Mine exploitation practical training method and system
By constructing a cloud-based 3D virtual spatiotemporal model and Bayesian network evaluation, the problems of high risk, high cost, and inaccurate evaluation in mining training have been solved, enabling safe and low-cost large-scale multi-role collaborative training and improving teaching effectiveness and training efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- WUHAN INST OF TECH
- Filing Date
- 2026-03-06
- Publication Date
- 2026-06-19
Smart Images
Figure SMS_1 
Figure SMS_2 
Figure SMS_10
Abstract
Description
Technical Field
[0001] This invention relates to the field of virtual simulation training technology, specifically to a mining training method and system. Background Technology
[0002] Open-pit mining is a core branch of mining engineering, and the industry places stringent demands on practitioners' cross-role collaboration abilities, time-space planning capabilities, and career adaptability. Currently, the practical training teaching and existing training systems for mining engineering majors in universities suffer from the following key technical deficiencies: 1. Key pain points in on-site training Practical training in open-pit mines involves high-risk scenarios such as blasting operations and inspection of steep slopes, with an average annual accident rate of about 0.3%. The entire life cycle of a mine spans more than 5 years, which cannot be completed within the 3-4 year teaching cycle of universities. At the same time, the cost of equipment rental and site coordination for on-site training is as high as 500,000 yuan per batch, making it difficult for most universities to carry out such training regularly.
[0003] 2. The problem of spacetime separation Existing mining simulation systems are mostly stand-alone or local area network VR systems, lacking a unified GIS spatial coordinate system and standardized time axis. The training operations of different roles cannot be synchronized in the same spatiotemporal scenario, and cannot simulate the real collaborative operation process of mine design, production, and supervision.
[0004] 3. Limited deployment and scalability Traditional systems are mainly deployed on local PCs and do not support elastic scaling, which cannot meet the large-scale training needs of college classes (≥50 people).
[0005] 4. Lack of career orientation and individualization The existing system uses fixed task templates and does not customize training content based on students' career plans (mining companies / regulatory departments / design institutes). The matching degree between training results and employment positions is less than 40%, which fails to achieve precise talent cultivation.
[0006] 5. The evaluation mechanism is imperfect. Current practical training evaluations only focus on basic indicators such as task completion and compliance, lacking process-oriented evaluation of higher-order abilities such as critical thinking, innovative design, teamwork, and problem-solving; evaluation data only collects the final operation results and does not record the behavioral trajectory of the entire practical training process, making it impossible to achieve dynamic and accurate ability assessment.
[0007] The development of cloud computing, 3D virtual simulation and AI large model technology has provided a foundation for solving the above problems. However, there is currently no technical solution that can deeply integrate a unified spatiotemporal collaboration mechanism, career planning-driven personalized education, high-level competency process evaluation and open-pit mining training. This has become a technical gap that urgently needs to be filled in this field. Summary of the Invention
[0008] The purpose of this application is to address the shortcomings of the aforementioned background technology and provide a mining training method and system.
[0009] The technical solution of this application is: a mining training method, including: Construct a three-dimensional virtual spatiotemporal model of the case mine, and configure the time axis and associated spatial region parameters for each stage of the mine's life cycle; Collect trainees' career planning information and assign them roles that match their career plans; Synchronize the spatiotemporal reference of all terminals, and assign collaborative training tasks with unique spatiotemporal stamps to different roles based on trainees' career plans and competency profiles. Trainees interact with the 3D virtual mine model on the terminal. The operation data is generated with a corresponding spatiotemporal stamp and uploaded to the cloud. The cloud collects process data of the entire training process in real time. Based on operation records with time and space stamps and process data, trainees are dynamically and quantitatively evaluated from two dimensions: process indicators and core competencies.
[0010] According to the mining training method provided in this application, the method for constructing a three-dimensional virtual spatiotemporal model of a case mine includes: constructing an interactive three-dimensional spatial model based on real geological data and GIS spatial data of the mine, the model having a unified spatial coordinate system and a standardized time axis; the spatial coordinate system is used to accurately calibrate the geographical coordinates of the mine slope location, mining face, and spoil heap area.
[0011] According to the mining training method provided in this application, the process indicators include timeliness achievement rate, task completion quality, inter-role communication and collaboration, compliance and innovation improvement, and different weights are assigned to each indicator for three roles: mine design engineer, mine technical administrator and mining supervision specialist.
[0012] According to the mining training method provided in this application, the core competencies include critical thinking, innovative design, teamwork, and problem-solving. The multi-dimensional quantitative evaluation adopts a Bayesian network model, with the four core competencies as hidden nodes and problem diagnosis accuracy, parameter adjustment innovation, communication response efficiency, conflict resolution timeliness, and conflict resolution effectiveness as observable nodes. Through prior probability initialization and posterior probability dynamic updating, the core competencies of trainees are dynamically quantitatively evaluated.
[0013] According to the mining training method provided in this application, the prior probability is calculated based on the role professional ability benchmark requirements and the results of the pre-training cognitive test; the posterior probability is updated based on the observation node data of the corresponding stage in the initial, middle and final stages of the training, so as to dynamically optimize the core ability score.
[0014] According to the mining training method provided in this application, cross-role collaborative verification is performed on the operation data to verify whether the operation content of different roles in the same time and space area meets the preset collaborative rules and safety compliance requirements; if operation conflicts or violations are detected, early warning information is generated and the conflict location is visually marked in the three-dimensional virtual mine model.
[0015] According to the mining training method provided in this application, the cross-role collaborative verification method includes: verifying whether the mining boundary parameters set by the design engineer in the three-dimensional model are coordinated with the production equipment scheduling plan formulated by the technical administrator in the same spatial area and at the same time node; and verifying whether the above parameters and plan meet the safety and compliance constraints proposed by the supervisor based on the preset safety procedures.
