red representing high-value areas and blue representing low-value areas. At the same time, stress distribution is updated in real time in the form of a cloud map, and any area with increased local stress will flash an alarm.

[0044] When heavy rainfall begins, the rainfall parameters captured by the multi-source data acquisition module are input into the system in real time. The dynamic monitoring and balance control module immediately recalculates the hydraulic gradient. Since rainfall causes the reservoir water level to rise and the dry beach length to shorten, the system automatically issues instructions to temporarily suspend deep-water mining operations in high-value areas and instead guide the floating sand mining platform to shallow clearing in low-value replacement areas to free up more effective flood control capacity.

[0045] At this point, the spatial displacement logic enters emergency mode. The synchronous filling and reinforcement module increases the filling pressure on the goaf, rapidly injecting materials with fast-setting properties to improve the initial strength of the fill and prevent rainwater infiltration that could cause tailings liquefaction around the goaf. The system accurately calculates the hydraulic connection between the mining area, the filling area, and the clarifier, and uses a flow guide simulator on the terminal interface to guide on-site personnel in deploying temporary water-retaining dams.

[0046] In the hydraulic balance control mechanism, the system makes millisecond-level adjustments to the height and angle of the temporary diversion weir. This adjustment is based on a real-time comparison of the kinetic energy of the water flow and the scour resistance of the beach surface. By guiding rainwater to the pre-set reinforced drainage channel, the system avoids direct scouring of the incompletely consolidated fill area.

[0047] Furthermore, the decision support terminal also features scenario simulation capabilities. Under extreme weather conditions, managers can manually set extreme parameters such as rainfall intensity and drainage pump station failure rates, allowing the system to compare the safety factors under different emergency recovery paths. The system calculates the time step for the infiltration line to reach the critical height at different drainage rates. This data-driven quantitative analysis ensures that management decisions no longer rely on intuition but are based on rigorous mechanical logic and probability distributions.

[0048] As the rainfall subsides, the system automatically and smoothly switches back to normal mining mode based on the slope of the water level drop. It re-plans the floating platform's sand-mining trajectory based on the latest beach topography data, ensuring that any elevation changes caused by localized scouring after the rain are promptly corrected. The entire process demonstrates the system's exceptional resilience and adaptability, minimizing the impact of extreme weather events that could otherwise lead to production stoppages or safety accidents through data-driven, precise control.

[0049] Example 4 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically: In this embodiment, the various logical operations and processing mechanisms involved in the present invention are described in detail in pure text form to support the engineering implementation of the system, and any formula symbols are strictly prohibited.

[0050] In constructing a detailed 3D digital model, the system employs an attribute fusion logic. First, the position vector of the target region in the 3D coordinate system is obtained. Then, a set of sample points surrounding that position is extracted. For each sample point, the system acquires its corresponding physical parameter value. During spatial interpolation, the system determines the weight coefficient of each sample point's contribution to the target location's attributes based on the geometric distance between the sample point and the target location. Points closer to the target have higher weight coefficients. This weight coefficient allocation follows an inverse decay relationship, meaning the contribution decreases non-linearly with increasing distance. Ultimately, the attribute value of the target location is determined by the sum of the products of the attribute values ​​of all relevant sample points and their corresponding weights.

[0051] When conducting value zoning evaluation, the generation of the comprehensive value score involves parameter ratio calculations across multiple dimensions. The system first obtains the raw metal grade value of the target unit and divides it by the benchmark price step of that mineral in the current market, obtaining a standardized component reflecting economic potential. Simultaneously, the system calculates the combined horizontal and vertical displacement from the unit's center to the dam crest edge. This distance is then compared to a preset safety warning radius. If the ratio is less than 1, it indicates that the unit is in a high-risk sensitive area, and the system will forcibly lower the unit's comprehensive value score through a penalty function. Furthermore, the mining difficulty coefficient is calculated by obtaining the overburden thickness value and the maximum excavation depth of the sand mining equipment, performing a difference calculation and mapping to the remaining proportion. Only when the thickness value is within the effective operating range of the equipment will the unit be assigned a higher mineable score.

