A method and system for dynamically generating training plans and scoring movement quality with smart dumbbells.

Abstract

This application relates to the field of intelligent fitness equipment and sports training control technology, and provides a method and system for dynamically generating training plans and scoring movement quality using intelligent dumbbells. The method utilizes the accelerometer, angular velocity sensor, and pressure sensor built into the intelligent dumbbell to collect real-time movement data and personalized parameters input by the user during training. The movement data and personalized parameters are input into a preset training plan generation algorithm module to generate a phased training plan that includes movement type, training duration for each set, rest intervals, and weight adjustment strategies. A movement quality scoring model is constructed, which pre-stores standard movement templates. These templates include standard trajectory curves, speed threshold ranges, amplitude baselines, and force symmetry parameters corresponding to different training movements. The movement data is then matched and analyzed against the standard movement templates in multiple dimensions, and the score is dynamically adjusted based on the user's current fatigue level.

Description

Technical Field

[0001] This application relates to the field of intelligent fitness equipment and sports training control technology, and in particular to a method and system for dynamically generating training plans and scoring movement quality using intelligent dumbbells. Background Technology

[0002] With the increasing popularity of smart fitness equipment, users' demand for personalized and scientific training is growing. Traditional fitness training, particularly dumbbell training plans, is typically user-defined or relies on fixed templates, presenting two major technical bottlenecks: First, training plans lack dynamic adaptability. Existing solutions often generate fixed plans based on static user fitness parameters (such as weight and age), failing to perceive real-time changes in user fitness, force stability, and movement accuracy during training. This leads to a mismatch between training intensity and the user's actual ability, resulting in poor training outcomes or the risk of injury. Second, movement quality assessment methods are limited. Existing technologies often rely on single sensors (such as accelerometers) to collect movement trajectories, only able to determine whether the movement is completed. They lack multi-dimensional analysis of core performance indicators such as trajectory consistency, speed uniformity, and force symmetry, and do not dynamically adjust scores based on the user's real-time fatigue level. This makes it difficult to accurately assess movement quality and adjust training plans accordingly. Furthermore, existing smart dumbbell systems generally lack a closed-loop control mechanism of "training plan generation - movement quality feedback - dynamic plan adjustment," resulting in a lack of effective linkage between training plans and actual training effects, hindering adaptive optimization of training programs.

[0003] To address the aforementioned issues, existing technologies offer some improvements, such as using sensors to collect motion data for action recognition or generating basic training plans based on user fitness parameters. However, none of these methods deeply integrate real-time motion data, personalized parameters, historical training adaptability, and fatigue assessment, nor do they construct a complete technical system that integrates "dynamic training plan generation, multi-dimensional motion quality scoring, and closed-loop feedback adjustment." Therefore, providing an intelligent method that can dynamically adjust training plans based on the user's real-time status and accurately assess motion quality through multi-dimensional data has become a pressing technical problem in this field. Summary of the Invention

[0004] This application provides a method and system for dynamically generating training plans and scoring movement quality using smart dumbbells. It aims to address the problems in existing technologies where some improvements are made, such as using sensors to collect motion data for movement recognition or generating basic training plans based on user fitness parameters. However, these methods do not deeply integrate real-time movement data, personalized parameters, historical training adaptability, and fatigue status assessment, nor do they construct a complete technical system that integrates "dynamic training plan generation, multi-dimensional movement quality scoring, and closed-loop feedback adjustment."

[0005] In a first aspect, embodiments of this application provide a method for dynamically generating training plans and scoring movement quality using smart dumbbells; the method includes:

[0006] The smart dumbbells use built-in accelerometers, angular velocity sensors, and pressure sensors to collect real-time motion data from users during training. This motion data includes dumbbell movement trajectory, movement speed, range of motion, grip pressure, and force angle. The system also acquires personalized parameters input by the user, including age, weight, body fat percentage, muscle strength level, training goals, and historical training records.

[0007] The motion data and personalized parameters are input into the preset training plan generation algorithm module to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals. The weight adjustment strategy is also adaptively adjusted according to the user's real-time force stability and historical strength growth curve.

[0008] A motion quality scoring model is constructed, in which standard motion templates are pre-stored. These templates include standard trajectory curves, speed threshold ranges, amplitude benchmarks, and force symmetry parameters corresponding to different training movements. Motion data is matched and analyzed against the standard motion templates in multiple dimensions. Motion compliance scores are calculated from four dimensions: motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry. The user's current fatigue level is obtained based on continuous training duration, historical heart rate data, and grip pressure fluctuations. The scores are then dynamically adjusted based on these fatigue parameters.

