Driving simulation training effective class hour evaluation method and device

Abstract

In order to overcome the defects that in the prior art, driving simulation training duration statistics is extensive, training quality cannot be effectively evaluated, and system fault interference is likely to happen, the invention provides a driving simulation training effective class hour evaluation method and device, and the method comprises the following steps: S1, collecting operation parameters of a driving simulator and current training scene information; s2, dividing the collected data into independent time slices according to a preset time interval, and calculating an operation characteristic index in each time slice; s3, performing effective class hour and comprehensive quality evaluation by utilizing an intelligent evaluation module, wherein whether the operation characteristic indexes in each time slice are effective or not is preliminarily judged on the basis of the basic threshold parameters; in combination with current training scene information, querying a scene-operation rule base, dynamically adjusting a basic threshold parameter, and re-evaluating whether the operation characteristic index in each time slice is valid or not; and performing quality evaluation and effective class hour evaluation based on a multi-dimensional data fusion algorithm, and outputting an accumulated effective class hour and comprehensive quality evaluation report.

Description

Technical Field

[0001] The present invention relates to the field of driving simulation training, and in particular to a method and device for evaluating effective learning hours of driving simulation training. Background Art

[0002] Driving simulators are widely used in motor vehicle driving training. They provide a safe, controllable, and repeatable training environment. Accurately recording students' training time is crucial to ensuring the authenticity and effectiveness of training. However, existing driving simulation training systems have shortcomings in evaluating training effectiveness. Most systems simply record students' total online time or simulation time. This crude timing method fails to distinguish between students' actual engagement and the quality of their performance in the simulator.

[0003] During simulation training, trainees may engage in idle, inefficient, or incorrect operation. These situations result in the total recorded time not truly reflecting the trainee's effective learning time. Furthermore, the lack of quantitative analysis of trainees' specific operational behaviors (such as steering wheel control stability, throttle and clutch coordination, and rational use of brakes) means that the assessment of training quality relies heavily on the instructor's subjective observations and empirical judgment, which is not only inefficient but also difficult to standardize and objectively assess.

[0004] Furthermore, the simulator system itself may experience technical glitches, such as program freezes, delayed or lost sensor data transmission, or data anomalies caused by sensor failure. If the timing system doesn't account for these factors, time periods caused by these technical issues may be incorrectly included in or excluded from valid credit hours, distorting the assessment results and affecting the fairness and accuracy of the assessment. Summary of the Invention

[0005] In response to the shortcomings of the existing technology in driving simulation training, such as rough statistics on duration, inability to effectively evaluate training quality, and susceptibility to interference from system failures, the present invention provides a method and device for evaluating effective learning hours of driving simulation training. The method uses multi-dimensional operation signal parameters and combines simulation scenario information and fault tolerance mechanisms to quantitatively evaluate the effectiveness and operational rationality of driving simulation training. It can not only effectively analyze the quality of students' operational behavior, but also judge the effectiveness and rationality of operations in combination with training scenarios, and can identify and handle the impact of system failures on timing, thereby more accurately and reliably reflecting students' learning status and training effects.

[0006] The present invention mainly achieves the above-mentioned purpose through the following technical solutions: The present invention provides a method for evaluating effective learning hours of driving simulation training, comprising the following steps: S1: Collect the operating parameters of the driving simulator and the current training scenario information through the data acquisition module; S2: Using the data processing module to divide the data collected by the data acquisition module into independent time slices according to the preset time intervals, and calculate the operating characteristic index within each time slice; S3: Use the intelligent evaluation module to perform effective learning hours and comprehensive quality evaluation, wherein the intelligent evaluation module includes an effectiveness judgment unit, a scenario association unit, an effective learning hour accumulation unit and a quality evaluation unit; specifically, the effectiveness judgment unit preliminarily judges whether the operation characteristic indicators in each time slice are valid based on the basic threshold parameters; the scenario association unit queries the preset scenario-operation rule library in combination with the current training scenario information, wherein the scenario-operation rule library stores the adjustment instructions for the basic threshold parameters and the expectations for the rationality judgment of the operation under different training scenarios; the basic threshold parameters are dynamically adjusted based on the current training scenario information, and the effectiveness of the operation characteristic indicators in each time slice are evaluated again; quality evaluation and effective learning hours evaluation are performed based on the multidimensional data fusion algorithm, and the cumulative effective learning hours and comprehensive quality evaluation report are output.

