Method and system for realizing practical training teaching accurate management and control and data-driven evaluation

Abstract

The invention relates to the technical field of teaching management, in particular to a method and a system for realizing accurate management and control and data-driven evaluation of practical training teaching, which enforce students to operate according to a standard process from the system level through a step state machine and role interlocking technical mechanism, and improve the practical training teaching accuracy. According to the technical scheme, traceable and leapfrogging-prevention refined management and control of the whole practical training process are achieved, boring single-person practice is converted into team tasks with mutual supervision and cooperation significance through an operation-inspection team cooperation mode, interestingness of the practical training process is enhanced through gamification design such as integral, ranking and value-added reward, and the practical training effect is improved. The learning motivation of students is effectively stimulated, and through the technical means of introducing a training admission rule, the students are forced from the system level to have to complete sufficient and up-to-standard training to obtain the examination qualification. The problem that training behaviors of students cannot be effectively restrained in traditional teaching is solved, and the base line of basic teaching quality is ensured.

Description

Technical Field

[0001] This invention relates to the field of teaching management technology, and more specifically, to a method and system for achieving precise control and data-driven evaluation of practical training. Background Technology

[0002] Currently, the following technical bottlenecks commonly exist in vocational education practical training (such as aircraft maintenance, precision manufacturing, and other fields): 1. Lack of process monitoring: Existing practical training management systems or online examination platforms mostly focus on theoretical assessment and result recording, failing to monitor and collect data on students' practical operations in real time and step by step. Teachers cannot trace the complete operation chain of students, resulting in a lack of process evaluation.

[0003] 2. High subjectivity in evaluation: The evaluation of operational standards, professional qualities and other dimensions relies heavily on teachers' subjective experience and on-site observation, lacks objective data support, and makes it difficult to ensure the fairness and consistency of the evaluation.

[0004] 3. Low teaching efficiency: Teachers need to circulate and guide a large number of students, resulting in an extremely heavy workload. It is also difficult for them to pay attention to the operational details of all students at the same time, leading to insufficient coverage of teaching guidance and inconsistent teaching results.

[0005] 4. Insufficient student motivation and low participation: Traditional individual practical training is monotonous, and students easily underestimate the training because they feel it is "simple," lacking intrinsic motivation to practice actively. Research found that in practical training classes without effective interaction, nearly half of the students were inattentive (e.g., playing on their phones), failing to practice effectively, resulting in frequent errors and weak skill mastery during assessments.

[0006] 5. Lack of access control in the training process: The existing system fails to establish a mandatory technical link between training and assessment. Students can skip necessary training steps and directly enter the assessment, leading to assessment failure due to insufficient training, thus failing to guarantee the basic quality of teaching at the system level.

[0007] 6. Rigid system and poor scalability: Existing technical solutions are mostly closed systems customized for specific courses, which cannot support teachers of different disciplines and courses to flexibly edit training processes (work cards) according to their own teaching content. The system has weak universality and scalability.

[0008] Therefore, there is an urgent need for a technological solution that can digitally, standardize, and intelligently manage the entire training process, stimulate students' enthusiasm for participation, and has broad applicability. Summary of the Invention

[0009] In view of the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for achieving precise control and data-driven evaluation of practical training.

[0010] To achieve the above objectives, the present invention provides the following technical solution: A method for achieving precise control and data-driven evaluation of practical training includes the following steps: Teachers can edit new content and publish practical training tasks using configurable work card templates; and set training mode access rules for practical training tasks. In training and assessment modes, operators and inspectors are teamed up, and interfaces with different permissions are assigned to each of them. The operator first enters training mode to practice, and the system records their training data. The system verifies whether the student's training data meets the admission rules, and automatically unlocks the assessment mode permission for the training task if it does. Based on the step state machine, a mandatory interlocking of the operation process between the student operation terminal and the student inspection terminal is implemented. The system synchronously collects operator step data, inspector inspection data, smart device data, and teacher evaluation data; The authenticity of the collected data is verified through an anti-fraud model; Based on a weighted model that integrates multi-source data, comprehensive evaluation results are automatically generated and presented visually.