[0009] In some embodiments, after dynamically correcting the score by combining the user's current fatigue level parameters, the method further includes: during the training process, feeding back the action quality score results to the training plan generation algorithm module in real time; if the action standard score is lower than a preset threshold multiple times in a row, triggering a dynamic adjustment mechanism for the training plan, automatically reducing the training weight or shortening the training time of a single set until the action standard score rises back to a reasonable range, thus forming a closed-loop optimization control between the training plan and the action quality.

[0010] In some embodiments, the real-time acquisition of user movement data during training via the built-in accelerometer, angular velocity sensor, and pressure sensor of the smart dumbbell includes: real-time acquisition of the dumbbell's acceleration data in three-dimensional space using the accelerometer; real-time acquisition of the dumbbell's angular velocity data around three rotation axes using the angular velocity sensor; and real-time acquisition of the user's grip pressure data when holding the dumbbell using the pressure sensor; calculating the dumbbell's trajectory, speed, and amplitude of motion based on the acceleration and angular velocity data; and calculating the user's force angle when exerting force based on the acceleration, angular velocity, and grip pressure data.

[0011] In some embodiments, the step of dynamically generating a phased training plan based on user motion data, personalized parameters, historical training fitness, and training goals, including motion type, training duration per set, rest interval, and weight adjustment strategy, includes: the training plan generation algorithm module first analyzes the motion completion rate, muscle recovery cycle, and training effect data in the user's historical training records to determine historical training fitness; combining the user's current motion data reflecting real-time physical condition, training goals in personalized parameters, and muscle strength level, it matches suitable motion types from a preset motion library; it sets the training duration per set based on the fatigue tolerance time of similar motions in the user's historical training; it sets the rest interval between sets based on the recovery law of exercise physiology; and it uses the weight adjustment strategy as a variable parameter of the training plan, so that the generated phased training plan includes a weight change strategy that is dynamically adjusted with the training stage.

[0012] In some embodiments, the adaptive weight adjustment strategy based on the user's real-time force stability and historical strength growth curve includes: judging the user's real-time force stability by analyzing the fluctuation range of real-time collected grip pressure data and force angle data; if the fluctuation range is within a preset stability threshold, the force is determined to be stable; otherwise, the force is determined to be unstable; retrieving weight increase data of the same movement type from the user's historical training records to form a historical strength growth curve; when the force is stable and the historical strength growth curve shows an upward trend, gradually increasing the training weight according to a preset incremental rule; when the force is unstable or the historical strength growth curve shows a plateau, maintaining the current training weight or reducing the training weight according to a preset reduction rule.

[0013] In some embodiments, constructing the motion quality scoring model includes: collecting standard motion data from professional trainees performing different training movements through sports biomechanics experiments; the standard motion data includes standard motion trajectory, standard motion speed range, standard motion amplitude, and left-right force symmetry benchmark values; classifying the standard motion data according to motion type to generate standard motion templates containing corresponding standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters; and storing the standard motion templates in the database of the motion quality scoring model.

[0014] In some embodiments, the multi-dimensional matching analysis of motion data with a standard motion template, calculating a motion standardization score from four dimensions—motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry—includes: calculating the coordinate point fitting degree between the real-time collected dumbbell motion trajectory and the standard trajectory curve in the standard motion template to obtain a motion trajectory conformity score; comparing the real-time motion speed with the speed threshold range in the standard motion template, calculating the percentage of time the speed value is within the threshold range as a speed uniformity score; comparing the real-time motion amplitude with the amplitude benchmark value in the standard motion template, calculating the percentage of movements that reach or exceed the amplitude benchmark value as an amplitude compliance rate score; acquiring grip pressure data and force angle data of the smart dumbbells on both sides, calculating the difference between the data on both sides and the deviation from the symmetry benchmark value as a left-right force symmetry score; and calculating the total motion standardization score by combining the scores of the four dimensions.