[0007] In this technical solution, a data acquisition module captures multi-dimensional operational parameters, such as steering wheel angle, clutch pedal depth, accelerator pedal depth, brake pedal depth, current gear, vehicle speed, accumulated mileage, and engine speed. This data is then segmented into independent time slices at regular intervals, and the operational characteristic indicators within each time slice are calculated. The validity determination unit, combined with pre-defined threshold parameters, accurately distinguishes between periods of actual student operation and periods of idle or inefficient operation, preventing invalid time from being counted as valid learning hours. Furthermore, the scenario association unit dynamically adjusts thresholds based on the operational expectations of different scenarios within the scenario-operation rule library, such as extending idle time for red light waiting scenarios and relaxing steering wheel movement requirements for high-speed cruising scenarios. This reassesses the validity of the operational characteristic indicators within each time slice, thereby comprehensively evaluating the rationality of the student's operations across each training scenario. This ensures that learning hour statistics are more aligned with actual training scenario requirements, significantly improving their accuracy and authenticity, and providing a reliable data foundation for training quality assessment. The effective learning hour accumulation unit in the intelligent assessment module combines the output results of the validity determination unit and the scenario association unit, and uses a multidimensional data fusion algorithm to determine the effective time slice, thereby obtaining the accumulated effective learning hours. The quality assessment unit, based on the multidimensional data fusion algorithm, uses multidimensional operating parameters and combines training scenario information to quantitatively analyze the quality of driving simulation training. This changes the previous single evaluation method, making the evaluation results more objective and accurate, and also helps improve the quality and efficiency of training, promoting the effective improvement of students' driving skills. The current training scenario information includes: road type, road geometry, traffic conditions, weather conditions, current training task instructions, and vehicle model information.

[0008] As a technical solution, in step S1, a timestamp alignment algorithm is used to achieve time synchronization of the collected data through hardware interruption or software interpolation. The specific error can be controlled within ±5ms, eliminating the calculation error of the operating characteristic indicators caused by timing misalignment; based on the preset physical range (such as pedal depth 0-100%) and dynamic rationality rules (such as average vehicle speed, steering wheel activity, pedal activity, gear position and other logical constraints), the collected data is verified in real time for validity, outliers are filtered and missing data is marked.

[0009] As a technical solution, after the continuous data stream collected by the data acquisition module is divided into independent time slices at certain time intervals, corresponding feature calculations are performed in each time slice based on the collected operating parameters to obtain operating feature indicators in each time slice. The operating feature indicators include but are not limited to: steering wheel activity, pedal activity, pedal coordination index, vehicle dynamic changes, gear rationality, special event detection, and trajectory offset. For example, based on the collected steering wheel angle parameters, the angle change rate, steering frequency, and angle standard deviation are calculated to obtain the steering wheel activity; based on the pedal displacement parameter analysis, the pedal depth change rate and frequency characteristics are calculated to construct a pedal activity index; by quantifying the oil-clutch coordination timing error (such as the time difference between clutch release and throttle increment in the starting phase), a pedal coordination evaluation index is established; combined with vehicle speed and mileage data, the speed change, average acceleration and jerk are calculated to form a quantitative indicator of vehicle dynamic changes; the real-time vehicle acceleration is calculated based on the vehicle speed parameters, and the emergency acceleration and deceleration operation characteristic indicators are extracted; the engine speed, ignition status and clutch position information are integrated to construct an engine stall judgment model; based on the multi-dimensional correlation between vehicle speed, engine speed and gear, a gear rationality evaluation system is established to obtain a gear rationality index; safety indicators, such as counting the number of collisions per effective hour.

[0010] As a technical solution, the validity judgment unit preliminarily judges whether the operating characteristic indicators within each time slice are valid based on the basic threshold parameters, specifically including: when the average vehicle speed is greater than or equal to the set threshold, and the steering wheel activity and pedal activity are greater than or equal to the set threshold, it is judged to be valid; when the average vehicle speed is less than the set threshold, and the steering wheel activity and pedal activity are less than the set threshold, and the duration is greater than the set threshold, it is judged to be invalid; the basic threshold parameters are set in advance by the threshold calibration module.

[0011] As a technical solution, the effective learning hours of each time slice are evaluated based on a multidimensional data fusion algorithm, specifically including: the effective learning hours accumulation unit is combined with the output results of the effectiveness judgment unit and the scene association unit, and the multidimensional data fusion algorithm is used to assign weights to the parameters of each operation characteristic indicator, and finally evaluate whether each time slice is a valid time slice; the duration of all valid time slices is accumulated to obtain the accumulated effective learning hours. In a specific embodiment, the effective learning hours accumulation unit uses a multidimensional data fusion algorithm combined with a hierarchical judgment model (basic layer + enhancement layer + scene correction layer) to evaluate the effective learning hours, including: dynamic parameter weight allocation: according to the training scene type (road type, road geometry, traffic conditions, weather conditions, currently ongoing training task instructions and vehicle model information, etc.), the weight coefficients of vehicle speed, steering wheel activity, pedal activity, pedal coordination, gear rationality, trajectory offset and safety indicators are differentially configured; hierarchical judgment model: the effective operation period is screened through the base layer, based on speed The Boolean logic of speed and operational activity determines effective time slices. The enhancement layer introduces a deduction mechanism, such as deducting the corresponding duration for collision events. The scenario correction layer dynamically adjusts the judgment threshold parameters or adds judgment dimensions based on the current scenario, such as relaxing the steering wheel fluctuation limit in high-speed scenarios. Cross-dimensional correlation analysis: A three-dimensional matrix of speed, gear, and speed is constructed to detect the rationality of the power combination. Based on the long short-term memory network model (LSTM model), pedal timing data of continuous time slices is output and analyzed to identify misoperation patterns (such as mistaking the accelerator for the brake). A mileage credibility factor is introduced to quantify the continuity of driving behavior. A nonlinear accumulation algorithm is used to comprehensively determine the accumulated effective learning hours.