[0011] Furthermore, a system for achieving precise control and data-driven evaluation of practical training includes: Teacher management terminal: used for work card editing, task posting, class management, roving scoring, and data dashboard viewing; Student operation terminal and student inspection terminal: respectively used by operators to execute work card steps and by inspectors to perform inspection scoring and signing, and view data on the dashboard. The interfaces and permissions of the two are isolated, and the process interlocking is achieved through the step state machine. Workcard Configuration and Task Scheduling Module: As the task definition and allocation hub of the system, it provides structured and editable workcard templates, receives task creation and configuration instructions from the teacher management terminal, publishes and schedules tasks according to the preset class-project mapping relationship, and manages the lifecycle of task status. Training access control module: It has a built-in rule configuration unit and a real-time verification unit. The training access control module is used to set and store the minimum training number threshold and the maximum score threshold of the training mode for each task. When a student requests to enter the assessment, it compares his / her historical training data with the rule threshold in real time and dynamically authorizes or prohibits his / her assessment permission. Role Interlocking and Process Control Module: This module is responsible for assigning operator or inspector role permissions to users and maintaining a process control model based on a step state machine. This module enforces the sequential execution of operation steps and the signing interlock between roles through technical mechanisms, ensuring that subsequent steps cannot be executed if the previous step has not been confirmed by the inspector. Multi-source data acquisition and storage module: It is responsible for asynchronously acquiring and standardizing operation sequence, test results, equipment readings, and teacher evaluations from multiple heterogeneous data sources such as student operation terminals, student testing terminals, teacher management terminals, and external intelligent testing equipment, and storing them in the database. At the same time, it establishes correlation indexes between data for traceability analysis. Data fusion and analysis engine: Built-in anti-cheating model and intelligent evaluation algorithm; Layered visualization and spatial scheduling module: Based on the data model, it dynamically generates three levels of data dashboard views: overall class, zoned project tasks, and individual details. According to the preset mapping strategy between logical dashboards and physical display devices, it pushes different dashboard content to the corresponding display devices in the training room via the network and renders them.

[0012] Furthermore, verification of the training mode access rules is a necessary condition for students to enter the assessment mode; those who do not meet the requirements cannot start the assessment task.

[0013] Furthermore, the anti-fraud model verifies data authenticity by analyzing at least one of the following: the rationality of the operation sequence, the consistency of data logic among multiple roles, and the matching degree between device data and human data.

[0014] Further detailed steps for building an anti-cheating model: Step S501: Feature vector definition and extraction; Define the feature vector of a single operation step for: ; Step S502: Anomaly scoring based on Mahalanobis distance; To overcome the influence of different feature dimensions and correlations, Mahalanobis distance is used as the comprehensive anomaly score. ; First, calculate the mean vector of the normal feature vector set from historical normal operation data. Covariance Matrix ; For a new feature vector Its abnormality score is: ; Step S503: Dynamic threshold determination and data labeling; The decision rule function can be expressed as: ; Marked as Data will have points deducted from the total score according to preset rules, or will be directly excluded from the statistics of valid training sessions.

[0015] Furthermore, The standard time taken for this step is the actual time taken for that step. Compared with the historical standard time for this step The ratio, i.e. ; Step interval anomaly: Calculates the time interval between the start time of the current step and the end time of the previous step. and apply Normalize the function: ,in Sensitivity coefficient; The deviation is measured by the absolute error between the inspector's deduction value (Sstudent) and the teacher's pre-set typical error deduction value (Satandard) for this step, normalized to the [0,1] interval. ; : Equipment data consistency flag. If the quantitative data returned by the intelligent detection equipment is consistent with the result of the operator's manual selection or judgment, it is 1; otherwise, it is 0.

[0016] Furthermore, the detailed steps of the intelligent evaluation algorithm execution are as follows: Step S601: Formal definition of value-added events; Each value-added event Define a triple as follows: ; Step S602: Event scanning and execution engine; The system maintains a value-added event library. ; When a certain When an event occurs, the algorithm uses the relevant student chronology event database. Execute each event function; Step S603: Summarize reward points; student The total bonus points during this trigger period are the sum of the reward points for all triggered events: ; The brackets [·] represent Iverson brackets, with a value of 1 if the condition is true and 0 otherwise.

[0017] Furthermore, Triggering timing; : Boolean trigger condition function, returns or ; : Reward points.

[0018] Compared with the prior art, the present invention has the following beneficial effects: Technical Effects - Refined Process Control: Through the technical mechanism of step state machine and role interlock, students are forced to operate according to standard procedures at the system level, realizing refined control of the entire training process with traceability and prevention of skipping steps.

[0019] Technical Effects - Objectification and Efficiency Improvement in Evaluation: Through the automatic collection and weighted fusion of multi-source data (operation, inspection, equipment, teachers), an objective and comprehensive evaluation system was constructed, which significantly reduced the subjective judgment load of teachers and expanded the coverage of evaluation dimensions from a single result to the whole process and multiple dimensions.

[0020] Technical Effects - Enhanced Learning Participation and Motivation: Through a team collaboration model of operation and verification, the tedious individual practice is transformed into team tasks with mutual supervision and cooperation. Gamification design, such as points, rankings, and value-added rewards, enhances the fun of the training process and effectively stimulates students' learning motivation.

[0021] Technological Effectiveness – Ensuring Basic Teaching Quality: By introducing technical means of training admission rules, students are required at the system level to complete sufficient and qualified training before they can qualify for assessment. This addresses the pain point of traditional teaching methods, which cannot effectively regulate student training behavior, and ensures a basic standard of teaching quality.

