A cognitive motor adaptive training method and system

Abstract

The application relates to a cognitive motor adaptive training method, which comprises the following steps: S101, determining an initial training level corresponding to a basic action level of a subject, and determining a training intensity parameter boundary corresponding to the initial training level; S102, determining a current training parameter combination within the training intensity parameter boundary, and generating a current training task according to the current training parameter combination; S103, acquiring a response feature of the subject when the subject executes the current training task, and converting the response feature into a quantitative performance index for evaluating real-time performance of the subject; S104, comparing the quantitative performance index with a preset threshold to obtain a comparison result, and incrementally adjusting the current training parameter combination according to the comparison result to obtain a next training parameter combination; and S105, judging whether a training termination condition is reached, and if not, applying the next training parameter combination to generate a new current training task in the next training iteration, and returning to the execution step S103 based on the new current training task.

Description

Technical Field

[0001] This application relates to the field of intelligent sports and rehabilitation training technology, and in particular to a cognitive movement adaptive training method and system. Background Technology

[0002] With the development of rehabilitation medicine and sports science, cognitive and motor dual-task training for children has become an important means to enhance neuroplasticity. This type of training usually requires subjects to complete specific cognitive processing tasks (such as symbol recognition and logical decision-making) while performing physical movements (such as balance maintenance and limb extension).

[0003] Existing technologies typically employ a tiered training model. This model pre-sets several task levels with fixed difficulty. After completing the task objectives for a specific stage, the system or a human operator switches to a higher difficulty level according to a predetermined step-by-step progression.

[0004] However, due to the high individual differences and dynamic fluctuations in the physiological and cognitive states of subjects in a dual-task environment, this rigid adjustment mechanism based on fixed levels is difficult to match the actual ability range of subjects in real time. This causes the training load to frequently oscillate between "insufficient stimulation" and "overload," making it difficult to accurately reach the subject's ability boundaries, thus limiting the improvement of overall training efficiency. Summary of the Invention

[0005] Therefore, it is necessary to provide a cognitive motion adaptive training method and system that improves training accuracy to address the aforementioned technical problems.

[0006] In a first aspect, this application provides a cognitive motion adaptive training method, which is applied to a task design controller in a cognitive motion adaptive training system. The method includes the following steps: Step S101: Determine the corresponding initial training level based on the subject's basic action level, and determine the training intensity parameter boundary corresponding to the initial training level; Step S102: Within the training intensity parameter boundary, determine the current training parameter combination and generate the current training task based on the current training parameter combination; Step S103: Obtain the response features of the subject when performing the current training task, and convert the response features into a quantitative performance index to evaluate the subject's real-time performance. ; Step S104: Compare the quantitative performance index with the preset threshold to obtain the comparison result, and make incremental adjustments to the current training parameter combination based on the comparison result to obtain the next training parameter combination. Step S105: Determine whether the training termination condition has been met. If not, in the next training iteration, apply the next training parameter combination to generate a new current training task, and return to step S103 based on the new current training task.

[0007] Secondly, this application provides a cognitive-motor adaptive training system, the training system including a task design controller, the task design controller including: Baseline Level Assessment Module: Used to determine the corresponding initial training level based on the subject's basic action level, and to determine the training intensity parameter boundary corresponding to the initial training level; Multidimensional task generation module: Used to determine the current combination of training parameters within the training intensity parameter boundary, and generate the current training task based on the current combination of training parameters; Interactive Level Quantization Module: Used to acquire the response features of subjects when performing the current training task, and to convert the response features into a quantitative performance index to evaluate the subjects' real-time performance. ; The adaptive iteration module is used to compare the quantitative performance index with a preset threshold, obtain the comparison result, and make incremental adjustments to the current training parameter combination based on the comparison result to obtain the next training parameter combination. It is also used to determine whether the training termination condition has been met. If not, it triggers the multi-dimensional task generation module to generate a new training task using the next training parameter combination until the training termination condition is met.

[0008] The beneficial effects of this application are as follows: By using the logic of "quantitative index + incremental adjustment", continuous adjustment of training intensity parameters is achieved, effectively avoiding the intensity abrupt changes caused by excessively large step sizes in level switching in traditional graded training, and enabling the training load to more accurately adapt to the real-time physiological and cognitive state of the subjects; iterative processing based on real-time performance index of response characteristics enables the system to promptly identify fluctuations in the subjects' performance and make parameter corrections, improving the response speed and dynamic adaptability of the adaptive process; by setting initial levels and their corresponding intensity parameter boundaries, while ensuring training safety, iterative logic is used to achieve smooth tracking of the subject's ability boundaries, optimizing the overall training efficiency. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the accompanying drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below.

[0010] Figure 1 This is a flowchart illustrating the cognitive-motor adaptive training method in the embodiments of this application; Figure 2 An architecture diagram of the task design controller in the cognitive motion adaptive training system of this application is shown. Figure 3 This is an architecture diagram of the instantaneous damping control subsystem applied to adaptive training in the embodiments of this application. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0012] In one exemplary embodiment, a cognitive motion adaptive training method is provided, characterized in that the method is applied to a cognitive motion adaptive training system, specifically executed by a task design controller within the cognitive motion adaptive training system, such as... Figure 1 As shown, the method includes the following steps: Step S101: Determine the corresponding initial training level based on the subject's basic action level, and determine the training intensity parameter boundary corresponding to the initial training level; This step aims to establish an "initial coordinate system" for the subject's abilities. It addresses the technical shortcomings of traditional sports training, such as the "blindness of initial difficulty," which leads to frustration due to tasks being too difficult or insufficient stimulation for neural remodeling due to tasks being too easy. Based on the "proximal developmental zone" theory, an individual's cognitive and motor performance exist in a dynamic equilibrium zone. Through a pre-set baseline test, the maximum load envelope that the individual can withstand is determined, preventing subsequent tasks from exceeding this limit.

[0013] The implementation process is as follows: First, the interactive interface initiates the baseline assessment procedure. Subjects must complete a set of standardized action responses (e.g., physically touching a specific light spot signal in an undisturbed state). The average response time and accuracy of the subjects are recorded using sensors and defined as the "basic action level."