[0052] In the hydraulic balance control mechanism, the system executes a dynamic control logic based on volume conservation. First, the volume of tailings removed per unit time during the mining operation is calculated in real time, and this volume is used as an input parameter for the goaf expansion rate. Simultaneously, the system acquires the slurry flow rate input through the filling pipeline within the same unit time and, combined with the percentage of solids in the slurry, calculates the actual volume of solids replaced. The system continuously compares the difference between the removed volume and the filled volume. If the difference exceeds a preset reservoir capacity fluctuation threshold, it indicates a challenge to the stress balance within the reservoir. In this case, the control mechanism increases the filling rate by increasing the operating frequency of the pumping motor.

[0053] For predicting the depth of the seepage line, the system employs a recursive time-series analysis method. It acquires water level observation data streams over a specific time period extending forward from the current moment and analyzes the lag effect between the water level rise rate and the seepage line response time. The system describes this lag effect as a damping coefficient reflecting the reservoir's permeability. When predicting future conditions, the system uses the predicted rainfall inflow as a driving factor, combined with the damping coefficient, to calculate the vertical coordinate value of the seepage line at future time points. When the difference between this coordinate value and the actual coordinate value of the beach surface—i.e., the depth—is less than the preset minimum safety protection distance, the system generates a "true" warning signal, triggering an emergency pumping and drainage procedure.

[0054] In the path planning of a floating sand mining platform, the system implements an obstacle avoidance logic and an optimal trajectory selection mechanism. The system acquires the platform's current three-dimensional coordinates and the coordinates of the target sand mining unit, calculating the azimuth angle between them. Simultaneously, it obtains an underwater obstacle distribution map along the operating path using sonar scanning. Among multiple candidate path schemes, the system calculates the length, cumulative turning angle, and expected energy consumption for each path. Through a multi-objective weighted optimization logic, the system selects the scheme that minimizes the comprehensive weighted result of the above values ​​as the execution command and sends it to the platform's propulsion system.

[0055] Example 5 For a big data-driven optimization method and system for tailings dam recovery schemes, please refer to [link / reference]. Figure 2 Specifically, in this embodiment, the hardware components and their physical connections of the entire tailings dam recovery scheme optimization system based on big data are described in detail to reflect the system characteristics of mechatronics.

[0056] The multi-source data acquisition module consists of fixed monitoring stations and mobile monitoring units deployed in different locations within the reservoir area. The fixed monitoring stations include total stations and laser scanners mounted on reinforced concrete bases, which are connected to the central control room via industrial Ethernet. The mobile monitoring units primarily utilize sensors mounted on drones and floating platforms, achieving real-time data transmission via wireless bridges or 5G communication modules.

[0057] The digital modeling module consists of a high-performance computing cluster, with a central processing unit (CPU) at its core capable of massively parallel processing. This module receives raw hexadecimal data streams from the multi-source data acquisition module via a high-speed bus interface and stores them in a distributed storage array with multi-level redundancy. The modeling logic runs in the computing cluster's memory, performing millisecond-level fusion processing of static geographic coordinate data and dynamic sensor response data.

[0058] The intelligent zoning assessment module is a decision-making logic unit running on the central server. It has an independent parameter configuration interface, allowing administrators to adjust the weight parameters in the assessment logic based on market conditions and safety regulations. This module maintains real-time communication with the digital modeling module; whenever the reservoir model is updated, it automatically reruns the zoning calculation process to ensure the real-time nature of the zoning results.

[0059] The core of the longwall mining operation module is the floating sand mining platform. This platform employs a catamaran structure to enhance stability, and the deck houses sand mining pump units, jet pump units, and a power frequency converter control cabinet. The vertical displacement of the sand mining head is controlled by a high-precision hydraulic lifting system, with its lifting stroke digitally fed back via an encoder. The pipeline transportation subsystem consists of multi-stage pump stations and high-strength wear-resistant composite pipes, with each pump station equipped with independent pressure monitoring and pressure relief protection devices.