[0012] As a technical solution, a quality evaluation is performed on each time slice based on a multidimensional data fusion algorithm. Specifically, this includes: establishing a quality evaluation system, in which a quality evaluation unit collects operational characteristic indicators within the valid time slice, performs a quality evaluation on each operational characteristic indicator based on the current training scenario information, and obtains a quality score for each operational characteristic indicator; a multidimensional data fusion algorithm is used to assign weights to each operational characteristic indicator parameter and generate a comprehensive quality evaluation report. For example, the quality evaluation unit calculates an operational stability index based on the average operational smoothness score calculated from the standard deviation of jerk and pedal depth; calculates task-specific indicators based on the operational efficiency (e.g., task completion time) and standardization (e.g., number of line crossings) scores for specific training tasks (e.g., reverse parking, emergency avoidance); and calculates training quality evaluation indicators such as safety risk indicators based on the number of collisions, frequency of sudden braking, and proportion of dangerous operations within the valid time slice.

[0013] As a technical solution, the basic threshold parameters are pre-set, including: collecting operation data of drivers of different levels in different training scenarios through standardized tests, grouping the collected operation data according to driver level and training scenario information, performing statistical analysis on each operation data, setting initial threshold parameters, calibrating the initial threshold parameters based on expert experience, and conducting actual measurement verification and iterative optimization to obtain the basic threshold parameters. This technical solution adopts a four-stage calibration method (data collection → statistical modeling → expert calibration → online optimization). Through standardized testing, a large amount of operation data from drivers of different levels and in multiple scenarios is collected, and real driving behavior is used as the basis for threshold setting to avoid subjective conjecture; based on statistical analysis, a data model is constructed to extract operation rules from massive data, so that the threshold parameters fit the actual driving data distribution, providing objective and accurate quantitative standards for driving training and evaluation; expert experience is introduced to calibrate the initial threshold parameters to make the threshold more in line with driving teaching needs, enhance the guidance and practicality of the threshold in actual training scenarios, and significantly reduce subjective judgment errors in the threshold setting process; in the actual measurement verification and online optimization stages, feedback data from actual applications is continuously collected, and the threshold parameters are dynamically adjusted according to new scenarios, new needs and changes in students' operation data to ensure that the threshold conforms to the actual driving data distribution and effectively avoids subjective judgment errors.

[0014] As a technical solution, the system also utilizes a state monitoring and fault handling module to monitor the operating status of the driving simulator, the data update frequency of the data acquisition module, the validity of operational characteristic indicators, and the presence of "stuck" data that remains unchanged for an extended period. This module then determines whether a fault has occurred based on pre-set fault determination criteria. If a fault is detected, the accumulation of valid learning hours is suspended, and the fault time and type are marked. The system ultimately outputs the accumulated valid learning hours. The system utilizes the state monitoring and fault handling module to accurately identify technical faults such as simulator program freezes and sensor data anomalies. A double-buffered data queue and a 1-second stabilization period recovery strategy ensure assessment continuity even when the data loss rate is ≤10%, preventing miscalculation or omission of valid learning hours due to system issues. This ensures that students' training time data is authentic and valid, eliminates assessment bias caused by technical failures in traditional timing systems, and improves the reliability of assessment results. Furthermore, clear fault records assist in analyzing equipment operating conditions, allowing managers to quickly locate and resolve equipment issues, reduce disruptions to normal teaching procedures due to equipment failures, and provide data support for preventive maintenance and system optimization. As a technical solution, the current training scene information includes: road type (such as city, highway, field, etc.), road geometry (such as straight line, curve curvature, slope, etc.), traffic conditions (such as traffic light status, surrounding vehicle density, etc.), weather conditions (such as sunny, rainy, cloudy, etc.), current ongoing training task instructions (such as starting, turning, reversing, etc.) and vehicle model information, etc.

[0015] The present invention also provides an evaluation device for the effective learning hours of driving simulation training, including a data acquisition module, a data processing module, and an intelligent evaluation module. The data acquisition module is used to collect the operating parameters of the driving simulator and the current training scene information; the data processing module is used to divide the collected data into independent time slices according to preset time intervals, and calculate the operating characteristic indicators within each time slice; the intelligent evaluation module is used to perform effective training hours and comprehensive quality evaluation on each time slice.

[0016] As a technical solution, it also includes a status monitoring and fault handling module, which is used to monitor the operating status of the driving simulator, the data update frequency of the data acquisition module, whether the operating characteristic indicators are valid, whether there is "stuck" data that has not changed for a long time, etc., and judge whether a fault has occurred based on the preset fault judgment conditions; if a fault is detected, the accumulation of valid learning hours will be suspended, and the fault time and fault type will be marked.

[0017] As a technical solution, it also includes a threshold calibration module for calibrating various basic threshold parameters and scenario-operation rule libraries used in the intelligent evaluation module.

[0018] As a technical solution, the intelligent evaluation module includes a validity judgment unit, a scenario association unit, an effective learning hour accumulation unit and a quality evaluation unit. The validity judgment unit is used to preliminarily judge whether the operation characteristic indicators within each time slice are valid; the scenario association unit is used to combine the current training scenario information, query the preset scenario-operation rule library, dynamically adjust the basic threshold parameters, and re-evaluate whether the operation characteristic indicators within each time slice are valid; the effective learning hour accumulation unit is used to calculate and output the accumulated effective learning hours; the quality evaluation unit is used to generate a quality evaluation report.