[0014] The internally stored level mapping matrix is ​​then retrieved. This matrix discretizes continuous capability values ​​into multiple tiered levels. For a given level, a set of "training intensity parameter boundaries" is locked. These boundaries physically define the "upper limit" and "lower limit" for subsequent task generation (e.g., the maximum amplitude of motor torque fluctuations, the maximum frequency of visual signal occurrences). This boundary setting at the logic layer ensures that adaptive adjustment always evolves within a safe and effective range, avoiding system oscillations.

[0015] Step S102: Within the training intensity parameter boundary, determine the current training parameter combination and generate the current training task based on the current training parameter combination; This step realizes the instantiation of "abstract cognitive parameters" into "concrete physical tasks." Its core logic lies in using "parametric task generation technology" to decompose complex motor behaviors into multiple controlled physical variables. Through the logical combination of multidimensional parameters, specific executive function load pressure is induced in the subject's nervous system. Within the training levels (L1-L4) and their corresponding boundary envelopes determined in S101, the system selects a specific set of parameter values ​​(i.e., the current training parameter combination). These parameters encompass spatial location, temporal rhythm, and logical mapping rules.

[0016] First, the system breaks down the selected complexity index (C / Ct) into executable hardware control instructions.

[0017] For example, when the system sets the complexity parameters N=2 (visual + auditory dual stimulation), R=2 (dual rule), and T=0.8s (reaction time limit), the generated training task will directly affect the intelligent hardware on the field. Taking basketball dribbling training as an example, the system drives the field's light cues to randomly flash "blue" or "red" signals, while simultaneously driving an intelligent whistle generator to emit a whistle at a specific time. Subjects are required to: upon seeing the blue signal, immediately dribble in the direction of the signal; if they hear the whistle during the action, they must instantly suppress their current dribbling impulse and remain still while holding the ball. In this process, the NN parameter increases the difficulty of attention allocation through the light / sound dual modality, while the TT parameter strengthens the subject's "inhibition control" ability under high-speed motion through time pressure.

[0018] Secondly, the system constructs dynamic decision-making scenarios by combining spatial path parameters (P) and switching frequency (F).

[0019] Taking agility rope ladder training as an example, the system dynamically generates three passable paths at the end of the rope ladder using ground markers or projection equipment, based on the determined parameters P=3P=3 (three-way branching path). Simultaneously, the system retrieves voice commands from a multimodal stimulus library, requiring subjects to execute different gait rules at different stages of the rope ladder traversal (e.g., hopping in the red zone, crossing in the green zone). As the switching frequency FF increases, the system shortens the physical distance between different gait rules, forcing subjects to continuously "reorganize rules" during high-speed movement. This task generation method couples spatial displacement decisions with complex action sequences, effectively inducing an improvement in the subject's cognitive flexibility.

[0020] Finally, the system encapsulates the generated task sequence into a digital instruction package.

[0021] These instruction packages are pushed in real time to teaching terminals (such as tablets and mobile phones) and related hardware in the venue (such as automatic ball-serving machines and timing sensors). For example, for table tennis / badminton specific tasks, the instruction package will drive the automatic ball-serving machine to output balls of different colors (visual stimulation), requiring the subject to perform the corresponding return action (such as smashing the red ball and chopping the blue ball) according to the ball color (rule mapping). In this way, the system constructs an immediate obstacle scenario with specific cognitive challenges in the physical space, thus completing the automated deployment of the "current training task". This task generation mechanism based on parameter combination completely solves the technical problems of monotonous and repetitive content and unclear cognitive intervention goals in traditional sports training. Step S103: Obtain the response characteristics of the subject when performing the current training task, and convert the response characteristics into a quantitative performance index E to evaluate the subject's real-time performance; This step extracts digital features from the subject's "biofeedback." It addresses the issues of subjectivity and lag in motor performance evaluation, achieving millisecond-level characterization of the subject's real-time state. Specifically, it is based on the "Speed-Accuracy Trade-off (SAT)" physical model. Individual performance should not be measured by a single indicator, but rather as a comprehensive function of the accuracy of movement execution and time cost.

[0022] The implementation process is as follows: During the interaction between the subject and the hardware execution terminal, the feedback sensor group (such as a high-precision encoder and force sensor) captures physical signals in real time. Key "response features" are extracted, including: the time interval between the electrical signal triggering and the subject's physical response (reaction time), the spatial deviation between the movement trajectory and the target point (positional accuracy), and the judgment accuracy in the presence of logical interference.

[0023] These heterogeneous features are then fused using a pre-defined quantization operator. By normalizing the temporal features and incorporating spatial precision weights, a scalar value E is calculated. This index E physically represents the subject's "functional redundancy" under the current workload: a higher E value indicates that the subject still has the potential to cope with more challenging tasks.

[0024] Step S104: Compare the quantitative performance index with the preset threshold to obtain the comparison result, and make incremental adjustments to the current training parameter combination based on the comparison result to obtain the next training parameter combination. This step achieves "negative feedback closed-loop regulation" of the system load. It addresses the problem of the training process being unable to achieve real-time dynamic convergence based on the subject's fatigue level or progress rate. Specifically, it employs an "incremental search algorithm." Based on the deviation between the current output (performance) and the target interval (preset threshold), the input (load parameters) are dynamically adjusted to ensure the subject remains in an optimal state of stimulation.

[0025] The implementation process is as follows: The index generated in step S103 Align with the preset "comfort zone threshold range" in real time.

[0026] like If the load remains above the upper limit of the threshold, it is determined that the current task is no longer challenging enough for the subject, and the "load increase logic" is triggered. Within the boundary set by S101, the difficulty weight in the parameter combination is increased by a preset step size (such as shortening the signal duration or increasing the resistance switching frequency). like If the value falls below the lower threshold, the subject is deemed to have cognitive overload or physiological fatigue, thus triggering the "load release logic" to reduce the parameter value and lower the difficulty.

[0027] Through this real-time incremental adjustment, the system generates the "next training parameter combination," ensuring that the gradient of training intensity always matches the subject's real-time state curve.

[0028] Step S105: Determine whether the training termination condition has been met. If not, in the next training iteration, apply the next training parameter combination to generate a new current training task, and return to step S103 based on the new current training task.

[0029] This step maintains the system's continuous evolutionary cycle, addressing the issues of logical stability and safe exit mechanisms. Specifically, it's based on "state machine loop control logic." Through continuous parameter iteration, the subject's cognitive-motor abilities are strengthened within the loop.