[0060] The synchronous backfilling and reinforcement module includes a slurry preparation system located on the shore. This system comprises a tailings hydrocyclone, a high-efficiency thickener, and a horizontal mixing tank. Low-concentration tailings discharged from the concentrator are concentrated by the hydrocyclone and then enter the thickener, where solid-liquid separation is achieved after the addition of flocculants. The resulting high-concentration slurry from the bottom is mixed with dry sand transported from the tail end of the reservoir in the mixing tank. The backfilling pump station transports the slurry to the goaf behind the mining platform via flexible backfilling pipelines laid on the floating bridge.

[0061] The dynamic monitoring and balance control module is a distributed control system. It connects not only to various physical sensors within the reservoir but also, via relay output modules, to the circuit breakers of the return water pumping station and the electro-hydraulic actuators of the guide weir. When the system determines that hydraulic balance adjustment is necessary, control commands are directly applied to these actuators, achieving an automated closed loop from monitoring to execution.

[0062] The decision support terminal is a physical terminal equipped with a high-resolution large-screen display and a professional graphics workstation. Deployed in the mine's dispatch center, it displays a digital twin image of the entire storage area through an interactive interface. Managers use this terminal to issue global mining plans, monitor the operational status of each module, and, when necessary, take over the automatic control logic for remote manual intervention.

[0063] The various modules mentioned above achieve seamless information flow through a fiber optic backbone network and dedicated industrial communication protocols. The entire system forms an organic whole, achieving precise navigation, dynamic balancing, and inherent safety in tailings dam mining operations under complex conditions through in-depth big data mining and real-time driving.

[0064] This embodiment describes in detail how the system maintains the overall volume and capacity stability of the reservoir area through material balance logic when mining high-value tailings.

[0065] As a whole, the safe operation of a tailings dam depends on maintaining sufficient flood control capacity and a stable dry beach. During the mining process, the removal of high-value tailings can cause a momentary increase in the local volume within the dam. Without scientific replacement, this volume change can lead to drastic fluctuations in the water level. This invention addresses this problem through a closed-loop logic based on mass conservation and volume compensation.

[0066] First, the system acquires the extraction flow rate of the floating sand mining platform in real time. By acquiring the flow velocity and concentration of the slurry in the delivery pipeline, the total mass of solid material removed from the storage area in the past minute is calculated. Since the tailings are saturated underwater, the system converts the mass value into the corresponding in-situ stockpile volume value based on the preset saturation density value.

[0067] Simultaneously, the synchronous filling and reinforcement module feeds back its filling data within the same time window to the system. The system acquires the instantaneous flow rate and solids content of the filling slurry and calculates the mass of solid material newly injected into the goaf. Considering that the filling slurry will settle and dehydrate within the goaf, the system introduces a volume conversion coefficient reflecting the filling efficiency. This coefficient is obtained through online monitoring of the early consolidation state of the filling body. The system multiplies the injected mass value by this volume conversion coefficient to obtain the actual filling volume occupying the goaf.

[0068] The system continuously subtracts the mined volume from the filled volume. If the difference is positive, it indicates that the goaf is expanding and the local stress in the reservoir area may be decreasing. In this case, the system will automatically send an acceleration command to the filling pump station to increase the filling rate by increasing the slurry delivery pressure. If the difference is negative, it indicates that the filling speed is too fast, which may cause abnormal rise in the local beach elevation and interfere with the water flow direction. In this case, the system will reduce the frequency of the filling pump.

[0069] Under this dynamic volume balance control, the effective volume of the tailings pond is always maintained within the designed safe range. This replacement is not a simple spatial exchange, but a stress field reconstruction process based on mechanical equilibrium. The system controls the gradient distribution of the filling material, so that the filling material near the dam body has a higher modulus, while the filling material near the tail end of the pond has better permeability. This differentiated physical characteristic configuration not only compensates for the spatial loss caused by mining, but also optimizes the distribution structure of the seepage lines inside the pond, making the tailings pond after mining even better than before mining in some key safety indicators.