[0019] As a technical solution, it also includes a user interface module, which provides users (such as coaches and students) with an interactive interface to view real-time training evaluation results, study hour statistics, error analysis, etc., and real-time visualization of the current effective study hour progress bar, real-time deduction reasons, 3D trajectory playback and offset heat map, etc. Through the data interaction function, it supports the retrieval of abnormal events (such as the moment of engine shutdown) by timeline and the associated scenarios for replay analysis.

[0020] The present invention provides a method and device for evaluating effective learning hours in driving simulation training. Based on the trainee's multi-dimensional real-time operation signal parameters and the current simulator training scenario information, the method intelligently determines effective learning hours, evaluates operational effectiveness, and effectively eliminates interference from system failures on timing evaluation. Compared with existing technologies, the present invention has the following advantages: (1) Dynamic scenario adaptation capability: A scenario-operation rule library is used to achieve differentiated evaluation in multiple scenarios (such as extending the tolerance time for waiting at a red light and comparing theoretical values ​​for turning on a curve), thus avoiding evaluation bias caused by traditional fixed thresholds and improving the targeted nature of teaching.

[0021] (2) Multi-dimensional quality fusion assessment "combines the hierarchical judgment model (basic layer + enhancement layer + scene correction layer)" with LSTM time series analysis, which not only counts the effective time, but also identifies hidden misoperations (such as mistaking the accelerator for the brake), and realizes the differentiated accumulation of learning hours through quality score evaluation.

[0022] (3) Highly robust fault-tolerant mechanism: The double-buffered data queue and 1-second stable period recovery strategy are used to ensure the continuity of evaluation when the data loss rate is ≤10%, avoiding miscalculation of learning hours due to simulator freezes or sensor abnormalities.

[0023] (4) Scientific threshold calibration process: Through the four-stage calibration method (data collection → statistical modeling → expert calibration → online optimization), the threshold is ensured to be consistent with the actual driving data distribution and to incorporate teaching experience, thereby reducing subjective judgment errors.

[0024] (5) Real-time visualization and interactive analysis: The user interface supports 3D trajectory heat maps and timeline event retrieval, helping instructors quickly locate students' operational weaknesses (such as frequent engine stalls) and improve teaching efficiency.

[0025] (6) Effective and reasonable quality evaluation: Based on the effectiveness judgment, quality score evaluation is introduced, and a multidimensional data fusion algorithm is used to assign weights to various operation characteristic indicator parameters to generate a comprehensive quality assessment report, providing multidimensional data support for skill certification. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Attachment Figure 1 This is a structural block diagram of a device for evaluating effective learning hours for driving simulation training according to an embodiment of the present invention; Attachment Figure 2 Schematic diagram of a flow chart of a method for evaluating effective learning hours of driving simulation training in an embodiment of the present invention; Attachment Figure 3 The figure is a schematic flow chart of the status monitoring and fault handling logic in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] This embodiment provides a driving simulation training effective learning time evaluation device, which is deployed on the control computer of the driving simulator, such as Figure 1 As shown, the evaluation device includes a data acquisition module, a data processing module, an intelligent evaluation module, a status monitoring and fault processing module, a threshold calibration module and a user interface module.

[0028] The data acquisition module is responsible for communicating with the driving simulator software or hardware interface to collect real-time, high-frequency operating parameters of the driving simulator, such as steering wheel angle (SteeringAngle), clutch pedal depth (ClutchDepth), accelerator pedal depth (ThrottleDepth), brake pedal depth (BrakeDepth), current gear (Gear), vehicle speed (VehicleSpeed), accumulated mileage (DistanceTraveled), engine speed (EngineRPM), etc. It also obtains current training scenario information provided by the driving simulator, such as road type (city, highway, field), road geometry (straight, curve curvature, slope), traffic conditions (signal status, surrounding vehicle density), weather, and the current training task instructions (such as starting, turning, and reverse parking). To avoid feature calculation errors caused by acquisition delays, a timestamp synchronization mechanism is used to ensure the time alignment of multi-sensor data and to verify data validity, for example, by checking whether the pedal depth value is within the range of 0-100%.

[0029] The data processing module is configured to receive the raw data stream from the data acquisition module, segment the collected continuous data stream into independent time slices at preset time intervals Δt, and calculate operational characteristic indicators within each time slice Δt. For example, based on the collected steering wheel angle parameters, the angle change rate, steering frequency, and angle standard deviation are calculated to obtain the steering wheel activity. The pedal depth change rate and frequency characteristics are calculated based on the pedal displacement parameters to construct a pedal activity index. A pedal coordination evaluation index is established by quantifying the timing error between clutch release and throttle increase during the start phase. Speed ​​change, average acceleration, and jerk are calculated based on vehicle speed and mileage data to form a quantitative indicator of vehicle dynamic changes. The real-time vehicle acceleration is calculated based on the vehicle speed parameter to extract emergency acceleration and deceleration operation characteristic indicators. An engine stall determination model is constructed by integrating engine speed, ignition status, and clutch position information. A gear rationality evaluation system is established based on the multi-dimensional correlation between vehicle speed, engine speed, and gear position to obtain a gear rationality index. Safety indicators such as the number of collisions per effective hour are also calculated.