[0030] The implementation process is as follows: Detect global variables (such as cumulative training time, total number of tasks completed, or physiological limit alarm signals of the subjects). If the termination threshold is not reached, use the "next training parameter combination" generated by S104 as input to re-drive the task generation engine of S102.

[0031] At this point, the subject will face a completely new task with an optimized difficulty structure. Through this infinite recursive loop of "perception-evaluation-adjustment-re-execution", the system guides the subject's nervous system to continuously migrate to a higher level of steady state until the termination condition is met (such as reaching a preset 20-minute efficient training duration), and finally safely and smoothly stops hardware output.

[0032] In an exemplary embodiment, step S101 specifically includes: In the first execution phase, the participants' inhibitory control scores were obtained. Standardized scores of working memory and cognitive flexibility were used to determine the basic behavioral level of the subjects. Traditional grading methods rely solely on single motor performance indicators and fail to recognize the heterogeneity of children's cognitive development (such as the "high motor ability - low inhibitory control" type). This embodiment is based on the core definition of executive function in neuropsychology. An individual's motor performance is not driven by a single muscle force, but is the result of the synergistic effect of three core factors controlled by the frontal lobe of the brain: inhibitory control (IC), working memory (WM), and cognitive flexibility (CF).

[0033] The implementation process is as follows: First, the raw performance data of the subjects in the pre-set assessment stage is obtained through a data interface. This data is transformed into standardized scores in three dimensions through a standard normal distribution transformation (Z-scroe transformation): Inhibition control score (IC): characterizing the subject's ability to eliminate irrelevant stimuli and maintain goal-oriented behavior in a highly interfering environment; Working memory score (WM): characterizing the subject's ability to store and dynamically process multiple instructions in a short period of time during movement; Cognitive flexibility score (CF): characterizing the subject's ability to flexibly adjust action plans according to instantaneous switching of training rules. These scores constitute the subject's "executive function baseline data," providing high signal-to-noise ratio feature input for subsequent personalized grading.

[0034] The second execution phase involves retrieving a preset weight template. And use equation (1) to calculate the ability level score. : (1) This step achieves a non-linear fusion from "multi-dimensional cognitive features" to "single ability scalars." It addresses the issue of inconsistent contributions of various cognitive abilities to performance under different training orientations (e.g., resting tasks emphasizing focus vs. dynamic adversarial tasks emphasizing agility). Specifically, it employs a "task-oriented weight compensation model." By adjusting the weight template, the calculated score objectively reflects the subject's "functional fit" in a specific training scenario.

[0035] The implementation process is as follows: Based on the current training subject attributes, retrieve the matching weight template from the internal memory. For example, in the "Multi-Rule Obstacle Avoidance Training" subject, the system will automatically adjust the height... The proportion of (cognitive flexibility). Subsequently, the driving computation unit executes formula (1): The calculated ability level score As the sole criterion for measuring a subject's integrated cognitive-motor potential, its physical meaning represents the comprehensive bandwidth an individual can process complex interactive information flows per unit of time.

[0036] In the third execution phase, the ability level score is compared with the preset tiered threshold range to determine the corresponding initial training level. Based on the initial training level, the preset training intensity parameter boundary table is retrieved to determine the first value boundary corresponding to the sensory dimension, the second value boundary corresponding to the logical dimension, the third value boundary corresponding to the spatial dimension, the fourth value boundary corresponding to the rule switching frequency dimension per unit time, and the fifth value boundary corresponding to the time pressure compression ratio dimension. Based on the first, second, third, fourth, and fifth value boundaries, determine the training intensity parameter boundaries corresponding to the initial training level.

[0037] This embodiment transforms a continuous distribution of abilities into discrete training stages that are controllable in an engineering manner. Specifically, it is based on the logic of "stepped stress load." Human cognitive tolerance is divided into different energy levels, ensuring that each level of task provides appropriate training incentives without overloading the nervous system.

[0038] The implementation process is as follows: The value is within the preset stepped threshold range (e.g.: , , Compare them. If If the subject falls within a certain range, its corresponding initial training level is immediately established (e.g., Level 1 - beginner level, or Level 4 - competitive level). Once this level is established, the "benchmark difficulty level" for this subject's current training is locked at the logical level.

[0039] In an exemplary embodiment, step S102, within the training intensity parameter boundary, determines the current training parameter combination, specifically including: Obtain an initial complexity target value that matches the initial training level. ; Equation (2) is the relationship between the pre-defined extended complexity index and the parameter values ​​corresponding to each dimension: (2) in For the parameter values ​​of the sensory dimension, For parameter values ​​of logical dimensions, For the parameter value of spatial dimension, This refers to the parameter value representing the rule switching frequency per unit of time. The parameter value represents the time-pressure compression ratio dimension; and These are the preset weight coefficients corresponding to the rule switching frequency dimension and the time pressure compression ratio dimension, respectively; based on the initial training level, and under the premise of satisfying the constraints, a set of values ​​is determined to make the calculation result of equation (2) approach the initial complexity target value. The set of parameter values ​​corresponding to each dimension is used as the current training parameter combination; This embodiment addresses the "blindness" in the task generation process by providing the algorithm with a definite convergence endpoint. Specifically, it is based on "load steady-state control." Each training level corresponds to a complexity range within which a subject can maintain a steady-state response, determined by setting... This allows the system to search for the most suitable configuration combination in a multi-dimensional parameter space.

[0040] The implementation process is as follows: Based on the initial training level determined in step S101 (e.g., level two), the corresponding target value is retrieved from the preset level / complexity mapping table. This value represents the "total cognitive work" that the subject should undertake at the current level.

[0041] The purpose of the constraint mechanism is to address the "cognitive collapse" problem caused by an excessive number of dimensions in the learning process, and to implement a tiered scaffolding education strategy. Specifically, it follows the principle of separating the "proximal developmental zone." Beginners should not face variables from all dimensions simultaneously. The system limits the number of variables through physical locking, allowing learners to focus on mastering the core competencies at their current level.

[0042] If the initial training level is Level 1 (Logic Enlightenment Period): the system will forcibly lock the system. (single-modal) (Fixed point) (No switching) and (No compression). At this point, only the logical dimension is fine-tuned within the second value boundary. In a physical sense, this is equivalent to letting a child learn the pure logical mapping of "seeing A and doing action A" in a fixed position without interference.