[0022] Technical Advantages - High System Versatility and Promotability: The core of this invention lies in providing a universal training management mode that supports configurable work cards, rather than software with fixed content. Teachers of any discipline can use the system's work card editing function to quickly create training content that meets the needs of their course and immediately apply all the advanced functions such as role interlocking, data acquisition, and intelligent analysis. This gives the invention strong promotional value across different professional fields and solves the pain point of the difficulty in promoting dedicated systems. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the structure of a method and system for achieving precise control and data-driven evaluation in practical training; Figure 2 A flowchart illustrating the construction process of a dynamic evaluation model based on fuzzy logic and capability vectors; Figure 3 A flowchart illustrating the construction process of an optimization model based on constraint satisfaction and critical path. Detailed Implementation

[0024] Example 1: Refer to Figure 1 A method for achieving precise control and data-driven evaluation of practical training includes the following steps: S1: Task and training admission rule configuration; Teachers edit and create training tasks containing multiple ordered operation steps based on configurable work card templates on the teacher management terminal, and specify the class to be executed; At the same time, training mode admission rules are set for the task, which include at least: minimum training number threshold and maximum score threshold in the training mode. S2: Access control for training and assessment modes; after receiving a task on the student's terminal, the student first enters training mode. In training mode, the student can practice according to the procedures in S3 to S5, but the system records the number of training sessions and the score for each session.

[0025] The system has a built-in training achievement verification module that continuously compares the student's training records with the admission rules. Only when the system verifies that the student has completed the minimum number of training sessions for this task and that the highest score achieved in the training mode has reached the highest score threshold, will the system automatically unlock the student's assessment mode operation privileges.

[0026] S3: Role Binding and Process Interlocking; This step applies to both training and assessment modes, but only the score generated in assessment mode is the final valid assessment score. After the student's operating terminal receives the task, the first student (operator) selects the second student (inspector) to form a team; the system assigns different operating permissions and interface views to the operator and inspector; the operation process of the student operating terminal and the student inspection terminal is forcibly interlocked based on the step state machine, that is: after the operator completes the current step and electronically signs, the system sets the status of the step to pending inspection and unlocks the corresponding signing interface of the student inspection terminal; after the inspector completes the inspection and signing, the system sets the status of the step to completed and unlocks the operator's next step operation interface; and so on, until the process ends.

[0027] S4: Multi-source data acquisition and synchronization; During the execution of the steps, the system synchronously collects and associates the following data: the operator's step operation sequence data and electronic signature data; the inspector's inspection result data (pass / fail) and deduction data for each step; the measurement data automatically uploaded by integrated intelligent testing equipment (such as digital calipers) via wireless transmission; and the scoring and evaluation data entered by teachers through the teacher management terminal during their rounds of guidance.

[0028] S5: Data Authenticity Verification and Intelligent Analysis; The system backend uses the following algorithm models to verify and analyze the validity of the collected data: Anti-cheating model: By analyzing the deviation between the time distribution of operation steps and the standard time, the logical rationality of the timestamps signed by the operator and inspector, and the consistency between smart device data and manually entered data, the system determines the authenticity of the data and marks or deducts points for abnormal data. Intelligent value-added evaluation algorithm: The system automatically performs sorting and threshold judgment, automatically assigning bonus points to situations such as entering the top N in the class for the first time, having a high accuracy rate in deducting points as an inspector, and being among the top ten in task completion speed.

[0029] S6: Multi-level data visualization and real-time ranking incentives; The system automatically integrates theoretical assessment scores (30%), practical training operation scores (50%, derived from step scores and inspector deductions), inspector role performance scores (20%), and intelligent value-added scores (bonus points) according to a preset weight model to generate a student's comprehensive score; and displays information such as class average score, individual ranking, error type distribution, step time curve, and three-dimensional capability radar chart on the visualization dashboards of the teacher management terminal and the student operation terminal.

[0030] The system generates and displays multi-level visual data dashboards, including at least: a class-wide data dashboard: displaying the class's overall score, typical error types, task progress, and information on the top N (e.g., top six) students in the class. A project data dashboard: displaying the class's mastery data for a specific project, the average time and score for each step, and information on the top N (e.g., top six) students in the project. An individual data dashboard: displaying each student's individual ranking, scores for each task, comparison of step times, error types, and a 3D ability radar chart. The ranking data in the class and regional project dashboards is automatically calculated and refreshed by the system backend based on real-time updated student performance data using a dynamic sorting algorithm, thereby creating a competitive and supportive learning atmosphere within the class.

[0031] Example 2: Refer to Figures 2-3 A system for achieving precise control and data-driven evaluation of practical training, comprising: Teacher Management Terminal: Used for work card editing, task posting, class management, roving scoring, and data dashboard viewing.

[0032] Student operation terminal and student inspection terminal: respectively used by operators to execute work card steps and by inspectors to perform inspection scoring and signing, view data dashboards, etc. The interfaces and permissions of the two are isolated, and the process is interlocked through the step state machine.

[0033] Workcard Configuration and Task Scheduling Module: As the central hub for task definition and allocation in the system, it provides structured and editable workcard templates, receives task creation and configuration instructions from the teacher management terminal, publishes and schedules tasks according to the preset class-project mapping relationship, and manages the lifecycle of task status.