[0016] According to the mining training method provided in this application, the process data includes, but is not limited to: collaborative communication records between roles, design parameter modification records, conflict resolution operation trajectories, and problem diagnosis behavior data. All data are associated with the spatiotemporal stamps of their corresponding operations and stored in a cloud database. The interactive operations include adjusting the parameters of the 3D model, drawing or editing the mining design scheme, and adding regulatory inspection marks.
[0017] According to the mining training method provided in this application, the time nodes, spatial scope and task complexity of subsequent training tasks are dynamically adjusted based on the evaluation results; for the role of mine design engineer, the complexity of innovative design elements in the tasks is increased and their weight in the evaluation is increased; for the role of mine technical administrator, more complex equipment scheduling and fault handling scenarios are introduced; and for the role of mining supervision specialist, the depth and breadth of safety compliance review are strengthened.
[0018] This application also relates to a mining training system for implementing the above-mentioned mining training methods, including: The cloud service layer, deployed on a cloud server cluster, includes a unified spatiotemporal engine, a cloud database, a core algorithm cluster, and a 3D virtual mine model module. The unified spatiotemporal engine provides a unified spatial coordinate system and a standardized time axis, and generates corresponding spatiotemporal stamps for all user operations. The core algorithm cluster includes a personalized recommendation engine, a cross-role collaborative verification algorithm, and a high-order capability process evaluation module. The network communication layer uses the WebSocket protocol to achieve low-latency bidirectional communication between the cloud service layer and the terminal. The terminal interaction layer includes multiple terminal devices used to bind different professional roles, receive training tasks, upload operation data, and display a 3D virtual mine model, collaborative conflict prompts, and dynamic evaluation results.
[0019] The advantages of this application are as follows: 1. This application relates to a mining training method. By constructing a three-dimensional virtual spatiotemporal model of a case study mine, this application fundamentally avoids on-site training in real high-risk mines, directly solving the core problems of high risk and high cost. The virtual model can compress the long life cycle of a mine. By configuring the time axis and associated spatial area parameters and synchronizing the spatiotemporal reference of all terminals, a unified virtual spatiotemporal reference system is established for all participating roles, ensuring that different roles operate in the same virtual mine and at the same time, solving the problems of spatiotemporal fragmentation and poor collaboration in existing systems. By collecting career planning information and assigning matching roles and distributing collaborative tasks based on career planning and competency profiles, the training tasks are transformed from generalized to personalized. The training content is directly linked to students' career development goals, realizing precise talent cultivation driven by career planning. By uploading operation data with spatiotemporal stamps and conducting dynamic quantitative evaluation based on the data with spatiotemporal stamps, the evaluation no longer depends solely on the final result, but is based on traceable behavioral data throughout the entire process. This provides a data foundation for objectively and quantitatively evaluating process indicators and core competencies, forming a closed-loop evaluation process for precise talent cultivation. 2. This application ensures the authenticity, accuracy, and interactivity of the virtual training environment; the model is constructed based on real geological data and GIS spatial data, making the virtual environment highly consistent with engineering practice, enhancing the immersion and teaching effectiveness of the training, and enabling students to directly transfer the skills they acquire to real work scenarios; the unified spatial coordinate system and accurate geographic coordinate calibration ensure the accurate location information of key areas such as slopes and working faces; this provides the technical prerequisite for subsequent collaborative operation verification (such as whether the equipment is operating in the designated area), safety compliance review (such as safety distance calculation), and accurate spatial data analysis, and is the spatial foundation for improving the "accuracy of spatiotemporal collaboration"; 3. This application clarifies the evaluation indicators and their personalized configurations, making the evaluation system more targeted, scientific, and fair; it clarifies five process indicators, including timeliness achievement rate and task completion quality, covering multiple professional competence levels such as skills operation, compliance awareness, communication, and innovation, resulting in a more comprehensive evaluation; it assigns different weights to each indicator for the three roles of design engineer, technical administrator, and regulatory specialist; the weight for innovation and improvement may be higher for design engineers, while the weight for compliance is most critical for regulatory specialists; this differentiated weight design allows the evaluation results to more accurately reflect the job competence of specific roles, strengthening the precision of career orientation; 4. This application constructs an advanced evaluation model that enables a scientific, dynamic, and quantifiable assessment of abstract higher-order abilities. It employs a Bayesian network model, designating four abstract core abilities, including critical thinking and innovative design, as hidden nodes, while specific behavioral data such as problem diagnosis accuracy and parameter adjustment innovation are designated as observable nodes. This establishes a scientific reasoning mechanism for inferring intrinsic hidden abilities from externally observable behaviors. Bayesian networks can handle uncertainty and simulate the mutual influence between abilities through probabilistic relationships, making them more scientific than simple weighted scoring and representing an advanced technological application in the field of ability evaluation. 5. This application further refines the dynamic nature of the evaluation model, realizing personalized initialization and accompanying growth tracking of the evaluation; the prior probability is calculated based on the weighted average of the role professional ability benchmark and the results of the pre-training cognitive test, which means that the system can identify the differences in the initial abilities of different trainees and establish a personalized ability assessment baseline for each person, rather than a one-size-fits-all approach; making the ability evaluation not a static endpoint score, but a dynamic curve that is constantly corrected and approaches the true level as the training progresses, intuitively reflecting the trainee's ability growth trajectory throughout the entire cycle, and constituting key feedback information in the precise talent cultivation closed loop; 6. This application adds a real-time collaborative verification and conflict resolution mechanism. Its advantages lie in forcibly cultivating students' awareness of collaborative work and safety compliance, and providing immediate feedback. By verifying whether the operations of different roles in the same spatiotemporal area meet the collaborative rules, it directly simulates the planning conflicts, resource conflicts, and safety conflicts that may occur when multiple departments collaborate in real mine production. By generating early warning information and visually annotating the conflict locations in the 3D model, it provides low-latency (thanks to WebSocket) real-time feedback. This allows students to immediately recognize the contradictions in their operations and intuitively see the conflict points in the virtual environment, greatly improving teaching efficiency and immersion. It is one of the core technical guarantees for the significant improvement in the accuracy of spatiotemporal collaboration. 