[0043] If the initial training level is Level 2 (Perceptual Space Expansion Period): The system unlocks sensory dimensions. ( ) and spatial dimension Simultaneous adjustment within the first, second, and third boundaries. This allows subjects to begin processing multimodal information across a spatial span, but does not yet involve high-frequency rule switching or extreme time compression.

[0044] If the initial training level is Level 3 (dynamic transition to advanced stage): the system further unlocks rule switching dimensions. ( At this point, the subjects must cope with sudden changes in logical rules during the exercise (e.g., suddenly changing from "chasing the red ball" to "avoiding the red ball"), thus training the flexibility of their nervous system.

[0045] If the initial training level is Level 4 (Full-Dimensional Practical Pressure Period): The system is fully unlocked, and time pressure compression is introduced. By maximally reducing the response window, subjects were forced to complete multidimensional task processing under high-pressure, near-real-world conditions.

[0046] Within a subspace satisfying specific level constraints, a heuristic search algorithm is used to select a set of candidate parameter values ​​from five boundary tables. Substitute this set of values ​​into equation (2) for trial calculation. If the calculated... and The deviation is within a preset threshold (e.g.) If the parameter set is locked as the "current training parameter combination", the actuator (motor, display screen) will convert it into a specific physical task.

[0047] Furthermore, step S102 generates the current training task based on the current combination of training parameters, specifically including: according to Define the combination of visual, auditory, and tactile stimuli in the current training task to configure the first subtask corresponding to the sensory information intake load; This embodiment addresses the insufficient neural arousal caused by single-modal stimulation by training the brain's ability to process multi-source information concurrently. Specifically, it is based on the "multisensory integration" effect. When processing mixed signals, the brain requires additional electrophysiological resources for filtering and integration. (Based on parameters...) The value is used to retrieve the corresponding stimulus modality from the hardware library. For example, if... Simultaneously, three types of physical signals are generated: a flashing RGB light array in the visual dimension, a specific frequency sine wave audio in the auditory dimension, and a high-frequency vibration of the handle in the tactile dimension. These three types of signals constitute the "background perception field" of the task, and the subject must extract the core target information from the interference.

[0048] according to Set the number of mapping rules between stimulus signals and response actions in the current training task to configure the second subtask corresponding to the information processing load; This embodiment addresses the problem of insufficient cognitive mapping depth by training subjects' conditional logic judgment ability. Specifically, it is based on "Logical Mapping Cost (Mapping Cst)". The more rules there are in the transition from perception to action, the longer the synaptic transmission delay in the neural pathway, and the greater the energy loss. The implementation process is as follows: Based on parameters... The numerical value is used to define the logical judgment equation for the current task. For example, if... The system sets the following rules: ① Pressing key A is required when seeing a red light; ② Pressing key B is required when seeing a blue light; ③ No operation is allowed when seeing a red light accompanied by a high-frequency sound. These logical mappings are loaded into temporary memory, and the system determines in real time whether the subject's action response conforms to the current Boolean logic during task execution.

[0049] according to Set the number or distribution density of target points in the physical space for the response action in the current training task, so as to configure the third subtask corresponding to the spatial decision and action control load; This embodiment addresses the problem of low efficiency in spatial exploration by training the coordinated control of proprioception and visual space. Specifically, it is based on Fitts' Law and spatial entropy theory. The more target points there are or the denser their distribution, the higher the difficulty index of the movement. The higher the value, the better. The implementation process is as follows: based on the parameters... The value is used to calculate the coordinates of the target point within the physical workspace. If Larger, the instruction execution mechanism has a radius of High-density deployment within the hemispherical area A specific interaction point (or the drive motor moves to a specific location) (Each physical phase). This forces the subject to make large-scale or fine-grained physical displacements, significantly increasing the precision requirements for motion control.

[0050] according to In the current training task, a corresponding number of rule-transformation actions are arranged within a unit of training time to configure the fourth subtask corresponding to the cognitive flexibility load; This embodiment addresses the problem of cognitive perseveratin by training subjects' ability to quickly switch between sets of mental tasks. Specifically, it is based on "switch cost." When old rules are suppressed and new rules are activated, subjects experience a momentary lag in response. The implementation process is as follows: Within a unit time period of the current task (e.g., 60 seconds), according to parameters... The instruction to change the rule for random insertion of values. For example, if During training, the logical mapping in the second subtask will be suddenly altered five times (e.g., red light originally corresponds to the A key, but suddenly becomes the B key). This physical-level rule oscillation forces the subjects to constantly reconstruct their behavioral patterns.

[0051] When obtaining the subject's baseline response in the previous stable phase and combined , using Calculate the upper limit of the reaction time for the current task. ,Will Set a time limit for subjects to make effective response actions to configure the fifth subtask corresponding to the inhibitory control load; This embodiment addresses impulsive behavior in subjects under time pressure, training them to precisely inhibit behavior under extreme conditions. Specifically, it is based on the "Time Pressure Effect." By shortening the allowed reaction time window, the decision threshold of the nervous system is forced to shift forward. The implementation process is as follows: First, the baseline reaction time of the subject is extracted from the baseline test in S101. (For example: the average reaction time is 500 ms). Then, using the formula... Perform the calculation. If (That is, a 20% compression) sets the current task's duration to 400ms. The threshold for determining the "validity of an action" is that only actions completed within 400 ms after the signal is emitted are considered valid. This millisecond-level time compression places extremely strong dynamic pressure on the subject's inhibitory control system.

[0052] Generate the current training task based on the first to fifth subtasks.

[0053] The instruction sets corresponding to the five sub-tasks mentioned above are time-aligned and packaged to generate a complete data packet for the "current training task". This data packet is then sent to the hardware peripherals via the driving engine, placing the subject in a comprehensive physical interaction field integrating "sensory interference, logical transformation, spatial displacement, regular oscillation, and time pressure". In this embodiment, a high-speed bus polls the sensor group in real time (including but not limited to: a high-precision photoelectric encoder built into the servo motor, a timestamp counter of the interactive touch array, and tension / compression sensors for monitoring limb pressure). Every physical movement of the subject generates a level transition at the sensor layer. Using "spatiotemporal synchronous sampling technology", the moment the signal is emitted from the interactive interface (excitation point) is aligned with the moment the subject produces a physical response (response point). By calculating the time difference between the two points and the matching degree of the signal logic, the ambiguous biological movements are transformed into discrete response features.