[0034] Work card configuration and task scheduling module: based on an optimization model of constraint satisfaction and critical path; The core technical purpose of this module is to solve the technical problems of complex step dependencies, frequent resource conflicts, and low scheduling efficiency in the arrangement of practical training tasks.

[0035] Detailed steps for constructing an optimization model based on constraint satisfaction and critical path: Step S101: Model the directed graph of the work card; The work card is abstracted as a directed acyclic graph (DAG) G=(V,E).

[0036] The node v_i∈V represents the operation steps, and its attributes include: std_time (standard time), tools (required toolset), prerequisites (prerequisites), and error_set (preset error and deduction set).

[0037] The edge e_ij=(v_i,v_j)∈E represents the order dependency from step v_i to v_j.

[0038] Step S102: Formalize the constraint satisfaction problem (CSP); Define the work card configuration as a constraint satisfaction problem P=(X,D,C).

[0039] X = {x_1, x_2, ..., x_n} is a set of editable attribute variables for each step.

[0040] D is the range of the variable, such as std_time>θ.

[0041] C is the constraint set, which includes: Order constraint: ∀e_ij∈E, v_i must be completed before v_j.

[0042] Resource constraint: ∀t,Σ[v_i uses tool r at time t]≤R_r, where R_r is the number of tools r.

[0043] Time constraint: Σstd_time(v_i)≤T_max (total duration limit).

[0044] Step S103: Critical path scheduling and resource allocation; The Critical Path Method (CPM) is used to calculate the earliest start time (ES), latest start time (LS), and float time.

[0045] ES(v_j) = max{ES(v_i) + std_time(v_i) | v_i is the predecessor of v_j} LS(v_i) = min{LS(v_j) - std_time(v_i) | v_j is the successor of v_i} The sequence of steps on the critical path, ES(v_i) = LS(v_i), determines the shortest completion time of the work card.

[0046] Based on the critical path, mixed-integer linear programming (MILP) is used for scheduling optimization under resource constraints: Objective function: Minimize the maximum completion time (Makespan).

[0047] Decision variable: x[s,t,i]∈{θ,1}, representing whether student s performs step i at time t.

[0048] Constraints: These include sequence constraints, resource constraints, and the requirement that each person executes each step only once.

[0049] 1.2 Related Functions and Algorithms: Cycle detection function: HasCycle(G)->Boolean, uses depth-first search (DFS) to detect whether there are cyclic dependencies in the graph.

[0050] The critical path calculation function is: CalculateCriticalPath(G)->(Path,Duration), which returns the critical path and its total duration.

[0051] MILP objective function: MinimizeC_max, where C_max≥completion_time(s)∀s∈Students.

[0052] Training Access Control Module: Acting as the gatekeeper for assessment permissions, this module includes a built-in rule configuration unit and a real-time verification unit. It sets and stores the minimum training iteration threshold and the maximum score threshold for each training mode for each task. When a student requests access to the assessment, it dynamically authorizes or denies their assessment access by comparing their historical training data with the rule thresholds in real time.

[0053] Training admission control module: a dynamic evaluation model based on fuzzy logic and capability vectors.

[0054] The core technical purpose of this module is to solve the technical problems of the rigidity of traditional admission mechanisms and their inability to comprehensively evaluate students' training quality and preparation status.

[0055] Detailed steps for building a dynamic evaluation model: Step S201: Student ability vector modeling; Construct a dynamic capability vector for student s for task t: C_s^t=[k,score_max,consistency,t_avg] k: Number of training sessions completed.

[0056] score_max: The highest score in the history of training.

[0057] consistency: operational stability, calculated as 1 - (σ_scores / μ_scores).

[0058] t_avg: Average completion time.

[0059] Step S202: Construction of the fuzzy logic system; Define a fuzzy logic system with inputs being the dimensions of a capability vector and outputs fuzzy values ​​for "admission qualifications".

[0060] Fuzzification: Define a membership function for each input (such as a triangular membership function for "few / medium / many training times").

[0061] Rule base: Create fuzzy rules, for example: IF(kisHigh)AND(score_maxisHigh)AND(consistencyisHigh)THEN(EligibilityisHigh) IF(kisLow)OR(score_maxisLow)OR(consistencyisLow)THEN(EligibilityisReject) Defuzzification: The centroid method is used to convert the output fuzzy set into an accurate admission score E_score.

[0062] Step S203: Dynamic decision function; The final decision is determined by the threshold function: GrantAccess(s,t)=1ifE_score(s,t)>=θ_eelseθ Where θ_e is the dynamic admission threshold, which can be fine-tuned according to the overall level of the class.

[0063] 2.2 Related Functions and Algorithms: The stability calculation function, CalculateConsistency(score_list)->Float, calculates the reciprocal of the coefficient of variation of the score list.