[0070] In addition, the system monitors the total discharge of the return water system. Since the filling slurry introduces a large amount of water, the return water system needs to promptly discharge excess water to maintain a constant water level within the reservoir. The system calculates the difference between the amount of water introduced during filling and the amount removed by the return sand, and adds this value to the real-time evaporation and leakage rates to obtain a target discharge value reflecting the increase in water volume within the reservoir. By controlling the number of return water pumps started and stopped, the system ensures that the water level fluctuation within the reservoir does not exceed a preset centimeter range, thereby guaranteeing the stability of the dry beach length.

[0071] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A tailings recovery scheme optimization method based on big data driving, characterized in that: The method comprises the following steps: S1. Obtain the physical and mechanical characteristics and spatial distribution characteristics of tailings in the tailings area through a multi-source heterogeneous data acquisition terminal, and construct a three-dimensional digital fine model containing geometric topological relationship and mechanical property distribution based on a big data processing platform, wherein macroscopic terrain data and microscopic physical and mechanical parameters are associated in multiple dimensions by using a spatial interpolation algorithm, and discrete sampling points are fitted into a continuous tailings attribute gradient field in a three-dimensional coordinate system; S2. Value zoning evaluation is performed based on the three-dimensional digital fine model, a value evaluation index system is set, including tailings metal grade, particle size composition, mining difficulty coefficient, and environmental governance cost, the comprehensive value score of each spatial unit is calculated by using a preset weight distribution logic, and the tailings area is divided into a high-value recovery area to be mined, a low-value replacement area for replacement, and a safety monitoring and warning area to be protected according to the comprehensive value score and a preset threshold range; S3. A spatial replacement and recovery strategy is developed according to the zoning result, a floating sand mining platform is driven to reach a predetermined point corresponding to the high-value recovery area through a remote control instruction, a jet flow sand suction device or a cutter suction device is used for beach mining, the flow state of the material in the conveying pipeline is monitored through a flowmeter and a concentration meter during the mining process, and the pumping pressure is adjusted in real time, the extracted high-value tailings are transferred to a beneficiation plant through the conveying pipeline; S4. A goaf synchronous filling and stress reinforcement program is executed, the volume and geometric shape change of the goaf formed by the high-value recovery area are monitored in real time, the total amount of filling material required according to the mechanical stability requirement of the goaf is calculated, the low-value material generated by the recovery or the tailings dewatering material in the tailings area is pumped to the goaf through a replacement pipeline for spatial replacement, and the self-weight stress of the filling material is used to compensate for the in-situ stress field and maintain the stress balance in the tailings area; S5. The beach elevation, pore water pressure, and water edge line position data are obtained in real time through a full-tailings-area intelligent monitoring system, the real-time monitoring data are compared with a preset safety benchmark model to determine the safety redundancy under the current recovery working condition, a hydraulic balance regulation and control mechanism is dynamically triggered, the pumping amount of the water return system or the interception direction of the temporary diversion weir is adjusted to control the water level in the tailings area, so that the dry beach length and the buried depth of the wetting line are within the safety preset range.

2. The big data driven tailings dam recovery plan optimization method according to claim 1, characterized in that: Step S1 specifically comprises: Macroscopic terrain data of the tailings pond is obtained by using high-precision satellite positioning technology and unmanned aerial vehicle aerial survey technology; Tailings moisture content, density, particle size distribution, and shear strength parameters at different depths are collected by a sensor array arranged inside the tailings area, wherein each sensor node has an independent code and transmits the collected analog signals to a data concentrator after converting them into digital signals; The position vector in the three-dimensional coordinate system is extracted, a sample point set around the position vector is obtained, and the weight coefficient of each sample point to the target position attribute is determined according to the geometric distance between the sample point and the target position, the weight coefficient follows an inverse proportional decay relationship, and finally the attribute value of the target position is determined by the sum of the products of all relevant sample point attribute values and corresponding weights; By adopting multi-scale feature extraction logic, discrete point data obtained by drilling sampling is taken as a hard constraint, and wave velocity cloud obtained by geophysical exploration is taken as a soft constraint, a continuous tensor field representing material heterogeneity of the whole reservoir area is constructed by deep fusion, and the continuous tensor field records the evolution history of the reservoir area at different accumulation stages through time series index.