[0054] In an exemplary embodiment, step S103 specifically includes: acquiring the response features of the subject when performing the current training task; wherein, the response features include at least the rule execution accuracy. Average reaction time and impulsive reaction error rate ;right Normalization was performed to obtain the effective reaction time. The quantitative performance index of the subjects' real-time performance was calculated using equation (3). : (3) in, The preset evaluation weights, and satisfy the following conditions: .

[0055] Within the current task cycle (e.g., a 60-second training run), the following three key physical characteristics are statistically analyzed and calculated in real time: Rule execution accuracy (ACC): This is the ratio of the total number of actions performed by the subject that conform to the mapping logic to the total number of instructions. Physically, it represents the subject's ability to maintain stable cognitive mapping under multidimensional load.

[0056] Mean Reaction Time (RT): The average time difference (in milliseconds) for all correct responses. Physically, it represents the overall rate of neural conduction and decision-making pathways in the subject.

[0057] Impulsive Response Error Rate (Err): This is a statistical measure of the proportion of incorrect triggers that occur when a subject performs an inappropriate action or during a rule-switching moment. Its physical meaning represents the frequency of missed alerts in a subject's inhibitory control function.

[0058] This ensures that the faster the response, the better. The greater the positive contribution; This item, acting as a "penalty item," ensures that test takers cannot achieve high scores by blindly increasing speed (impulsive response). It is set according to the current training phase (e.g., a "consolidation period" emphasizing accuracy or a "reinforcement period" emphasizing explosive power). The specific proportions. For example, during the enhancement period, adjust... The percentage. Calculated. The value is pushed to the decision buffer in real time. At this time, The size directly reflects the subject's current five-dimensional workload ( "Functional redundancy" under )

[0059] like A value close to 1 indicates that the physical layer means the subject completed a complex multidimensional interaction flawlessly within a very short time, and the system determines that they are in their "comfort zone" and ready to increase their workload. If If the value is close to 0 or negative, it means that the subject has experienced a large number of impulsive errors or delayed actions. The system determines that the subject is in the "collapse zone" and the load release mechanism needs to be activated immediately.

[0060] Among them, for Normalization was performed to obtain the effective reaction time. Specifically, this includes: obtaining the extreme range of reaction time for subjects within the initial training level or within a preset sliding window in the current training cycle. ; Calculate the effective reaction time using equation (4) : (4) in, This is the minimum reaction time. This represents the maximum reaction time. A "linear mapping normalization" method is employed. The millisecond value, which has absolute dimensions, is transformed into a value within a range relative to the individual baseline or rank limit by scaling. A dimensionless scalar within the interval ensures the mathematical consistency of the evaluation system. The implementation process is as follows: retrieve the maximum allowable time under the current level. (Calculated from S102). Using a linear scaling function, the real-time observed... Mapped to effective reaction time For example, when Approaching hour, Approaching 1; when In a very short time, Approaching 0. This processing allows the "speed" feature to be weighted at the same order of magnitude as the "accuracy" feature.

[0061] In an exemplary embodiment, step S104 specifically includes: responding to the subject continuously performing Quantitative performance index corresponding to the current training task All are greater than or equal to the preset threshold. This triggers a load escalation strategy; when executing the load escalation strategy, incremental adjustments are made to the current training parameter combination in the following order; The purpose of the above steps is to address the problem of unwarranted workload escalation caused by occasional bursts of performance from test subjects (such as "extraordinary performance"). Specifically, this is based on the principle of "Performance Consistency." Only when the performance index... In continuous Only when the subject maintains a high level in the task does the system determine that the subject has fully internalized the skill in the current dimension and entered the true "skill automation" stage.

[0062] The implementation process is as follows: A sliding counter is built in. After each training task is completed, the calculated... Values ​​compared to the preset "excellent threshold" (like Compare them. If The counter increments by 1; if there is any interruption... The counter is immediately reset to zero. When the counter value reaches... (like When this happens, a "Level-up" command is sent to the task generation engine.

[0063] First priority: Maintain Unchanged, increased or until Reaching the upper limit of the second value boundary, or Reaching the upper limit of the third value boundary; Specifically, it follows the training principle of "quality first, quantity later." In the sensory input environment ( Without changing the underlying logic, prioritize increasing logical complexity. or spatial displacement difficulty The implementation process is as follows: First, check the parameters. and Maintaining sensory dimensions Under constant conditions, incrementally increase the number of mapping rules. (e.g., changing from "choose 1 out of 2" to "choose 1 out of 3") or increasing the distribution density of target points. At this point, the subjects process more complex logical judgments or a wider range of physical movements in a familiar sensory environment. Until... Reaching the upper limit of the second value boundary, and The upper limit of the third value boundary has been reached. This indicates that the subject's logical / spatial processing ability in a "static environment" has reached the limit of this level.

[0064] Second priority: Responding to and If all values ​​reach the corresponding upper limit of the value boundary, then increase. Or increase until Reaching the upper limit of the fourth value boundary, or Reaching the upper limit of the fifth value boundary; Specifically, it is based on the theory of "reaction-time compression and task switching". In the logical architecture ( After the initial priority parameter reaches its maximum, a time-based game theory is introduced. The implementation process is as follows: Once the first priority parameter reaches its maximum, the rule switching frequency is adjusted. Or time pressure compression ratio For example, to Increase the switching frequency from 2 times per minute to 4 times per minute, or... Increase from 0.1 to 0.15 (i.e., further shorten the allowable reaction time) Subjects must increase their neural firing rate and inhibitory control precision in familiar complex logic tasks to cope with faster pace and more varied rules. Until... and It reaches the upper limit of the fourth and fifth value boundaries.

[0065] Third priority: Responding to and If all values ​​reach the corresponding upper limit of the value boundary, then increase. until The upper limit of the first value boundary is reached.

[0066] This is the highest level of difficulty challenge. When the subjects are already "at ease" with the current sensory environment, increasing the variety of information input sources ( This reconstructs the entire perceptual field. The implementation process is as follows: after all four parameters mentioned above have reached their maximum values, the final step is to execute... Incremental operations (such as increasing from "bimodal" to "trimodal", introducing tactile or background noise interference). This is equivalent to changing the "background" of the task. An increase in complexity can produce a "complexity spillover effect," which often leads to a decrease in the subject's... The value drops in the next round of tasks, forcing the system to re-enter the first-priority fine-tuning loop, thus achieving a spiral upward.