7. This application takes the coordination of mining boundary parameters, production equipment scheduling schemes, and safety compliance constraints as an example to demonstrate the multi-dimensional game between technical feasibility, economic efficiency, and safety management in mine design; the system automatically verifies this complex relationship through algorithms, training students to establish system thinking and a global perspective, which is an advanced teaching goal that cannot be achieved by traditional independent training based on job positions; 8. This application clarifies the types of process data, reveals the depth and breadth of data collection, and provides rich material for high-precision evaluation; by associating all data with the spatiotemporal stamps of the corresponding operations, it ensures that every communication record and every parameter modification can be traced back to a specific scenario in virtual spacetime, making the behavior analysis highly contextual; it clarifies that interactive operations include parameter adjustment, scheme drawing, and the addition of regulatory markers, defines the scope of user behavior in the system response, and reflects the complete functionality of the system and the clear boundaries of patent protection; 9. This application realizes an adaptive personalized training path. By dynamically adjusting subsequent training tasks based on evaluation results, it enhances the complexity of innovative design elements for design engineers, which is a direct response to the evaluation results of their innovative design capabilities. This dynamic adjustment mechanism enables the system to intelligently strengthen training for each trainee's weaknesses or provide advanced challenges for their strengths, truly realizing personalized and precise talent cultivation under large-scale training, and greatly improving training efficiency. 10. This application defines the hardware and software architecture for implementing the above method, ensuring the efficient, stable, and scalable operation of the method process at the system level; the cloud service layer is deployed on a cloud server cluster, integrating a spatiotemporal engine, database, core algorithms, and models; this architecture provides powerful computing capabilities, supports concurrent access from multiple terminals and real-time operation of complex algorithms (such as Bayesian networks and collaborative verification), ensuring system performance and stability, and significantly reducing the cost of single-batch training; the network communication layer specifically ensures real-time bidirectional communication, which is the network foundation for realizing real-time collaborative operation and instant conflict warning; the terminal interaction layer supports multi-terminal access, demonstrating the system's cross-platform compatibility and ease of use, and supporting various training scenarios (such as classrooms and remote locations). Detailed Implementation
[0020] The embodiments of this application are described in detail below, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The following embodiments are exemplary and intended to explain this application, and should not be construed as limiting this application.
[0021] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0022] The present application will be further described in detail below with reference to specific embodiments.
[0023] This application relates to a mining training method. It constructs a unified spatial coordinate system and virtual timeline through a cloud service layer, providing synchronized spatiotemporal references for all terminals and generating operation spatiotemporal stamps. Combined with low-latency communication guaranteed by the network communication layer and multi-role access at the terminal interaction layer, it enables real-time collaborative work among mine design engineers, technical administrators, and supervisors in the same 3D virtual mine scenario. The method customizes personalized training tasks by collecting career planning information, collects and verifies multi-role operation data with spatiotemporal stamps in real time, and, based on this process data, integrates process indicators such as task completion rate with core competency indicators based on Bayesian network dynamic evaluation to conduct multi-dimensional quantitative evaluation of trainees. Finally, it dynamically optimizes subsequent training paths based on the evaluation results, thus constructing a career-oriented, precise talent cultivation closed loop integrating virtual simulation, real-time collaboration, process evaluation, and adaptive optimization. This solution effectively solves the technical problems of high risk and cost, long cycle, and spatiotemporal fragmentation and poor collaboration in traditional training, as well as existing simulation systems. Specifically, the mining training method of this application is carried out according to the following steps: S1. Construct a three-dimensional virtual spatiotemporal model of a case study mine, and configure timelines representing each stage of the mine's life cycle (such as exploration, design, production, and mine closure and reclamation) and spatial region parameters associated with each stage in the model. S2. Collect career planning information of trainees (such as students or employees) and assign corresponding virtual roles according to their plans (such as wanting to become a design engineer, technical administrator or supervisor). S3. The system synchronizes the spatiotemporal reference of all training terminals and assigns collaborative tasks with unique spatiotemporal stamps to different roles based on the trainees' career plans and historical ability profiles. S4. During the training, trainees interact with the 3D virtual mine model through the terminal (such as adjusting parameters and drawing schemes). Their operation data is generated with a corresponding time stamp and uploaded to the cloud in real time. The cloud synchronously collects various process data of the whole process. S5. Based on these operation records and process data with time and space stamps, the system dynamically and quantitatively evaluates trainees from two dimensions: process indicators (such as task completion status) and core competencies (such as innovation ability).
[0024] The training method described in this application constructs a digital twin training environment. Through a unified spatiotemporal benchmark and spatiotemporal stamp technology, it integrates the traditionally fragmented and disparate offline training processes into a coherent and collaborative virtual spacetime. Career planning drives role allocation, ensuring personalized training goals. Full-process data collection provides a data foundation for objective evaluation, while dual-dimensional dynamic evaluation enables comprehensive monitoring of the learning process and skill development.
[0025] The training method proposed in this application completely avoids the high risks of on-site training and compresses the mine lifecycle training, which can take several years, into a few weeks, significantly reducing the cost per batch. The unified spatiotemporal benchmark solves the fundamental problem of existing simulation systems where training is conducted independently and there is a lack of coordination, laying the foundation for true multi-role collaborative operations. Starting from career planning, it achieves job-specific training and job-specific training, with training content closely aligned with individual development goals. Based on dynamic quantitative evaluation using full-process, multi-dimensional data, it shifts from a result-oriented approach to a developmental evaluation that emphasizes both process and ability.
[0026] In some embodiments of this application, this embodiment specifically illustrates a method for constructing a three-dimensional virtual spatiotemporal model. Based on real geological exploration data (such as borehole data, strata information) and Geographic Information System (GIS) data (such as topographic maps, administrative divisions) of the target mine, a high-fidelity, interactive three-dimensional spatial model is constructed using a three-dimensional modeling engine (such as Unity3D, Unreal Engine, or a professional mining software platform). This model incorporates a unified spatial coordinate system (such as the National Geodetic Coordinate System 2000) and a standardized time axis (which can compress real time proportionally, for example, 1 virtual day represents 1 real year). The spatial coordinate system is used to accurately georeference key areas in the model, such as the specific location of slopes, the corner coordinates of the current mining face, and the boundary range of the spoil heap.