[0030] The intelligent evaluation module is the core decision-making unit, which is used to evaluate the effective training hours and comprehensive quality of each time slice, including an effectiveness judgment unit, a scene association unit, an effective training hour accumulation unit and a quality evaluation unit. The effectiveness judgment unit makes a preliminary judgment on the operating characteristic indicators in each time slice based on the preset basic threshold parameters. If the average vehicle speed in a time slice is greater than the minimum speed (for example, 5km / h), or the steering wheel activity and pedal activity are greater than or equal to the set threshold although the speed is low, it is preliminarily judged to be valid. At the same time, it also contains exclusion rules. For example, if the average vehicle speed is less than the set threshold, and all activity indicators are lower than the threshold, and the duration exceeds the threshold (such as 30 seconds), it is judged to be invalid. These basic threshold parameters are determined by the threshold calibration module. The scenario association unit is used to query a preset scenario-operation rule library based on the current training scenario information. This rule library stores adjustment instructions for validity determination rules and / or expectations for operational rationality for different scenarios (such as "urban straight road," "curve," "hill start task," and "waiting at a red light"). The scenario association unit dynamically adjusts basic threshold parameters and re-evaluates the validity of the operational characteristic indicators within each time slice. For example, when detecting a "waiting at a red light" scenario, it instructs the validity determination unit to temporarily extend a certain threshold; when detecting a "high-speed cruising" scenario, it relaxes the steering wheel movement requirements; when detecting a "curve" scenario, it compares the current steering wheel angle with the theoretical angle calculated based on vehicle speed and curve curvature to assess steering suitability; when detecting a "hill start," it checks the throttle and clutch coordination timing, back-slip distance, and whether the engine is stalled, and assigns a rationality score. The effective learning hour accumulation unit, based on the output of the scenario association unit, uses a multidimensional data fusion algorithm to assign weights to each operational characteristic indicator parameter, ultimately evaluating whether each time slice is valid. This unit is controlled by the status monitoring and fault handling module and accumulates time during the pause of the faulty device.Preferably, the effective learning hour accumulation unit uses a multi-dimensional data fusion algorithm combined with a hierarchical judgment model (base layer + enhancement layer + scenario correction layer) to evaluate effective learning hours, including: dynamic parameter weight allocation: differentiated configuration of weight coefficients for vehicle speed, steering wheel activity, pedal activity, pedal coordination, gear rationality, trajectory offset, and safety indicators according to the training scenario type (road type, road geometry, traffic conditions, weather conditions, currently ongoing training task instructions, and vehicle model information, etc.); hierarchical judgment model: screening effective operation periods through the base layer, determining effective time slices based on Boolean logic of speed and operation activity, introducing a deduction mechanism in the enhancement layer, such as performing corresponding time deductions for collision events, and dynamically adjusting the judgment threshold parameters or adding judgment dimensions based on the current scenario in the scenario correction layer, such as relaxing steering wheel fluctuation restrictions in high-speed scenarios; cross-dimensional correlation analysis: constructing a three-dimensional matrix of speed-gear-speed to detect the rationality of the power combination, outputting and analyzing pedal timing data of continuous time slices based on the long short-term memory network model (LSTM model) to identify misoperation modes (such as using the accelerator as the brake), and introducing a mileage credibility factor to quantify the continuity of driving behavior. A nonlinear cumulative algorithm is used to comprehensively determine the accumulated effective learning hours. The quality assessment unit collects operational characteristic indicators within the effective time slice and performs a quality evaluation on each operational characteristic indicator based on the current training scenario information, obtaining a quality score for each operational characteristic indicator. A multidimensional data fusion algorithm is used to assign weights to each operational characteristic indicator parameter, ultimately generating a comprehensive quality assessment report. For example, the quality assessment unit calculates an operational stability index based on the average operational smoothness score calculated from the standard deviation of jerk and pedal depth. It also calculates task-specific indicators based on operational efficiency (e.g., task completion time) and standardization (e.g., number of line crossings) for specific training tasks (e.g., reverse parking, emergency avoidance). Training quality evaluation indicators such as safety risk indicators are calculated based on the number of collisions, frequency of sudden braking, and proportion of dangerous operations within the effective time slice.

[0031] The user interface module is used to visually display the current effective learning hour progress bar, real-time deduction reasons, 3D trajectory playback and offset heat map, etc. in real time. It can view real-time training evaluation results, learning hour statistics, error analysis, etc., support retrieval of abnormal events (such as engine shutdown time) by timeline, and associate scenes for replay analysis.