[0067] The training termination condition in step S104 includes at least one of the following: (1) the cumulative training duration of the current training cycle reaches a preset time limit threshold; (2) the total number of executions of the current training tasks reaches a preset task capacity threshold; (3) in response to Within a preset observation period, or within a preset continuous period During this training task, the fatigue level remained consistently below the preset warning threshold. The subject was determined to be in a state of fatigue.

[0068] In one exemplary embodiment, a cognitive-motor adaptive training system is provided, the training system including a task design controller, such as... Figure 2 As shown, the task design controller includes: Baseline Level Assessment Module 11: Used to determine the corresponding initial training level based on the subject's basic action level, and to determine the training intensity parameter boundary corresponding to the initial training level; Multidimensional task generation module 12: Used to determine the current combination of training parameters within the training intensity parameter boundary, and generate the current training task based on the current combination of training parameters; Interactive Level Quantization Module 13: Used to acquire the response features of the subject when performing the current training task, and to convert the response features into a quantitative performance index to evaluate the subject's real-time performance. ; Adaptive Iteration Module 14: It is used to compare the quantitative performance index with the preset threshold, obtain the comparison result, and make incremental adjustments to the current training parameter combination based on the comparison result to obtain the next training parameter combination; and to determine whether the training termination condition has been met. If not, it triggers the multi-dimensional task generation module to generate a new training task by applying the next training parameter combination until the training termination condition is met.

[0069] Currently, in the field of children's rehabilitation training and balance development, electric training equipment typically uses force sensors or position sensors to monitor the subject's movement status. This existing technology mainly relies on a feedback loop at the software logic level: by collecting changes in force or displacement deviations at the load end, it uses control algorithms such as PID to adjust the motor's output torque, thereby attempting to counteract the subject's tendency to lose balance. When the system detects that the subject is about to fall or lose balance, it usually achieves protection by increasing the motor's reverse resistance or cutting off the power.

[0070] However, in real-world children's balance training scenarios, the aforementioned existing technologies suffer from a significant "perception-execution delay" problem. When a child is at the critical point of instability, the tremors (Tremr) produced by their limbs are often high-frequency physical fluctuations with extremely high frequency and weak amplitude. The frequency of these fluctuations usually exceeds the sampling bandwidth of conventional force sensors and the response limit of software control loops (software loop cycles are typically on the order of 10ms-100ms). Heat conduction has thermal inertia, while mechanical motion feedback has "logical inertia." By the time the software layer completes signal filtering, algorithm calculation, and outputs instructions, the subject has often already undergone irreversible displacement, causing the protective action to lag behind the occurrence of instability. This deficiency, where the perception response speed lags behind the physical stall speed, makes it difficult for existing methods to provide sufficient rigid support at the instant instability occurs.

[0071] To address the aforementioned technical problems, the cognitive motion adaptive training system provided in this embodiment further includes an instantaneous damping control subsystem applied to adaptive training, such as... Figure 3 As shown, the subsystem includes a multi-dimensional task generation module, a power source motor, a load execution terminal, a power conversion circuit, and an instantaneous damping control device.

[0072] Specifically, the subsystem integrates a high-speed feedback loop at the physical hardware layer. The power source motor (such as a permanent magnet synchronous motor (PMSM) or a brushless DC motor (BLDC)) is connected to the load actuator (such as a pedal or handlebar) via a transmission mechanism. The power conversion circuit (Inverter), consisting of a phase bridge arm composed of six power transistors (MSFETs or IGBTs), is responsible for driving the motor. The instantaneous damping control device acts as the "overriding control core," directly intervening in the drive input of the power conversion circuit. The various hardware pathways achieve dynamic coordination by constructing a closed loop of "electromagnetic induction - hardware comparison - hard short circuit."

[0073] This subsystem bypasses the traditional software algorithm processing cycle by analyzing the micro-current fluctuations of the power source motor in real time, directly identifying the instability signs of the subject at the hardware level and triggering physical damping. Furthermore, the system also includes electric training equipment and an instantaneous damping control device that works in conjunction with the electric training equipment; the electric training equipment includes: a power source motor, a load actuator driven by the power source motor, and a power conversion circuit connected to the power source motor; The multidimensional task generation module is also used to: calculate the task complexity corresponding to the current training task and output the task complexity to the instantaneous damping control device; The instantaneous damping control device is equipped with: The first component, the threshold monitoring pathway, is used to configure a dynamic envelope threshold based on task complexity; wherein, the dynamic envelope threshold is associated with the subject's expected balance control under the current training task; First, obtain the task complexity corresponding to the subject's current training task. Logically, task complexity is a comprehensive indicator that encompasses the instability of the balance fulcrum, the frequency of interference load, and the accuracy requirements of the training objective.

[0074] The threshold monitoring path has a pre-set "damping sensitivity mapping table," which retrieves a matching damping sensitivity coefficient based on the current task complexity. This coefficient determines the subsystem's "tolerance bandwidth" for instability symptoms: based on the damping sensitivity coefficient, the subsystem dynamically adjusts the torque-speed gradient mapping model of the power source motor. This model, in a physical sense, establishes the dynamic response boundary of electromagnetic torque generated by rotor disturbance, thereby deriving the maximum allowable current fluctuation slope under the current task (i.e., The threshold monitoring path maps the maximum current fluctuation slope calculated by the software layer to a corresponding reference voltage signal via a digital-to-analog converter (DAC). This signal is configured as the "dynamic envelope threshold" input to the hardware comparison circuit. This process achieves "physical locking of control intent." By transforming the abstract task difficulty into a specific level reference value, subsequent anomaly determination no longer requires software instruction parsing, providing a priori criteria for microsecond-level hardware triggering.

[0075] The second component, the signal acquisition pathway, is used to capture in real time the induced ripple signal generated by the power source motor during the interaction with the subject; wherein, the induced ripple signal characterizes the electromagnetic feedback generated by the subject's tremor caused by unexpected instability to the power source motor; During the interaction between the subject and the training equipment, any minute mechanical disturbance at the load actuation end (such as the balance pedal) will be synchronously reflected in the electromagnetic field of the power source motor. The signal acquisition path captures this feedback through the following physical path: by acquiring the original phase current signal containing the driving fundamental component and the induced characteristic component through high-precision current sampling elements (such as constantan wire resistors or magnetic balance current sensors) deployed on the DC side bus or three-phase branch of the power conversion circuit (inverter bridge).