[0064] The fuzzy inference function, FuzzyInference(training_count,max_score,consistency)->E_score, performs fuzzy rule calculations.

[0065] Membership functions: μ_low(x), μ_medium(x), μ_high(x), define the membership degree of each language variable.

[0066] Role Interlocking and Process Control Module: As the execution engine of the system's business logic, this module is responsible for assigning operator or inspector role permissions to users and maintaining a process control model based on a step state machine (e.g., not started, pending operation, pending inspection, completed). This module enforces the sequential execution of operation steps and interlocking of signatures between roles through technical mechanisms, ensuring that subsequent steps cannot be executed unless the previous step is confirmed by the inspector.

[0067] Role interlocking and flow control module: a formal verification model based on time-based Petri nets; The core technical purpose of this module is to solve the technical problems of confused role permissions, operation step jumps, and untraceable process in the training process.

[0068] Detailed steps for constructing a formal verification model; Step S301: Time-based Petri net modeling; The work card execution process is formally modeled using a Timed Petri Net: PN=(P,T,F,M_θ,D).

[0069] P: Place set, representing the operation status (such as "pending operation", "pending inspection", "completed").

[0070] T: A set of transitions, representing events that trigger a change in state (such as "operator signature" or "inspector confirmation").

[0071] F: A set of arcs (Flows) that define the direction of state flow.

[0072] M_θ: Initial Marking, indicating the start state of the process.

[0073] D:T->R+: Delay function, which associates a time (such as the time required for operation / test) with each transition.

[0074] Step S302: Implementation of interlock logic; Interlocking is implemented using a Petri net structure. The key design is that the "check complete" transition of one step is a prerequisite for the "operation start" transition of the next step.

[0075] The system state S is uniquely determined by the current identifier M. Any user action (clicking a button) triggers a transition t if and only if t is enabled under M.

[0076] Enable condition: ∀p∈•t:M(p)≥W(p,t), where•t is the set of input libraries of t, and W is the weight function.

[0077] Step S303: State transition and guard conditions; When the enabled transition t is triggered, the system state transitions from M to M': ∀p∈P:M'(p)=M(p)-W(p,t)+W(t,p) Guard conditions: Add pre-guard functions to critical transitions (such as signing, verification), such as: operator_permission_guard(user_context)->Boolean signature_valid_guard(signature_data)->Boolean 3.2 Related Functions and Algorithms: Transition enable function: IsEnabled(t,M)->Boolean, checks whether transition t is enabled under label M.

[0078] State transition function: FireTransition(t,M)->M', executes the enabled transition and updates the network state.

[0079] Guard functions: operator_guard, inspector_guard, signature_guard, verify user permissions and the legitimacy of operations.

[0080] Multi-source data acquisition and storage module: As the system's data hub, it is responsible for asynchronously acquiring and standardizing process data such as operation sequence, test results, equipment readings, and teacher evaluations from multiple heterogeneous data sources, including student operation terminals, student testing terminals, teacher management terminals, and external intelligent testing equipment. It also persists the data to the database and establishes correlation indexes between the data for traceability analysis.

[0081] Multi-source data acquisition and storage module: a stream processing architecture based on a unified spatiotemporal data model; The core technical objective of this module is to solve the technical problems of inconsistent formats, asynchronous acquisition, scattered storage, and difficulty in correlation analysis of multi-source heterogeneous training data.

[0082] Detailed steps for building a stream processing architecture based on a unified spatiotemporal data model: Step S401: Define a unified spatiotemporal data model; Define a unified data unit (DataUnit) structure: DU=(s_id,t_id,step_id,timestamp,location,data_type,payload) s_id, t_id, step_id: Student, task, and step identifiers, respectively.

[0083] timestamp: A high-precision timestamp that ensures timeliness.

[0084] location: The location of the data source (such as workstation ID, device ID).

[0085] data_type: An enumerated type, such as OPERATOR_ACTION, INSPECTOR_JUDGEMENT, DEVICE_READING, TEACHER_SCORE.

[0086] payload: A JSON-formatted payload that stores the actual data.

[0087] Step S402: Time-series database storage and index optimization; All data units are stored in the time-series database using (s_id, timestamp) as the primary key.

[0088] Inverted indexes are created for high-frequency query conditions such as (t_id, step_id) and data_type to achieve sub-second multidimensional retrieval.

[0089] Columnar storage is used to optimize analytical query performance.

[0090] Step S403: Stream processing and real-time aggregation; A stream processing model is used to consume data streams in real time.

[0091] Define a sliding time window W, within which micro-batch processing or real-time aggregation of data is performed: Aggregated_Data={count,avg_duration,error_frequency,...} It provides real-time input to the upper-level analysis engine, supporting instant feedback and alerts.

[0092] 4.2 Related Functions and Algorithms: The data serialization function, SerializeData(DU)->ByteStream, serializes a data unit into a byte stream.

[0093] The window aggregation function, AggregateWindow(data_stream,W)->Aggregated_Data, performs time window aggregation on the data stream.