[0027] By leveraging real-world data to drive modeling, the engineering realism and geographical accuracy of the virtual environment are ensured. A unified spatial coordinate system serves as the benchmark for all spatial operations and data associations, and is a prerequisite for subsequent spatial analysis and conflict detection. A standardized timeline accelerates the simulation of the long mine lifecycle and enables all events and state changes to be managed and traced on a unified time scale.
[0028] In practice, we collect original geological reports, topographic maps, mining plans, and other data from the case mines; import GIS data into modeling software to generate digital elevation models (DEMs) and surface models; construct three-dimensional models of underground strata and ore grades based on geological data; set a global coordinate system in the software and bind the time axis system to model state changes (such as mining progress and spoil heap growth) in the programming; and perform spatial coordinate calibration and attribute definition for key areas such as slopes, working faces, transportation roads, and spoil heaps.
[0029] This embodiment ensures a high degree of simulation and data traceability in the training. Precise geographic coordinate calibration ensures a one-to-one correspondence between virtual operations and real-world locations, greatly enhancing the seriousness and technical value of the training. A unified spatiotemporal framework is the cornerstone of all subsequent collaboration, verification, and analysis functions, making it possible to reproduce and optimize real workflows in the virtual world.
[0030] In other embodiments of this application, this embodiment defines in detail the process evaluation indicators and their differentiated applications. The process indicators include: timeliness achievement rate (whether the task is completed within the specified virtual time), task completion quality (such as the resource recovery rate of the design scheme, the efficiency of the equipment scheduling scheme), inter-role communication and collaboration (the number and quality of proactively initiated / responded collaborations), compliance (whether the operation complies with virtual safety procedures and environmental standards), and innovation and improvement (whether a solution superior to the standard process has been proposed and implemented). For the three roles of mine design engineer, mine technical administrator, and mining supervision specialist, the system configures different weighting coefficients for the same set of indicators.
[0031] Specifically, the timeliness achievement rate (A) refers to the proportion of practical training tasks completed within the specified time. The time stamp is the core data basis, and the calculation method is: Timeliness achievement rate = (number of tasks completed on time / total number of tasks) × 100. Points are deducted proportionally for tasks that are overdue (50% deduction for overdue within 10%, 80% deduction for overdue between 10% and 30%, and 0 points for overdue more than 30%). The weighting coefficient for the timeliness achievement rate of technical administrators is 25%, that of design engineers is 20%, and that of supervisors is 30%.
[0032] Task completion quality (B) refers to the degree to which the completed task meets the training standards, industry norms, and role responsibilities. The calculation method is: Task completion quality = (Number of task deliverables that meet the standards / Total number of task deliverables) × 100. If a key deliverable fails to meet the standard, the item will receive a score of 0 (such as the blasting parameter design of mining technicians, the general layout plan of designers, and the compliance checklist of supervisors). The task completion quality weighting coefficient for technical administrators is 30%, for design engineers it is 35%, and for supervisors it is 25%.
[0033] Inter-role communication and collaboration (C) refers to the frequency of communication with the other two roles, the accuracy of information transmission, and the timeliness of collaborative response. The calculation method is: Communication and collaboration score = (Number of effective communications / Total number of communications) × 60 + (Number of timely collaborative responses / Number of collaborations required) × 40. Effective communication is defined as the complete transmission of information and clear feedback from the other party. The time stamp records the difference between the communication time and the response time (a response time difference ≤ 2 hours is considered timely). The inter-role communication and collaboration weight coefficient for technical administrators is 20%, for design engineers it is 25%, and for supervisors it is 20%.
[0034] Compliance (D) refers to the degree to which the operation process and results comply with the relevant laws, regulations, industry standards and training specifications for open-pit mining. The calculation method is: Compliance Score = 100 - (Number of violations × 10). Major violations (such as violations of safe mining specifications, mandatory clauses of design standards, and core regulatory requirements) will receive a score of 0. Violations will be determined by comparing operation records with the standard clauses. The compliance weighting coefficient for technical administrators is 15%, for design engineers it is 10%, and for regulatory specialists it is 20%.
[0035] Innovation and improvement (E) refers to the feasibility and value of optimization suggestions and improvement plans proposed during task execution. The calculation method is: Innovation and improvement score = (number of effective innovation suggestions × 25). Effective innovation suggestions must meet the requirements of "implementable, supported by data, and able to improve efficiency / quality / safety". They are reviewed and confirmed by the training instructor in combination with industry practice. The innovation and improvement weight coefficient for technical administrators is 10%, the innovation and improvement weight coefficient for design engineers is 10%, and the innovation and improvement weight coefficient for supervisors is 5%.
[0036] Formula for calculating the comprehensive score of process indicators: Technical Administrator V1 = A × 25% + B × 30% + C × 20% + D × 15% + E × 10%; Design Engineer V2 = A × 20% + B × 35% + C × 25% + D × 10% + E × 10%; Regulatory Commissioner V3 = A × 30% + B × 25% + C × 20% + D × 20% + E × 5%.
[0037] By establishing a multi-dimensional performance evaluation system that is strongly correlated with job responsibilities, and assigning differentiated indicator weights to different roles, the evaluation results can more accurately reflect the performance of the core responsibilities of the role, achieve job suitability in the evaluation, and guide trainees to focus on the key performance areas required for their future positions.
[0038] In practice, the system administrator presets the indicator weight templates for the three types of roles in the background evaluation module; when a trainee is assigned a role, the corresponding weight template automatically takes effect; the cloud algorithm calculates the trainee's score on each indicator based on the collected operation data; the final process evaluation score is obtained by multiplying the scores of each indicator by their respective role weights and then summing them.