3. The big data driven tailings dam recovery plan optimization method of claim 1, wherein: The step S2 specifically comprises: Obtaining the original value of the metal grade of the target unit and dividing it by the benchmark price step of the target mineral in the current market to obtain a standardized component reflecting economic potential; Calculating the combined distance of the horizontal displacement and the vertical displacement of the center position of the target unit to the edge of the dam top, and comparing the combined distance with the preset safety warning radius, if the ratio is less than one, the comprehensive value score of the target unit is adjusted downward through the penalty function; Obtaining the value of the thickness of the overburden layer and the value of the maximum digging depth of the sand mining equipment, and calculating the mining difficulty coefficient through difference operation and residual proportion mapping; Using a heuristic search algorithm based on fuzzy comprehensive evaluation, the market price fluctuation step of the tailings, the energy loss of the transportation path and the additional disturbance generated by the stoping operation to the surrounding structure are comprehensively considered when calculating the comprehensive value score, and the mining sequence with the maximum net economic benefit and the minimum impact on the stability of the dam body is identified through iterative optimization.

4. The big data driven tailings dam recovery plan optimization method of claim 1, wherein: The step S3 specifically comprises: According to the spatial coordinate information of the high-value stoping area, the operation path and the draft depth of the floating sand mining platform are planned; Using the Beidou satellite navigation terminal and the autonomous obstacle avoidance module equipped on the floating sand mining platform, the navigation track is corrected according to the real-time generated beach topographic map, and the geometric shape of the underwater mining interface is detected in real time through the multi-beam sonar system; The flow rate and concentration of the slurry in the conveying pipeline are monitored in real time, and the measured values are compared with the preset plugging critical value, if the flow rate is lower than the preset lower limit of the flow rate, the operating frequency of the pumping motor is increased; Determine whether the mining slope detected by the multi-beam sonar system exceeds the natural repose angle of the tailings, if it exceeds, limit the drilling depth of the jet sand suction device or the cutter suction device and send a risk prompt signal.

5. The big data driven tailings dam recovery plan optimization method of claim 1, wherein: The step S4 specifically comprises: Obtaining the flow rate and concentration of the slurry in the conveying pipeline, calculating the total mass of solid materials removed from the reservoir area per unit time, and converting the total mass into in-situ accumulation volume value according to the preset saturation density value; Obtaining the instantaneous flow rate and solid content rate of the filling slurry input into the replacement pipeline, calculating the mass of solid materials newly injected into the mined-out area, and calculating the actual filling volume value occupying the mined-out area in combination with the volume conversion coefficient reflecting the early consolidation state of the filling body; Performing material balance logic, calculating the difference between the in-situ accumulation volume value and the filling volume value, and adjusting the operating frequency of the filling pump station according to the difference to maintain the stress balance in the reservoir; Adding a solidifying agent in proportion to the low-value tailings, injecting high-concentration filling slurry with specific rheological properties into the bottom of the mined-out area through the replacement pipeline, so that the filling body forms a support body with structural strength after consolidation.