[0076] Because the fundamental frequency of the PWM driving motor is low and regular, while the limb tremors caused by the subject's instability (which typically generate a large number of high-frequency harmonic clusters in the 5Hz-50Hz frequency range) are highly transient and random, the signal acquisition path uses a high-pass filter circuit to remove the fundamental driving component from the original current, thereby completely preserving the induced ripple signal generated by the subject's limb instability through magnetic field coupling.

[0077] The induced ripple signal is fed into a differentiating circuit. The differential processing, in its physical sense, extracts the rate of change of tremor kinetic energy, significantly amplifying the "acceleration" characteristic at the moment of instability. The final generated electrical differential characteristic has an amplitude proportional to the severity of the subject's instability. This step achieves "self-decoupling of characteristic signals" at the physical layer, simplifying the complex electromechanical interaction signal into a pure differential voltage reflecting the intensity of instability, greatly improving the signal-to-noise ratio.

[0078] The third component, the logic drive path, is used to compare the differential characteristics of the induced ripple signal with the dynamic envelope threshold in real time. When the differential characteristics exceed the dynamic envelope threshold, the conduction state of the power conversion circuit is forcibly switched to form a closed short-circuit loop in the stator winding of the power source motor, so as to generate instantaneous damping at the load execution end.

[0079] Using a hardware comparator in the threshold monitoring path, the differential characteristic is input to the non-inverting input of the comparator in real time, and the dynamic envelope threshold is input to the inverting input. When the subject is in a steady state, the differential characteristic of the ripple generated by the micro-movement of their limbs is always lower than the dynamic envelope threshold, the comparator outputs a low level, and the subsystem maintains normal driving state. Once the subject experiences unexpected nonlinear tremors due to loss of balance, the differential characteristic level will break through the dynamic envelope threshold in a very short time (microseconds). The comparator instantly flips to output a high-level signal, which is directly fed to the logic driving path as an "instability confirmation" signal. Since the comparison logic is completed directly at the analog circuit hardware layer, the "perception blind spot" caused by the ADC sampling and quantization cycle and software code polling in conventional subsystems is completely eliminated.

[0080] Furthermore, in response to the differential characteristic exceeding the dynamic envelope threshold, the logic drive path immediately executes an "overriding intervention" action, forcibly switching the operating state of the power conversion circuit (H-bridge / three-phase bridge): the hard trigger signal output by the logic drive path has the highest physical priority and can instantly shield the controller's PWM modulation signal. The subsystem forcibly shuts down the upper-side MSFETs of all phase bridge arms in the power conversion circuit and fully turns on all lower-side MSFETs.

[0081] At this moment, the three-phase stator windings of the motor form a low-impedance closed short-circuit loop through the fully conducting lower bridge arm power transistors. According to Lenz's law, due to the inertial displacement caused by the subject's instability, the rotor induces a strong reverse electromotive force in the short-circuit windings, generating a huge short-circuit induced current in the loop. The reverse electromagnetic torque generated by this current produces a strong instantaneous damping at the load actuation end. This damping torque is not generated by external energy, but is "converted in place" from the kinetic energy of the subject's instability. Physically, the load actuation end seems to instantly switch from a "flexible sliding dynamic" to a "rigid locked state," providing the subject's limbs with a sense of support as if stepping on solid ground, effectively preventing falls or tipping.

[0082] The subsystem also enhances safety by pre-setting an initial physical viscosity. Based on the baseline evaluation results of the subjects, the control device sets the initial PWM frequency and extremely low duty cycle of the motor, generating a weak static self-locking torque. This "physical viscosity" acts as a safety base, ensuring that the dynamic envelope threshold is configured above a stable reference current, effectively filtering electromagnetic shock noise at the moment of motor startup, and ensuring the sensitivity and robustness of the subsystem during dynamic operation.

[0083] Furthermore, after damping protection is triggered, the subsystem does not simply shut down. Instead, it extracts the peak voltage and frequency density of the induced ripple at the moment of triggering to quantitatively assess the risk level of this instability. If the risk level is found to exceed the safety warning threshold (determining that the subject's current ability is insufficient to handle the complexity), the subsystem automatically lowers the complexity parameter for the next task cycle and restarts the threshold monitoring pathway to execute the process. Through this closed loop of "hardware protection - software correction," the training load is adapted to the subject's immediate ability.

[0084] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described cognitive-motor adaptive training method.

[0085] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described cognitive-motor adaptive training method.

[0086] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the above-described cognitive-motor adaptive training method.

[0087] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile deterministic machine-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0088] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A cognitive-motor adaptive training method, characterized in that, The method is applied to a task design controller in a cognitive motion adaptive training system, and the method includes the following steps: Step S101: Determine the corresponding initial training level based on the subject's basic action level, and determine the training intensity parameter boundary corresponding to the initial training level; Step S102: Within the training intensity parameter boundary, determine the current training parameter combination, and generate the current training task based on the current training parameter combination; Step S103: Obtain the response features of the subject when performing the current training task, and convert the response features into a quantitative performance index to evaluate the subject's real-time performance. ; Step S104: Compare the quantitative performance index with a preset threshold to obtain the comparison result, and make incremental adjustments to the current training parameter combination based on the comparison result to obtain the next training parameter combination; Step S105: Determine whether the training termination condition has been met. If not, in the next training iteration, apply the next training parameter combination to generate a new current training task, and return to execute step S103 based on the new current training task.

2. The method according to claim 1, characterized in that, Step S101 specifically includes: Based on the obtained scores of the subjects corresponding to their inhibitory control... Working memory score Standardized score of cognitive flexibility To determine the basic level of action of the subjects; Retrieve preset weight template And use equation (1) to calculate the ability level score. : （1） The ability level score is compared with a preset tiered threshold range to determine the corresponding initial training level. The preset training intensity parameter boundary table is retrieved based on the initial training level to determine the first value boundary corresponding to the sensory dimension, the second value boundary corresponding to the logical dimension, the third value boundary corresponding to the spatial dimension, the fourth value boundary corresponding to the rule switching frequency dimension per unit time, and the fifth value boundary corresponding to the time pressure compression ratio dimension. The training intensity parameter boundary corresponding to the initial training level is determined based on the first value boundary, the second value boundary, the third value boundary, the fourth value boundary, and the fifth value boundary.