[0039] Advantages: This embodiment achieves refined and targeted evaluation. It changes the one-size-fits-all evaluation method, enabling different types of positions, such as mine design, production management, and safety supervision, to receive fair and targeted assessments within the same system. This not only makes the evaluation results more convincing but also acts as a guide, directing trainees to focus on developing the core competencies valued in their target positions during practical training, thus strengthening the career-oriented educational role.
[0040] In a further embodiment of this application, this embodiment discloses a method for dynamically evaluating higher-order core competencies using a Bayesian network model. The core competencies include critical thinking, innovative design ability, teamwork ability, and problem-solving ability. A Bayesian network is constructed, with the above four core competencies as hidden nodes (not directly observable), and observable data generated during training as observation nodes. For example: problem diagnosis accuracy (related to critical thinking), parameter adjustment innovation (related to innovative design ability), communication response efficiency (related to teamwork ability), and conflict resolution timeliness and effectiveness (related to problem-solving ability). During model initialization, the prior probability is calculated by weighting the professional competency benchmark requirements for the role (general job standards) and the results of the cognitive test conducted before training (individual initial level). During the training process, the model dynamically updates the posterior probability: in the initial stage (e.g., the first 1 / 3 of the time), the middle stage (the middle 1 / 3 of the time), and the final stage (the last 1 / 3 of the time), observation node data within the corresponding time period are collected, input into the Bayesian network for inference, and the probability estimates of the trainee's four core competencies are updated, thereby dynamically optimizing their core competency scores.
[0041] Prior probability is calculated based on a weighted average of the role's professional competence benchmark requirements and the results of initial cognitive tests. For example, prior probability can be calculated using the following formula: in: P(C i ) —The prior probability of the i-th core competency. C 1 For critical thinking, C 2 For innovative design capabilities, C 3 For teamwork skills, C 4 For problem-solving ability; α —The weight of the professional competence benchmark is 0.6 (industry benchmarks are dominant). P std (C i )—The industry benchmark probability of the i-th ability for this role (e.g., for design engineers) P std (C i ) =Excellent percentage: 35% P std (C i ) =Good percentage: 40% P std (C i ) =Generally accounting for 20%, P std (C i ) =Poor performance accounted for 5%) 1-α —Initial test weight, valued at 0.4; P test (C i ) —The probability of assessing the i-th ability in the initial role cognition test of trainees (derived from the test answer accuracy rate).
[0042] The posterior probability of updating the core capability (hidden node) based on the data of the observed nodes can be calculated, for example, using the following formula: in: —Given data from n observation nodes O 1 Accuracy of problem diagnosis O 2 When modifying parameters (such as innovation), the posterior probability of the i-th core capability; —Given the i-th core capability, what is the joint conditional probability of all observed nodes occurring simultaneously, assuming the observed nodes are independent (Naive Bayes assumption)? = ; P(C i ) —The prior probability (or the posterior probability after the previous stage update) of the i-th core capability. —The marginal probability of all observed nodes occurring simultaneously, used for normalization, is given by the formula: = .
[0043] The posterior probability distribution is transformed into a quantitative score of 0-100 points, using a weighted summation formula: in: —Quantitative score of the i-th core competency; level —Ability level (1=Excellent, 2=Good, 3=Average, 4=Poor); S(level) — Corresponding scores for each level (Excellent 95 points, Good 82 points, Average 67 points, Poor 30 points, taking the midpoint of each level range).
[0044] The design of Bayesian network nodes is shown in the table below.
[0045] Table 1: Bayesian Network Node Design Based on the professional competency benchmarks for three roles in the open-pit mining industry, and combined with the results of role cognition tests during the initial training phase, the prior probability distributions of four core competencies were initialized. For example: Technical Administrator: Prior Probability of Critical Thinking (Excellent: 30%, Good: 40%, Average: 20%, Poor: 10%) Design Engineer: Prior Probability of Innovative Design Capability (Excellent: 35%, Good: 40%, Average: 20%, Poor: 5%) Regulatory Commissioner: Prior probability of critical thinking related to compliance (Excellent: 40%, Good: 35%, Average: 20%, Poor: 5%).
[0046] The practical training is divided into three stages: initial, intermediate, and final. After each stage, the observation node data (O1-O5) of that stage are extracted. Using Bayes' theorem P(hidden node|observed node) = P(observed node|hidden node) × P(hidden node) / P(observed node), the posterior probability distribution of the four core competencies is updated and converted into a quantitative score (Excellent: 90-100 points, Good: 75-89 points, Average: 60-74 points, Poor: 0-59 points). Specific process: 1. Initial Stage: Based on prior probabilities and initial operational records (such as problem diagnosis of basic tasks and initial collaborative communication), generate the first version of the competency score; 2. Mid-term: Incorporate mid-term operation records (such as parameter modification for complex tasks, multi-round collaborative conflict resolution), update posterior probabilities, and revise capability scores; 3. Post-processing: Integrate the entire cycle of operation records, finally update the posterior probability, and output a predictive quantitative score of core competence (predicting the role's performance in actual work).
[0047] Bayesian networks are a probabilistic graphical model-based method for uncertainty reasoning, well-suited for linking observable concrete behaviors with abstract, indirect abilities. The principle is to continuously incorporate new evidence (observational data) to adjust the reliability (probability) of latent variables (core competencies). Prior probabilities set an initial reliability that aligns with job requirements and the individual's starting point, while posterior probability updates enable evidence-based continuous learning and characterization of ability development trajectories.
[0048] In practice, domain experts (education experts, senior engineers) define the conditional probability relationship between core competencies and observed behaviors, and construct a Bayesian network structure; trainees complete an onboarding cognitive test, and the system generates a personalized prior probability distribution of competencies based on their role standards; the system automatically extracts observation data periodically (e.g., by training stage), such as analyzing response efficiency from communication records and analyzing innovation from solution version history; the evidence is input into the Bayesian network inference engine to calculate the updated posterior probability distribution, which serves as the competency score for that stage; the posterior probabilities of competencies at each stage are displayed to trainees and instructors in the form of trend charts or radar charts.