6. The big data driven tailings dam recovery plan optimization method of claim 1, wherein: The step S5 specifically comprises: The distributed optical fiber sensing technology is used to capture the micron-level deformation and vibration signals of the dam body by sensing the frequency drift of the light wave in the sensing optical fiber laid transversely in the dam body. The time series recursive analysis method is used to analyze the water level observation data stream, and the long short-term memory neural network algorithm is used to calculate the damping coefficient reflecting the permeability of the reservoir body, and the vertical coordinate value of the phreatic line at the future time point is predicted. The difference between the vertical coordinate value and the beach elevation data is calculated to obtain the predicted buried depth value, and if the predicted buried depth value is less than the preset minimum safety protection distance, an emergency water pumping program is triggered. The difference between the water quantity value brought in by filling and the water quantity value taken away by sand returning is obtained, combined with the real-time evaporation and leakage to calculate the water quantity increment in the reservoir, and the start-stop number of the water returning pump is controlled to make the water level fluctuation amplitude of the reservoir within the preset centimeter level range.

7. The big data driven tailings dam recovery plan optimization method of claim 1, wherein: The water balance regulation mechanism specifically includes: calculating the hydraulic connection relationship between the high-value recovery area, the low-value replacement area and the clarifying tank in the reservoir; According to the real-time comparison result of the water flow kinetic energy and the beach anti-scouring capacity, the height and angle of the temporary diversion weir are adjusted to guide the water flow to the preset reinforced flood discharge channel to avoid water flow scouring the unconsolidated filling area; When the monitoring data deviates from the safety preset range, the water pumping quantity of the water returning system or the interception direction of the temporary diversion weir is automatically adjusted to control the water level in the reservoir by changing the hydraulic gradient, and to ensure that the dry beach length meets the specific proportion requirement of the flood control standard.

8. The tailings recovery scheme optimization system based on big data driving, applied to the tailings recovery scheme optimization method based on big data driving according to any one of claims 1-7, characterized in that: It includes: A multi-source data acquisition module is used to acquire topographic data, tailings physical and mechanical parameters, and spatial distribution data of the tailings pond, and the multi-source data acquisition module includes a fixed monitoring station, a mobile monitoring unit, and a sensor array arranged inside the reservoir area; A digital modeling module is connected with the multi-source data acquisition module and is used to construct a dynamically updated three-dimensional digital fine model through a high-performance computing cluster; An intelligent partition evaluation module is used to perform value gradient analysis and safety risk evaluation based on the three-dimensional digital fine model to determine the high-value recovery area, the low-value replacement area, and the safety monitoring warning area; A recovery operation execution module includes the floating sand mining platform and the pipeline transportation subsystem, which is used to extract and transport high-value tailings; A synchronous filling and reinforcement module is configured to pump low-value materials into the empty space generated by recovery to achieve spatial replacement and mechanical compensation; A dynamic monitoring and balance control module is used to monitor key safety parameters of the reservoir area and feedback adjust the hydraulic system to maintain the overall dynamic balance of the reservoir area.

9. The tailings pond recovery scheme optimization system based on big data driving according to claim 8, characterized in that: The fixed monitoring station in the multi-source data acquisition module includes a total station and a laser scanner, and the mobile monitoring unit includes a multi-spectral camera-mounted unmanned aerial vehicle and a sensor mounted on the floating sand mining platform. The synchronous filling reinforcement module comprises a slurry preparation system located at the bank side, which is composed of a tailings cyclone, a high-efficiency thickener, a horizontal stirring tank and a filling pump station, and is connected with the goaf through flexible filling pipelines laid on the floating bridge; The dynamic monitoring and balance control module comprises a distributed control system connected to the circuit breaker of the backwater pump station and the electric hydraulic push rod of the temporary diversion weir through a relay output module.

10. The big data driven tailings dam recovery plan optimization system of claim 8, wherein: Further comprising: A decision support terminal configured with an interactive visualization interface for displaying a digital twin mirror image of the stockyard, wherein the digital twin mirror image comprises a color heat map of material grade, a stress distribution cloud chart and a safety risk early warning indication; The decision support terminal supports simulation of the stoping scheme, compares economic benefits and safety factors under different spatial displacement paths by adjusting simulation parameters, and issues execution instructions to the stoping operation execution module and the synchronous filling reinforcement module according to the simulation results.

Images