[0094] Index query functions: QueryByTimeRange(start,end)->[DU], QueryByStep(step_id)->[DU].

[0095] Data Fusion and Analysis Engine: Serving as the system's intelligent brain, this engine incorporates an anti-cheating model and intelligent evaluation algorithm. It cleans, fuses, and weights the collected raw data, including: The anti-fraud model identifies and filters abnormal data by analyzing features such as the rationality of the operation sequence and the consistency of data logic among multiple roles. Anti-cheating model: based on multi-dimensional temporal anomaly detection; The core technical purpose of this model is to solve the technical problem that existing systems cannot effectively distinguish the authenticity of training data. By using a computational model, it can automatically identify and mark false or invalid operational data, thereby ensuring the reliability of the data source for subsequent analysis.

[0096] Detailed steps for building an anti-cheating model: Step S501: Feature vector definition and extraction; Define the feature vector of a single operation step for: · Standard time ratio for each step. Actual time taken for this step. Compared with the historical standard time for this step The ratio, i.e. This ratio can eliminate the impact of inherent differences in difficulty between different steps.

[0097] · Step interval anomaly. Calculate the time interval between the start time of the current step and the end time of the previous step. and apply Normalize the function: in The sensitivity coefficient (can be set to 0.1) indicates that too short an interval may mean malicious practice, while too long an interval may mean a lack of concentration.

[0098] · : Inspection deviation. Inspector deduction value (Sstudent) and teacher's pre-set typical error deduction value for this step in the backend. The absolute error of Satandard, normalized to the [0,1] interval: .

[0099] · : Equipment data consistency flag. If the quantitative data returned by the intelligent detection equipment is consistent with the result manually selected or judged by the operator, it is 1; otherwise, it is 0.

[0100] • Step S502: Anomaly scoring based on Mahalanobis distance; To overcome the influence of different characteristic dimensions and correlations, Mahalanobis distance is used. ) as a comprehensive anomaly score .

[0101] First, calculate the mean vector of the normal feature vector set from massive historical normal operation data. Covariance Matrix .

[0102] For a new feature vector Its abnormality score is: Mahalanobis distance measures the current operational characteristics. The distance from the center point of the normal operating group and the covariance structure of the data are taken into account, making it more statistically accurate.

[0103] Step S503: Dynamic threshold determination and data labeling; Abnormal threshold It is not a fixed value, but rather dynamically adjusted based on the overall operational proficiency of the current class. The initial value is set to historical normal data. Distribution Quantiles.

[0104] The decision rule function can be expressed as: ; Marked as Data will have a certain number of points deducted from the total score according to preset rules (for example, the highest score for this operation will not exceed 60 points), or will be directly excluded from the statistics of valid training times.

[0105] The intelligent evaluation algorithm automatically performs value-added scoring and ranking calculations according to preset rules, and finally generates a comprehensive evaluation result.

[0106] Intelligent evaluation algorithm: Automatic value-added scoring based on event triggers; The core technical purpose of this algorithm is to solve the problem that existing teaching systems have a single evaluation dimension and cannot automatically identify and motivate students' personalized outstanding performance.

[0107] Detailed steps of the event-triggered automatic value-added scoring method: • Step S601: Formal definition of value-added events; Each value-added event Define a triple as follows: .

[0108] · Triggering timing. For example: after each task is completed, or at the end of daily training.

[0109] · : Boolean trigger condition function. Returns or .

[0110] · : Reward points.

[0111] Specific event examples: · Rank Breakthrough Award · After the results of a single task are released.

[0112] · .in Rank the classes The threshold for the leap (e.g., 5 places).

[0113] +2 points.

[0114] · Precision Inspector Award · After the results of a single task are released.

[0115] · • This section uses The similarity coefficient calculates the degree of overlap between student and teacher deduction items, which is better than the simple overlap rate and can penalize students for both missed and false detections.

[0116] · and These are the sets of deductions recorded for students and teachers, respectively.

[0117] · The similarity threshold is set (e.g., 0.7).

[0118] · +3 points.

[0119] Step S602: Event scanning and execution engine; The system maintains a value-added event library. .

[0120] When a certain When an event occurs, the algorithm uses the relevant student chronology event database. Execute each event function.

[0121] Step S603: Summarize reward points; student The total bonus points during this trigger period are the sum of the reward points for all triggered events: The brackets [·] represent Iverson brackets, with a value of 1 if the condition is true and 0 otherwise.

[0122] Data fusion engine: A two-layer weighted fusion model based on AHP-entropy weighting method; The core technical purpose of this engine is to solve the technical problem that the weight allocation of multi-source heterogeneous data (theory, practice, verification) is highly subjective and difficult to quantify scientifically.

[0123] Detailed steps for constructing a two-layer weighted fusion model; Step S701: Determining Subjective Weights (AHP Layer) The Analytic Hierarchy Process (AHP) is used to integrate the experience of teaching experts.