[0049] This embodiment achieves a scientific and dynamic assessment of abstract, higher-order abilities. Employing a probabilistic model objectively addresses the uncertainty and indirectness in ability assessment, making it more convincing than subjective scoring or simple weighted summation. Phased updates to the ability profile clearly present the trainee's ability growth curve during training, rather than a static final score. Based on prior knowledge of individual starting points and personalized behavioral evidence, the assessment results more accurately reflect an individual's progress and weaknesses, providing a precise basis for personalized feedback.
[0050] In some preferred embodiments of this application, methods for ensuring consistency and security in multi-role collaborative operations are described in detail. The system performs cross-role collaborative verification on uploaded operation data to verify whether the operations of different roles within the same spatiotemporal area meet preset collaborative rules and safety compliance requirements. Specific methods include: verifying whether the mining boundary parameters (such as the final slope angle) set by the design engineer in the 3D model are coordinated with the production equipment scheduling plan (such as the operating position of a large electric shovel) formulated by the technical administrator in the same spatial area and at the same time node (e.g., whether the electric shovel's operating range exceeds the designed safety boundary); and simultaneously verifying whether the above parameters and plans comply with the safety and compliance constraints (such as the minimum working platform width) proposed by the supervisor based on preset safety regulations (such as the "Safety Regulations for Metal and Non-metal Mines"). If operational conflicts or violations are detected, an early warning message is immediately generated, pushed to the terminals of the relevant roles, and the conflict location is visually marked with highlights, flashing, etc., in the 3D virtual mine model.
[0051] By introducing a virtual collaborative rule engine, which incorporates the working logic and safety red lines between various professions in mining production, and compares the operation instructions of different roles in the same time and space coordinates in real time, the engine can instantly discover the contradictions between design, planning and regulatory requirements, just like coordination meetings or safety inspections in real projects. It also visualizes abstract conflicts, forcing trainees to solve problems in a collaborative and compliant manner.
[0052] In actual operation, collaborative constraints are defined in the form of rule scripts in the system backend (e.g., the equipment working face must be within the designed mining boundary and at least 10 meters away from the slope line); the cloud service monitors all operation data streams with time and space stamps; when it is detected that key data (design parameters, scheduling instructions, monitoring markers) involving multiple roles are updated in the same spatiotemporal area, the verification algorithm is triggered; the algorithm calls the rules to make a judgment, and if the rules are violated, an alarm is generated, which includes the conflicting roles, conflicting rules, and conflicting locations (3D coordinates); in the 3D scene, the conflict area is rendered as a red warning box, and the relevant role terminals are refreshed and displayed synchronously.
[0053] This embodiment improves the accuracy of multi-role collaboration from relying on individual self-awareness to mandatory system verification, significantly enhancing the accuracy of spatiotemporal collaboration; it transforms safety procedures from textual clauses into hard constraints that take effect in real time within the system, making compliant operation the only feasible path and deeply internalizing safety awareness; visual conflict forces trainees to transcend their own role limitations and understand the complex relationships of engineering systems from a global perspective, which is an effective means of cultivating engineering systems thinking.
[0054] In some embodiments of this application, the key procedural data types collected by the system and the supported interactive operations are clearly defined. The procedural data includes, but is not limited to: records of collaborative communication between roles (text, voice), historical records of design parameter modifications, operational trajectories performed to resolve conflicts (such as the steps of repeatedly adjusting the plan), and behavioral data during problem diagnosis (such as which geological data was consulted and what analytical tools were used). All of this data is associated with a spatiotemporal stamp (time + spatial coordinates) generated by a unified spatiotemporal engine at the time of the operation and is structured and stored in a cloud database. The interactive operations supported by the terminal mainly include: adjusting parameters of geological bodies and equipment attributes in the 3D model (such as sliders and input boxes); drawing, editing, and annotating the boundary lines of the mining design plan; and having supervisors add safety inspection markers, violation screenshots, and rectification notices.
[0055] This embodiment not only collects results, but also focuses on the collection process and intent, giving each behavioral segment a precise spatiotemporal label, so that it can be reviewed later and the training scenario and decision-making process at any time and any place can be fully reconstructed, providing unprecedented rich materials for deep learning and analysis.
[0056] During operation, the system incorporates data acquisition interfaces at all possible interaction points during its design phase. Any interaction between the trainee and the model (clicking, dragging, inputting) will trigger a data acquisition event. The data acquisition event processor packages the action content, the current 3D scene camera perspective, the model status, and the spatiotemporal stamp into a data packet. The data packet is then uploaded in real time to the corresponding data table in the cloud database (such as the operation log table and communication record table) through the communication layer.
[0057] This embodiment establishes the data foundation for advanced competency assessment and instructional analysis. Rich, detailed, and contextualized process data forms the basis for advanced analytical algorithms such as Bayesian networks, enabling assessments to extend beyond final outcomes and delve into thought processes, collaborative strategies, and problem-solving pathways. This provides solid data support for achieving true process-oriented assessment and learning analysis.
[0058] In other embodiments of this application, this embodiment describes a method for dynamically optimizing training paths based on evaluation results. The system automatically adjusts the timeframes, spatial scope, and complexity of subsequent training tasks assigned to the trainee based on their process indicator scores and core competency dynamic evaluation results. Specific strategies include: for the role of a mine design engineer, if their innovative design capability score is high, increasing the complexity of innovative design elements in subsequent tasks (e.g., requiring multi-scheme economic comparison) and increasing the weight of innovative improvement indicators in the evaluation; for the role of a mine technical administrator, if their problem-solving ability is outstanding, introducing more complex scenarios such as concurrent equipment failures and severe weather; for the role of a mining regulatory specialist, if their critical thinking and compliance scores are high, strengthening the depth and breadth of safety and compliance reviews in subsequent tasks (e.g., requiring comprehensive reviews from multiple dimensions such as environmental protection and occupational health).