[0032] The state monitoring and fault handling module is used to monitor the operating status of the driving simulator, the data update frequency of the data acquisition module, whether the received sensor values ​​are within the valid range, whether there is "stuck" data that has not changed for a long time, and the communication heartbeat with the simulator; based on the preset fault judgment conditions (for example, the data update frequency is <5Hz for 2 seconds, or the braking data with NaN values ​​is continuously received), it is judged whether a fault has occurred; if a fault is detected, a "pause timing" signal is immediately sent to the effective learning hour accumulation unit to suspend the effective learning hour accumulation, and the fault time and fault type are marked. When the fault condition is detected to be lifted and stabilized for a period of time, a "resume timing" signal is sent. The fault detection logic is as follows: Data integrity check: A fault is triggered when NaN or out-of-range data is received for 10 consecutive frames; Update frequency detection: If the data frame rate is less than 5Hz for 2 seconds, it is considered a communication abnormality; Data freeze detection: Alarm when key parameters (such as vehicle speed) remain unchanged for 5 consecutive seconds.

[0033] The threshold calibration module is used to calibrate various basic threshold parameters and scenario-operation rule bases used in the intelligent assessment module. The calibration of basic threshold parameters includes: (1) Baseline data collection: Organize drivers of different skill levels (such as novices, experienced trainees, and instructors) to perform standardized driving tasks in different simulators (such as manual transmission and automatic transmission) and record their complete operating parameter data.

[0034] (2) Statistical analysis: The collected benchmark data are grouped according to the driver level and training scenario information, and statistical analysis (such as calculating the mean, standard deviation, percentile, and drawing a distribution chart) is performed on each operation data (such as the steering wheel angle change rate, the accelerator pedal change rate, etc.) to identify the parameter characteristics and their numerical ranges that can distinguish different levels or different operation states (such as smooth vs. impatient).

[0035] (3) Initial threshold setting: Based on the statistical analysis results, the threshold parameters required for each judgment rule are preliminarily set (for example, the 90th percentile value of a parameter of an experienced driver in a certain scenario is set as the threshold for distinguishing "unreasonable" operations).

[0036] (4) Calibration based on expert experience: The initially set threshold and its corresponding judgment logic are submitted to senior driving instructors for review, and their feedback based on teaching experience is collected (such as “Is the cornering steering deviation threshold reasonable?”), and the threshold is adjusted accordingly.

[0037] (5) Field test verification and iterative optimization: The evaluation system with calibrated threshold parameters is put into a small-scale actual training test. The effective learning hours and evaluation results output by the system are compared with the manual observation and evaluation of the coach. The reasons for the differences are analyzed. Based on the verification results, the threshold is continuously fine-tuned and optimized using the gradient descent method until the expected evaluation accuracy is achieved.

[0038] The scenario-operation rule library includes: (1) Expected operating mode: describes the sequence of driving behaviors or states that are generally expected in this scenario (e.g., the standard sequence of steps for starting on a hill).

[0039] (2) Reasonable parameter range: defines the reasonable or acceptable numerical range of various operating characteristic indicators (such as steering wheel stability, speed control, and pedal smoothness) in this scenario.

[0040] (3) Validity determination adjustment rules: define the modifications that need to be made to the basic validity determination threshold or logic in specific scenarios (for example, adjusting the stationary time threshold in the red light waiting scenario).

[0041] (4) Scenario-based threshold grouping: Different scenarios (e.g., city vs. highway) use independent threshold sets to avoid evaluation bias caused by global thresholds.

[0042] This embodiment also provides a method for evaluating effective learning hours of driving simulation training. The flowchart is as follows: Figure 2 As shown, the following steps are included: A1: Threshold Calibration. Before the system is officially used or periodically, the threshold calibration module is used to calibrate the various basic threshold parameters and scenario-operation rule base used in the intelligent assessment module. This includes: collecting operational data from drivers of different skill levels in various training scenarios through standardized testing; grouping the collected operational data by driver skill level and training scenario information; performing statistical analysis on each operational data item; setting initial threshold parameters; calibrating the initial threshold parameters based on expert experience; and conducting field testing, verification, and iterative optimization to determine the basic threshold parameters.

[0043] A2: Data Acquisition. While the system is running, the data acquisition module continuously acquires real-time multi-dimensional operating parameters and current training scenario information from the driving simulator. A timestamp alignment algorithm is used to achieve time synchronization of the collected data. The collected data is validated based on preset physical ranges and dynamic rationality rules, filtering outliers and marking missing data. The operating parameters include, but are not limited to, steering wheel angle, clutch pedal depth, accelerator pedal depth, brake pedal depth, current gear, vehicle speed, accumulated mileage, and engine speed. The current training scenario information includes, but is not limited to, road type (e.g., urban, highway, field), road geometry (e.g., straight, curve curvature, slope), traffic conditions (e.g., signal light status, surrounding vehicle density), weather conditions (e.g., sunny, rainy, cloudy), currently ongoing training task instructions (e.g., starting, turning, reversing), and vehicle model information.

[0044] A3: Feature Extraction. The data processing module segments the collected data into time slices based on Δt and calculates the operational characteristic indicators within each slice. These operational characteristic indicators include, but are not limited to, steering wheel activity, pedal activity, pedal coordination index, vehicle dynamics, gear position rationality, special event detection, and trajectory deviation.