[0124] Construct judgment matrix A (taking theoretical, practical, and verification criteria as an example): · Matrix elements Representation Criteria Relative to criteria Importance (scale 1-9).

[0125] Calculate the eigenvectors of the matrix and perform a consistency test (CR < 0.1). The resulting eigenvectors are... This refers to subjective weighting.

[0126] Step S702: Determining Objective Weights (Entropy Weight Method) The objective weights are determined based on the distribution of all students' grades in this task using the EntropyWeightMethod.

[0127] It has One student, Evaluation indicators (theory, practical operation, inspection). First, standardize the original score matrix to obtain .

[0128] Calculate the entropy value of the th indicator : Where , .

[0129] Calculate the difference coefficient of the th indicator , then the objective weight is: · Step S703: Combined weight calculation and score fusion Linearly combine the subjective and objective weights to obtain the final combined weight: Where is the preference coefficient, , usually 0.5 can be taken to show balance.

[0130] The final score fusion function is: This model dynamically adjusts the weights, respecting both the teaching rules (AHP) and responding to the data characteristics of the current task (entropy weight method), and is more scientific and self-adaptive than the fixed weight model.

[0131] Hierarchical visualization and space scheduling module: As the interaction interface between the system and the physical environment, it includes a kanban generation unit and a space scheduling unit. This module dynamically generates data kanban views at three levels: the overall class, partition project tasks, and personal details according to the data model, and based on the mapping strategy of the preset logical kanban and physical display devices, pushes and renders different kanban contents to the corresponding display devices in the training room (such as public large screens, area display screens, personal terminals) through the network, realizing the deep integration and linkage of data presentation and the physical teaching space.

[0132] Hierarchical visualization and space scheduling module: An optimization model based on visual coding and bipartite graph matching; The core technical purpose of this module is to solve the technical problems of chaotic data display content on multiple screens, unreasonable space layout, and high information cognitive load.

[0133] Detailed steps for constructing an optimization model based on visual coding and bipartite graph matching: Step S801: Visual coding and mapping model; Define the visual channel set Visual_Channels={Position,Length,Angle,Color_Hue,Color_Saturation,...}.

[0134] Based on data type (categorized, ordered, ordinal) and task type (overall trend, anomaly detection, detailed query), data attributes are mapped to the most effective visual channels through predefined mapping rules.

[0135] For example: Map_Rule: Class Ranking -> Position (Y-axis) Map_Rule: Task Score -> Color_Hue (Green-to-Redgradient) Map_Rule: Error frequency -> Size(Circleradius) Step S802: Space scheduling optimization model; The relationship between display devices and dashboard content is modeled as a bipartite graph matching problem.

[0136] Define a cost function Cost(Screen_j, Dashboard_i), and consider: Information importance Imp_i (the higher the importance, the lower the cost).

[0137] Spatial correlation Geo(Screen_j, Dashboard_i) (physical distance between the device and the relevant content area).

[0138] Cognitive load Cog(Screen_j,User) (avoiding information overload on a single screen).

[0139] The goal is to minimize the total cost ΣCost(Screen_j, Dashboard_i) through allocation. This weighted bipartite graph minimum weight matching problem is solved using the Kuhn-Munkres (Hungarian) algorithm.

[0140] Step S803: Real-time rendering and push mechanism; A publish-subscribe (Pub-Sub) model is established based on the WebSocket protocol.

[0141] Each physical screen acts as a subscriber, subscribing to a specific type of data stream.

[0142] When the backend data is updated, the new visualization data packet is pushed in real time to all clients that have subscribed to the data stream via WebSocket.send(JSON_Data).

[0143] 8.2 Related Functions and Algorithms; Visual mapping function: MapDataToVisual(data_attribute,data_type)->visual_channel.

[0144] The Hungarian algorithm matching function, Hungarian(cost_matrix)->assignment, solves for the optimal screen-kanban allocation.

[0145] Cost calculation function: CalculateCost(screen,dashboard)->Float, comprehensively calculates space, cognitive, and technological costs.

[0146] The above formulas are all dimensionless calculations, and the preset parameters in the formulas should be set by those skilled in the art according to the actual situation.

[0147] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0148] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0149] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0150] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0151] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0152] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0153] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for achieving precise control and data-driven evaluation of practical training, characterized in that, Includes the following steps: Teachers can edit new content and publish practical training tasks using configurable work card templates; and set training mode access rules for practical training tasks. In training and assessment modes, operators and inspectors are teamed up, and interfaces with different permissions are assigned to each of them. The operator first enters training mode to practice, and the system records their training data. The system verifies whether the student's training data meets the admission rules, and automatically unlocks the assessment mode permission for the training task if it does. Based on the step state machine, a mandatory interlocking of the operation process between the student operation terminal and the student inspection terminal is implemented. The system synchronously collects operator step data, inspector inspection data, smart device data, and teacher evaluation data; The authenticity of the collected data is verified through an anti-fraud model; Based on a weighted model that integrates multi-source data, comprehensive evaluation results are automatically generated and presented visually.