[0059] The evaluation system in this embodiment continuously evaluates and diagnoses the trainees' ability status and mastery level, and dynamically adjusts the difficulty, focus, and pace of the training plan accordingly. It follows the zone of proximal development theory and provides each trainee with a challenge that is within reach, thereby maximizing the training effect.
[0060] In practice, the system sets up a task library, and each task has multi-dimensional attribute tags (such as complexity level, emphasized ability dimension, and spatial region). After each key task node is completed, the system calls the evaluation module to generate the trainee's current ability profile. The personalized recommendation engine matches the ability profile with the attributes in the task library and selects the most suitable next task or group of tasks. When assigning tasks, personalized challenge goals can be attached (such as "Please try to increase the resource recovery rate by 2% for this task").
[0061] This embodiment achieves a leap from standardized to personalized practical training, solving the problem of traditional one-size-fits-all training. It ensures that highly capable trainees are adequately challenged, while less capable trainees keep up. This adaptive mechanism greatly improves training efficiency and individual growth speed, representing the ultimate intelligent manifestation of building a precise talent development loop, and making large-scale personalized education a reality in the field of engineering practical training.
[0062] The specific steps of the mining training method described in this application are as follows: Step 1: Cloud system initialization and personalized character binding At the cloud service layer, a three-dimensional virtual mine model with a unified spatiotemporal reference is constructed based on real case data; Collect career planning questionnaires and cognitive pre-test data from each trainee and input them into the personalized recommendation engine of the core algorithm cluster; Based on the planning and testing results, the engine matches trainees with roles such as mine design engineer, technical administrator, or mining supervision specialist, and binds them to their terminals; Step 2: Synchronizing the spatiotemporal reference and dispatching collaborative tasks The unified time engine sends synchronization instructions to all terminals to ensure that the virtual timeline of each terminal is consistent with the timeline of the cloud model. Based on the student's role and initial ability profile, tasks are extracted from the task library and assigned a unique time stamp by the time engine before being dispatched to the corresponding terminal. Step 3: Real-time terminal interaction and full-process data collection Trainees can perform interactive operations on the 3D model on the terminal in a manner consistent with their roles, such as adjusting design parameters, issuing equipment scheduling instructions, and marking safe areas. The network communication layer uses the WebSocket protocol to upload each operation (including operation content, type, terminal coordinates, and viewpoint) to the cloud in real time. The uploaded data packets are instantly appended with precise spatiotemporal stamps by the unified spatiotemporal engine. The cloud database not only stores the final operation results, but also continuously collects and correlates process data, including: communication logs between roles, the iteration history of design solutions, all exploratory operation sequences in the conflict resolution process, and records of consulting help documents; Step 4: Real-time collaborative verification and conflict visualization early warning The cross-role collaborative verification algorithm in the core algorithm cluster runs continuously, monitoring multi-role operations within the same spatiotemporal region; The algorithm is triggered immediately once it detects that the operation combination violates the preset coordination rules (such as the mining plan not matching the equipment capacity) or safety regulations (such as the slope angle exceeding the standard). The system generates structured early warning information and pushes it to the terminals of all relevant trainees in real time through the communication layer; at the same time, in the 3D model interface, the specific spatial location of the conflict is accurately visualized and marked with highlights and animations to guide trainees in locating the problem. Step 5: Multi-dimensional dynamic quantitative evaluation and capability profile update Process indicator calculation: Based on predefined formulas, the system automatically calculates the scores of each trainee on indicators such as timeliness, quality, collaboration, compliance, and innovation, and weights them according to their role weight template to obtain the process evaluation score. Core competency dynamic assessment: The high-order competency process evaluation module (Bayesian network model) is launched; the model is based on the trainees' initial competency prior probability, and inputs observable evidence (such as innovation and diagnostic accuracy) collected in the current training phase into the network for reasoning; The model outputs the posterior probability distribution of the trainee's four core competencies, which is the dynamically updated competency score. This process is repeated in key stages of training (such as the initial, middle and final stages) to form a dynamically evolving competency profile. Step 6: Adaptive Task Iterative Optimization Based on Evaluation Feedback The personalized recommendation engine receives the latest ability profiles and process performance of trainees from the evaluation module; Based on the principle of consolidating weaknesses and leveraging strengths, the engine dynamically adjusts the attributes of the trainee's subsequent pre-set tasks, including the task's time arrangement (earlier or later), spatial scope (changing to a more complex or more basic area), and core complexity (increasing or decreasing the difficulty of innovation, scheduling, and review). The system customizes in-depth training directions for different roles, such as pushing advanced design tasks containing economic optimization algorithms to promising design engineers; The updated tasks are then assigned again in the second step, initiating a new cycle of training-monitoring-evaluation-optimization until the overall training plan is completed.
[0063] Ultimately, through the closed-loop operation of the above six steps, this method achieves refined and intelligent monitoring and management of the entire mining training process. It not only ensures efficient and safe collaboration among multiple roles in a unified time and space, but also drives the training content to evolve towards personalization and adaptability through continuous tracking and evaluation of process data and core competencies. Ultimately, it achieves the goal of cultivating high-end technical and skilled talents in high-risk industries with low cost, high efficiency, and precision.
[0064] Furthermore, this application also relates to a mining training system, which discloses a system architecture for executing the aforementioned methods. The system adopts a three-layer architecture: a cloud service layer deployed on a cloud server cluster, including a unified spatiotemporal engine, a cloud database, a core algorithm cluster, and a 3D virtual mine model module. The unified spatiotemporal engine is responsible for providing a unified spatiotemporal benchmark and generating spatiotemporal stamps; the core algorithm cluster integrates a personalized recommendation engine, a cross-role collaborative verification algorithm, and a high-order capability process evaluation module. The network communication layer uses the WebSocket protocol to achieve bidirectional real-time communication between the cloud and the terminal with a latency of less than 100ms. The terminal interaction layer includes PCs, tablets, and other terminal devices connected to the system, used for binding roles, displaying 3D models, receiving tasks and warnings, and uploading operational data.