[0045] A4: Status monitoring and fault handling. This step is performed in parallel with or immediately after feature extraction or effectiveness evaluation. Use the status monitoring and fault handling module to monitor the operating status of the driving simulator, the data update frequency of the data acquisition module, whether the operating characteristic indicators are valid, whether there is "stuck" data that has not changed for a long time, etc., and judge whether a fault has occurred based on the preset fault judgment conditions; if a fault is detected, the accumulation of valid learning hours is suspended, and the fault time and fault type are marked; finally, the accumulated valid learning hours are output. If a fault condition is detected (judgment is "yes"), fault handling is performed, such as setting the "fault flag", notifying the pause timing, and cyclically waiting for the fault to be resolved. If no fault is detected (judgment is "no"), the process continues. If Figure 3 As shown, the specific logic of the status monitoring and fault handling module can be as follows: A41: Get data. Continuously receive simulator status reports and operation signal parameter data.

[0046] A42: Check the fault conditions. Check each preset fault condition to see if it is met. For example: IsSimulatorRunning (is it true?)? DataUpdateFrequency (is it >= MinFrequencyThreshold?)? AreDataValuesValid (has NaN or out-of-range data been received for 10 consecutive frames?)? IsDataFrozen (checks whether key parameters remain unchanged for an extended period of time?)? IsCommunicationOK (checks the heartbeat or connection status?) A43: Determine whether there is a fault. If any fault condition is met, it is determined to be a fault state. Otherwise, it is a normal state.

[0047] A44: Handle the fault. If it is determined to be a fault: set the internal "fault flag" to true; send a "timeout" signal to the effective learning hour accumulation unit; record the fault start time and possible fault type; return to A41 to continue monitoring.

[0048] A45: Handling normal status / fault recovery. If normal: Check whether the previous "fault flag" is true. If it is, it means the fault has just been resolved and you need to wait for a period of stabilization. If not, proceed normally.

[0049] A46: Wait for the stabilization period. Based on the simulator data refresh rate (≥10Hz), set a 1-second stabilization period to ensure data continuity after fault recovery.

[0050] A47: Resume timing. If stable recovery is confirmed, the "fault flag" is set to false, a "resume timing" signal is sent to the effective learning hour accumulation unit S133, and the fault end time is recorded.

[0051] A48: Process normally. Allow the intelligent assessment module to perform effectiveness assessment and credit accumulation normally.

[0052] Return to A41 to continue monitoring.

[0053] A5: Effectiveness Assessment. This step is only performed in full when no fault is detected, or conditionally performed based on the fault status. Use the intelligent assessment module to perform effective learning hours and comprehensive quality assessment, including: A51: Applying the basic threshold value: The validity determination unit preliminarily determines whether the operating characteristic indicator in each time slice is valid based on the preset basic threshold value parameters, and obtains a preliminary validity flag (valid / invalid).

[0054] A52: Scenario Association. The scenario association unit obtains the current training scenario information and queries the scenario-operation rule library, which stores the adjustment instructions for basic threshold parameters under different training scenarios and the expectations for judging the rationality of operations.

[0055] A53: Dynamically adjust thresholds. Dynamically adjusts basic threshold parameters based on the current training scenario information. If the rules require adjustment, modify the threshold parameters or decision logic. If the rules require a reasonableness assessment, compare the current operating characteristics with the scenario expectations to obtain a reasonableness score / grade.

[0056] A54: Comprehensive Determination. Combined with the current training scenario information, the system re-evaluates the validity of the operational characteristic indicators within each time slice and records the results. Even if the initial determination is invalid (e.g., low speed), if the scenario association indicates a "reverse parking" task and the operation is active, the final determination is valid.

[0057] A6: Accumulate valid learning hours. The valid learning hour accumulation unit combines the output of the scenario association unit and uses a multidimensional data fusion algorithm to assign weights to the various operational characteristic indicator parameters. Ultimately, it evaluates whether each time slice is valid. If no fault is detected and the current time slice is determined to be valid, the valid learning hour accumulation unit adds the valid time slice to the total valid learning hours. If a fault is detected, this step is paused.

[0058] A7: Comprehensive Quality Assessment. 6. Establish a quality assessment system. The quality assessment unit collects operational characteristic indicators within the valid time slice, performs a quality assessment on each operational characteristic indicator based on the current training scenario information, and obtains a quality score for each operational characteristic indicator. A multidimensional data fusion algorithm is used to assign weights to each operational characteristic indicator parameter and generate a comprehensive quality assessment report.

[0059] A8: Loop or End. Determine whether training is complete. If not, return to A2 to continue processing the next time slice. If complete, proceed to A9.

[0060] A9: Output report.

[0061] While the specific embodiments of the present invention have been described above with reference to the accompanying drawings, it should be understood by those skilled in the art that the above embodiments are for illustrative purposes only and are not intended to limit the scope of the present invention. It should be understood by those skilled in the art that modifications may be made to the above embodiments without departing from the scope and spirit of the present invention. The scope of protection of the present invention is defined by the appended claims.