3. The method according to claim 2, characterized in that, In step S102, determining the current training parameter combination within the training intensity parameter boundary specifically includes: Obtain an initial complexity target value that matches the initial training level. ; Preset the relationship between the extended complexity index and the parameter values ​​corresponding to each dimension, as shown in equation (2): （2） in The parameter value for the sensory dimension. The parameter value for the logical dimension. The parameter value for the spatial dimension. Here are the parameter values ​​for the rule switching frequency dimension per unit time. The parameter value for the time-pressure compression ratio dimension; and These are the preset weighting coefficients corresponding to the rule switching frequency dimension and the time pressure compression ratio dimension, respectively. Based on the initial training level, and under the premise of satisfying the constraints, a set of values ​​is determined that makes the calculation result of equation (2) approach the initial complexity target value. The set of parameter values ​​corresponding to each dimension is used as the current training parameter combination; Specifically, the limiting constraints include: If the initial training level is the first level, then the following is limited: , , and Adjustment is only permitted within the second value boundary. ; If the initial training level is the second level, then the following is limited: and Allows adjustment of the value within the first value boundary, the second value boundary, and the third value boundary, respectively. , and ; If the initial training level is level three, then the following is limited: and Allows adjustment of the value within the first to fourth value boundaries respectively. , , and ; If the initial training level is the fourth level, then training is allowed within the range from the first value boundary to the fifth value boundary. , , And make adjustments to the combination.

4. The method according to claim 3, characterized in that, Step S102, which generates the current training task based on the current training parameter combination, specifically includes: according to The types of combinations of visual, auditory, and tactile stimulus signals in the current training task are set to configure the first subtask corresponding to the sensory information intake load. according to Set the number of mapping rules between stimulus signals and response actions in the current training task to configure the second subtask corresponding to the information processing load; according to Set the number or distribution density of target points in the physical space for the response action in the current training task, so as to configure the third subtask corresponding to the spatial decision and action control load; according to In the current training task, a corresponding number of rule transformation actions are arranged within a unit training time to configure the fourth subtask corresponding to the cognitive flexibility load; When obtaining the baseline response of the subject in the previous stable phase and combined , using Calculate the upper limit of the reaction time for the current task. ,Will A time limit is set for the subject to make an effective response action in order to configure the fifth subtask corresponding to the inhibitory control load; The current training task is generated based on the first to the fifth subtasks.

5. The method according to claim 4, characterized in that, Step S103 specifically includes: Obtain the response features of the subject when performing the current training task; wherein, the response features include at least the rule execution accuracy. Average reaction time and impulsive reaction error rate ; right Normalization was performed to obtain the effective reaction time. ; The quantitative performance index of the subject's real-time performance is calculated using equation (3). : （3） in, The preset evaluation weights, and satisfy the following conditions: .

6. The method according to claim 5, characterized in that, The pair Normalization was performed to obtain the effective reaction time. Specifically, it includes: Obtain the extreme range of the subject's reaction time within the initial training level or within a preset sliding window of the current training cycle. The effective reaction time is calculated using equation (4). : （4） in, This is the minimum reaction time. This represents the maximum reaction time.

7. The method according to claim 5, characterized in that, Step S104 specifically includes: In response to the subject continuously performing The quantitative performance index corresponding to the current training task is described below. All are greater than or equal to the preset threshold. This triggers a load escalation strategy. When implementing the load boosting strategy, incremental adjustments are made to the current training parameter combination in the following order: First priority: Maintain Unchanged, increased or until Reaching the upper limit of the second value boundary, or Reaching the upper limit of the third value boundary; Second priority: Responding to and If all values ​​reach the corresponding upper limit of the value boundary, then increase. Or increase until Reaching the upper limit of the fourth value boundary, or Reaching the upper limit of the fifth value boundary; Third priority: Responding to and If all values ​​reach the corresponding upper limit of the value boundary, then increase. until The upper limit of the first value boundary is reached.

8. The method according to claim 7, characterized in that, The training termination condition in step S104 includes at least one of the following: (1) The cumulative training time of the current training cycle has reached the preset time limit threshold; (2) The total number of executions of the current training task reaches the preset task capacity threshold; (3) Responding to Within a preset observation period, or within a preset continuous period During this training task, the fatigue level remained consistently below the preset warning threshold. The subject was determined to be in a state of fatigue.

9. A cognitive-motor adaptive training system, characterized in that, The training system includes a task design controller, which includes: Baseline level assessment module: used to determine the corresponding initial training level based on the subject's basic action level, and to determine the training intensity parameter boundary corresponding to the initial training level; Multidimensional task generation module: used to determine the current combination of training parameters within the boundary of the training intensity parameters, and generate the current training task based on the current combination of training parameters; Interactive Level Quantization Module: Used to acquire the response features of the subject when performing the current training task, and convert the response features into a quantitative performance index to evaluate the subject's real-time performance. ; The adaptive iteration module is used to compare the quantitative performance index with a preset threshold to obtain a comparison result, and to incrementally adjust the current training parameter combination according to the comparison result to obtain the next training parameter combination; and to determine whether the training termination condition has been reached. If not, the multi-dimensional task generation module is triggered to generate a new training task using the next training parameter combination until the training termination condition is reached.

10. The system according to claim 9, characterized in that, The system also includes electric training equipment and an instantaneous damping control device that works in conjunction with the electric training equipment. The electric training equipment includes: a power source motor, a load actuator driven by the power source motor, and a power conversion circuit connected to the power source motor. The multidimensional task generation module is also used to: calculate the task complexity corresponding to the current training task, and output the task complexity to the instantaneous damping control device; The instantaneous damping control device is equipped with: Threshold monitoring pathway: used to configure a dynamic envelope threshold based on the task complexity; wherein the dynamic envelope threshold is associated with the subject's expected balance control under the current training task; Signal acquisition path: used to capture in real time the induced ripple signal generated by the power source motor during the interaction with the subject; wherein, the induced ripple signal characterizes the electromagnetic feedback generated by the subject's tremor caused by unexpected instability to the power source motor; Logic drive path: used to compare the differential characteristics of the induced ripple signal with the dynamic envelope threshold in real time, and when the differential characteristics exceed the dynamic envelope threshold, to force the switching of the conduction state of the power conversion circuit, so that the stator winding of the power source motor forms a closed short circuit loop, thereby generating instantaneous damping at the load execution end.

Images