[0065] The system architecture of this application follows the principle of centralized cloud processing and terminal-focused interaction. Computationally intensive model rendering, spatiotemporal computation, and intelligent algorithms are deployed in the cloud, leveraging the elastic computing power of cloud computing to ensure system performance. Optimized communication protocols ensure real-time interaction. The lightweight terminals facilitate deployment and maintenance, support access from multiple device types, and improve the system's accessibility and ease of use.
[0066] The training system presented in this application features a clear layered architecture and decoupled modules, facilitating subsequent functional upgrades and maintenance. The WebSocket protocol ensures smooth collaborative operations. Cloud deployment significantly reduces local hardware investment and maintenance costs for universities and colleges, supports remote training via the internet, and expands application scenarios. This architecture can easily support concurrent training for hundreds of people, meeting the needs of whole-class teaching in universities and large-scale employee training in enterprises, and possesses significant commercial application prospects.
[0067] The foregoing has shown and described the basic principles, main features, and advantages of this application. Those skilled in the art should understand that this application is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this application. Various changes and modifications can be made to this application without departing from the spirit and scope thereof, and all such changes and modifications fall within the scope of this application as claimed. The scope of protection of this application is defined by the appended claims and their equivalents.
Claims
1. A training method for mining operations, characterized in that, include: Construct a three-dimensional virtual spatiotemporal model of the case mine, and configure the time axis and associated spatial region parameters for each stage of the mine's life cycle; Collect trainees' career planning information and assign them roles that match their career plans; Synchronize the spatiotemporal reference of all terminals, and assign collaborative training tasks with unique spatiotemporal stamps to different roles based on trainees' career plans and competency profiles. Trainees interact with the 3D virtual mine model on the terminal. The operation data is generated with a corresponding spatiotemporal stamp and uploaded to the cloud. The cloud collects process data of the entire training process in real time. Based on operation records with time and space stamps and process data, trainees are dynamically and quantitatively evaluated from two dimensions: process indicators and core competencies.
2. The mining training method according to claim 1, characterized in that, The method for constructing a three-dimensional virtual spatiotemporal model of the case mine includes: constructing an interactive three-dimensional spatial model based on real geological data and GIS spatial data of the mine. This model has a unified spatial coordinate system and a standardized time axis. The spatial coordinate system is used to accurately calibrate the geographical coordinates of the mine slope location, mining face, and spoil heap area.
3. The mining training method according to claim 1, characterized in that, The process indicators include timeliness achievement rate, task completion quality, inter-role communication and collaboration, compliance and innovation improvement, and different weights are assigned to each indicator for the three roles of mine design engineer, mine technical administrator and mining supervision specialist.
4. The mining training method according to claim 1, characterized in that, The core competencies include critical thinking, innovative design, teamwork, and problem-solving. The multi-dimensional quantitative evaluation adopts a Bayesian network model, with the four core competencies as hidden nodes and problem diagnosis accuracy, parameter adjustment innovation, communication response efficiency, conflict resolution timeliness, and conflict resolution effectiveness as observable nodes. Through prior probability initialization and posterior probability dynamic updates, the trainee's core competencies are dynamically quantitatively evaluated.
5. A mining training method according to claim 4, characterized in that, The prior probability is calculated by weighting the role's professional competence benchmark requirements with the results of the pre-training cognitive test; the posterior probability is updated based on the observation node data of the corresponding stage in the initial, middle and final stages of the training, so as to dynamically optimize the core competence score.
6. The mining training method according to claim 1, characterized in that, Cross-role collaborative verification is performed on operational data to verify whether the operational content of different roles in the same time and space area meets the preset collaborative rules and safety compliance requirements; if operational conflicts or violations are detected, early warning information is generated and the conflict location is visually marked in the 3D virtual mine model.
7. A mining training method according to claim 6, characterized in that, The cross-role collaborative verification method includes: verifying whether the mining boundary parameters set by the design engineer in the 3D model are coordinated with the production equipment scheduling plan formulated by the technical administrator in the same spatial area and at the same time node; and verifying whether the above parameters and plans meet the safety and compliance constraints proposed by the regulatory specialist based on the preset safety procedures.
8. A mining training method according to claim 1, characterized in that, The process data includes, but is not limited to: records of collaborative communication between roles, records of design parameter modifications, operation trajectories for conflict resolution, and problem diagnosis behavior data. All data are associated with the spatiotemporal stamps of their corresponding operations and stored in a cloud database. The interactive operations include adjusting the parameters of the 3D model, drawing or editing mining design schemes, and adding regulatory inspection markers.
9. A mining training method according to claim 1, characterized in that, Based on the evaluation results, the time nodes, spatial scope, and task complexity of subsequent practical training tasks will be dynamically adjusted; for the role of mine design engineer, the complexity of innovative design elements in the tasks will be increased and their weight in the evaluation will be increased; for the role of mine technical administrator, more complex equipment scheduling and fault handling scenarios will be introduced; for the role of mining supervision specialist, the depth and breadth of safety compliance review will be strengthened.
10. A mining training system, characterized in that, A method for performing mining training as described in any one of claims 1 to 9, comprising: The cloud service layer, deployed on a cloud server cluster, includes a unified spatiotemporal engine, a cloud database, a core algorithm cluster, and a 3D virtual mine model module. The unified spatiotemporal engine provides a unified spatial coordinate system and a standardized time axis, and generates corresponding spatiotemporal stamps for all user operations. The core algorithm cluster includes a personalized recommendation engine, a cross-role collaborative verification algorithm, and a high-order capability process evaluation module. The network communication layer uses the WebSocket protocol to achieve low-latency bidirectional communication between the cloud service layer and the terminal. The terminal interaction layer includes multiple terminal devices used to bind different professional roles, receive training tasks, upload operation data, and display a 3D virtual mine model, collaborative conflict prompts, and dynamic evaluation results.