Claims

1. A method for evaluating effective learning hours of driving simulation training, characterized in that: The following steps are involved: S1: Collect the operating parameters of the driving simulator and the current training scenario information through the data acquisition module; S2: Using the data processing module to divide the data collected by the data acquisition module into independent time slices according to the preset time intervals, and calculate the operating characteristic index within each time slice; S3: Use the intelligent evaluation module to conduct effective learning hours and comprehensive quality evaluation, including: the effectiveness judgment unit preliminarily judges whether the operation characteristic indicators in each time slice are valid based on the basic threshold parameters; the scenario association unit combines the current training scenario information, queries the preset scenario-operation rule library, dynamically adjusts the basic threshold parameters, and re-evaluates whether the operation characteristic indicators in each time slice are valid; based on the multi-dimensional data fusion algorithm, each time slice is evaluated for effective learning hours and quality, and finally outputs the cumulative effective learning hours and comprehensive quality evaluation report.

2. A driving simulation training effective learning time evaluation method according to claim 1, characterized in that: In step S1, a timestamp alignment algorithm is used to achieve time synchronization of the collected data; the collected data is validated based on the preset physical range and dynamic rationality rules, outliers are filtered and missing data are marked.

3. The method for evaluating effective learning hours of driving simulation training according to claim 1, characterized in that: The operating parameters of the driving simulator include: steering wheel angle, clutch pedal depth, accelerator pedal depth, brake pedal depth, current gear, vehicle speed, accumulated mileage, and engine speed.

4. A driving simulation training effective learning time evaluation method according to claim 1 or 3, characterized in that: The data processing module calculates the operating characteristic indicators within each time slice based on the collected operating parameters. The operating characteristic indicators include: steering wheel activity, pedal activity, pedal coordination index, vehicle dynamic changes, gear rationality, special event detection, and trajectory offset.

5. The method for evaluating effective learning hours of driving simulation training according to claim 1, characterized in that: The effective learning hours of each time slice are evaluated based on the multidimensional data fusion algorithm, specifically including: the effective learning hours accumulation unit is combined with the output results of the scene association unit, and the multidimensional data fusion algorithm is used to assign weights to the parameters of each operation characteristic indicator, and finally evaluate whether each time slice is a valid time slice; the duration of all valid time slices is accumulated to obtain the cumulative effective learning hours.

6. A driving simulation training effective learning time evaluation method according to claim 5, characterized in that: The quality of each time slice is evaluated based on a multidimensional data fusion algorithm, specifically including: establishing a quality evaluation system, in which the quality evaluation unit collects the operation characteristic indicators within the effective time slice, and evaluates the quality of each operation characteristic indicator according to the current training scenario information to obtain the quality score of each operation characteristic indicator; using a multidimensional data fusion algorithm to assign weights to the parameters of each operation characteristic indicator and generate a comprehensive quality evaluation report.

7. The method for evaluating effective learning hours of driving simulation training according to claim 1, characterized in that: The basic threshold parameters are pre-set, including: collecting operation data of drivers of different levels in different training scenarios through standardized tests, grouping the collected operation data according to driver level and training scenario information, performing statistical analysis on each operation data, setting initial threshold parameters, calibrating the initial threshold parameters based on expert experience, and conducting actual measurement verification and iterative optimization to obtain the basic threshold parameters.

8. The method for evaluating effective learning hours of driving simulation training according to claim 1, characterized in that: It also includes using the status monitoring and fault handling module to monitor the operating status of the driving simulator, and determine whether a fault has occurred based on preset fault judgment conditions; if a fault is detected, the accumulation of effective learning hours is suspended, and the fault time and fault type are marked; and finally the accumulated effective learning hours are output.

9. The method for evaluating effective learning hours of driving simulation training according to claim 1, characterized in that: The current training scenario information includes: road type, road geometry, traffic conditions, weather conditions, currently ongoing training task instructions, and vehicle model information.

10. An evaluation device based on the driving simulation training effective learning hours evaluation method according to claim 1, characterized in that: It includes a data acquisition module, a data processing module, and an intelligent evaluation module. The data acquisition module is used to collect the operating parameters of the driving simulator and the current training scenario information; the data processing module is used to divide the collected data into independent time slices according to preset time intervals and calculate the operating characteristic indicators within each time slice; the intelligent evaluation module is used to evaluate the effective training hours and comprehensive quality of each time slice.

11. The evaluation device according to claim 10, characterized in that It also includes a status monitoring and fault handling module, which is used to monitor the operating status of the driving simulator and determine whether a fault has occurred based on preset fault judgment conditions; if a fault is detected, the accumulation of valid learning hours will be suspended, and the fault time and fault type will be marked.

12. The evaluation device according to claim 10, characterized in that It also includes a threshold calibration module for calibrating various basic threshold parameters and scenario-operation rule libraries used in the intelligent evaluation module.

13. The evaluation device according to claim 10, characterized in that The intelligent evaluation module includes a validity determination unit, a scenario association unit, a valid learning hour accumulation unit and a quality evaluation unit. The validity determination unit is used to preliminarily determine whether the operation characteristic indicator within each time slice is valid; The scenario association unit is used to combine the current training scenario information, query the preset scenario-operation rule library, dynamically adjust the basic threshold parameters, and re-evaluate whether the operation characteristic indicators within each time slice are valid; the effective learning hour accumulation unit is used to calculate and output the accumulated effective learning hours; the quality assessment unit is used to generate a quality assessment report.

Images