2. A system for achieving precise control and data-driven evaluation of practical training, applied to the method for achieving precise control and data-driven evaluation of practical training as described in claim 1, characterized in that, include: Teacher management terminal: used for work card editing, task posting, class management, roving scoring, and data dashboard viewing; Student operation terminal and student inspection terminal: respectively used by operators to execute work card steps and by inspectors to perform inspection scoring and signing, and view data on the dashboard. The interfaces and permissions of the two are isolated, and the process interlocking is achieved through the step state machine. Workcard Configuration and Task Scheduling Module: As the task definition and allocation hub of the system, it provides structured and editable workcard templates, receives task creation and configuration instructions from the teacher management terminal, publishes and schedules tasks according to the preset class-project mapping relationship, and manages the lifecycle of task status. Training access control module: It has a built-in rule configuration unit and a real-time verification unit. The training access control module is used to set and store the minimum training number threshold and the maximum score threshold of the training mode for each task. When a student requests to enter the assessment, it compares his / her historical training data with the rule threshold in real time and dynamically authorizes or prohibits his / her assessment permission. Role Interlocking and Process Control Module: This module is responsible for assigning operator or inspector role permissions to users and maintaining a process control model based on a step state machine. This module enforces the sequential execution of operation steps and the signing interlock between roles through technical mechanisms, ensuring that subsequent steps cannot be executed if the previous step has not been confirmed by the inspector. Multi-source data acquisition and storage module: It is responsible for asynchronously acquiring and standardizing operation sequence, test results, equipment readings, and teacher evaluations from multiple heterogeneous data sources such as student operation terminals, student testing terminals, teacher management terminals, and external intelligent testing equipment, and storing them in the database. At the same time, it establishes correlation indexes between data for traceability analysis. Data fusion and analysis engine: Built-in anti-cheating model and intelligent evaluation algorithm; Layered visualization and spatial scheduling module: Based on the data model, it dynamically generates three levels of data dashboard views: overall class, zoned project tasks, and individual details. According to the preset mapping strategy between logical dashboards and physical display devices, it pushes different dashboard content to the corresponding display devices in the training room via the network and renders them.

3. The system for achieving precise control and data-driven evaluation of practical training teaching according to claim 2, characterized in that, Verification of the training mode access rules is a necessary condition for students to enter the assessment mode; those who do not meet the requirements cannot start the assessment task.

4. The system for achieving precise control and data-driven evaluation of practical training teaching according to claim 2, characterized in that, The anti-fraud model verifies data authenticity by analyzing at least one of the following: the rationality of the operation sequence, the consistency of data logic among multiple roles, and the matching degree between device data and human data.

5. The system for achieving precise control and data-driven evaluation of practical training teaching according to claim 2, characterized in that, Detailed steps for building an anti-cheating model: Step S501: Feature vector definition and extraction; Define the feature vector of a single operation step for: ； Step S502: Anomaly scoring based on Mahalanobis distance; To overcome the influence of different feature dimensions and correlations, Mahalanobis distance is used as the comprehensive anomaly score. ; First, calculate the mean vector of the normal feature vector set from historical normal operation data. Covariance Matrix ; For a new feature vector Its abnormality score is: ； Step S503: Dynamic threshold determination and data labeling; The decision rule function can be expressed as: ； Marked as Data will have points deducted from the total score according to preset rules, or will be directly excluded from the statistics of valid training sessions.

6. The system for achieving precise control and data-driven evaluation of practical training teaching according to claim 5, characterized in that, The standard time taken for this step is the actual time taken for that step. Compared with the historical standard time for this step The ratio, i.e. ; Step interval anomaly: Calculates the time interval between the start time of the current step and the end time of the previous step. and apply Normalize the function: ,in Sensitivity coefficient; The deviation is measured by the absolute error between the inspector's deduction value (Sstudent) and the teacher's pre-set typical error deduction value (Satandard) for this step, normalized to the [0,1] interval. ; : Equipment data consistency flag. If the quantitative data returned by the intelligent detection equipment is consistent with the result of the operator's manual selection or judgment, it is 1; otherwise, it is 0.

7. The system for achieving precise control and data-driven evaluation of practical training teaching according to claim 2, characterized in that, Detailed steps of the intelligent evaluation algorithm: Step S601: Formal definition of value-added events; Each value-added event Define a triple as follows: ; Step S602: Event scanning and execution engine; The system maintains a value-added event library. ; When a certain When an event occurs, the algorithm uses the relevant student chronology event database. Execute each event function; Step S603: Summarize reward points; student The total bonus points during this trigger period are the sum of the reward points for all triggered events: ； The brackets [·] represent Iverson brackets, with a value of 1 if the condition is true and 0 otherwise.

8. The system for achieving precise control and data-driven evaluation of practical training teaching according to claim 7, characterized in that, Triggering timing; : Boolean trigger condition function, returns or ; : Reward points.

Images