[0015] In some embodiments, obtaining the user's current fatigue parameters based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude includes: calculating the uninterrupted continuous training duration during the user's current training phase, and marking fatigue risk when the continuous training duration exceeds a preset fatigue warning duration; retrieving heart rate data from the user's historical training records, analyzing the deviation between the current training heart rate and the historical average heart rate to determine the degree of cardiopulmonary fatigue; calculating the fluctuation amplitude of real-time grip pressure data, and determining a decline in muscle control ability when the fluctuation amplitude exceeds a preset stability threshold; and determining the user's current fatigue parameters by comprehensively considering the continuous training duration, heart rate deviation amplitude, and grip pressure fluctuation amplitude through preset fatigue assessment rules.

[0016] In some embodiments, the dynamic correction of the score based on the user's current fatigue level parameters includes: setting a correction coefficient for the action standardization score based on the user's current fatigue level parameters; when the fatigue level parameters indicate that the user is in a state of mild fatigue, the score thresholds for the consistency of the action trajectory and the uniformity of speed are adaptively relaxed; when the fatigue level parameters indicate that the user is in a state of moderate or severe fatigue, in addition to relaxing the score thresholds, an additional judgment on the safety of the action trajectory is added, and if a force deviation that may lead to sports injury occurs, the action standardization score is directly reduced; through the correspondence between the fatigue level parameters and the correction rules, the dynamic adjustment of the action quality score is realized.

[0017] Secondly, this application provides a dynamic training plan generation and movement quality scoring system for smart dumbbells, applied to smart dumbbells, the system comprising:

[0018] The parameter acquisition unit is used to collect the user's movement data in real time during training through the accelerometer, angular velocity sensor and pressure sensor built into the smart dumbbell. The movement data includes the dumbbell's movement trajectory, movement speed, movement range, grip pressure and force angle. It also acquires personalized parameters input by the user, including age, weight, body fat percentage, muscle strength level, training goals and historical training records.

[0019] The strategy acquisition unit is used to input motion data and personalized parameters into the preset training plan generation algorithm module, so as to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals, and adaptively adjust the weight adjustment strategy according to the user's real-time force stability and historical strength growth curve.

[0020] A dynamic correction unit is used to construct a motion quality scoring model. The scoring model pre-stores standard motion templates, which include standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters corresponding to different training movements. The unit performs multi-dimensional matching analysis between the motion data and the standard motion templates, and calculates the motion standardization score from four dimensions: motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry. The unit obtains the user's current fatigue parameter based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude, and dynamically corrects the score based on the user's current fatigue parameter.

[0021] This application provides a method and system for dynamically generating training plans and scoring movement quality using smart dumbbells. It acquires muscle response signals under static pressure using multiple sensors, including pressure, vibration, and displacement sensors, and extracts multi-dimensional feature parameters such as pressure peak, deformation recovery time, and vibration attenuation coefficient to accurately quantify muscle hardness and elasticity properties, overcoming the limitations of traditional single-signal detection. A machine learning-based intelligent matching model is constructed, integrating muscle hardness features, user body data, and striking preferences to output personalized parameter combinations. During striking, parameters are dynamically adjusted through real-time feedback signals, forming a closed-loop control of "detection-modeling-execution-feedback," achieving for the first time real-time intelligent adaptation of smart dumbbell parameters to muscle state. Through pre-striking verification, abnormal state monitoring, and parameter protection mechanisms, striking safety is ensured while allowing for personalized adjustments, overcoming the shortcomings of existing technologies where parameter settings are "one-size-fits-all" or rely on subjective judgment.

[0022] This application generates striking parameters tailored to individual differences based on multi-dimensional muscle stiffness characteristics and user body data, avoiding the blindness of traditional manual adjustments and improving muscle relaxation effects. By optimizing parameters in real time through feedback signals during the striking process, the striking force and frequency are always matched to the real-time state of the muscles (e.g., automatically increasing amplitude when muscles are stiff after exercise, and intelligently reducing force when fatigued muscles are over-pressed), significantly improving user comfort and safety. Users do not need to manually adjust parameters frequently; the device automatically completes detection, modeling, and dynamic control, lowering the barrier to entry and adapting to the differentiated needs of different exercise scenarios (e.g., pre-exercise activation, post-exercise recovery, and daily relaxation). Through pre-striking verification, abnormal vibration monitoring, and parameter protection mechanisms, the risk of muscle injury due to improper parameters is effectively avoided, broadening the device's applicable user group (e.g., the elderly, exercise beginners).

[0023] In summary, this invention fills the technological gap in the field of integrated control of "precise detection-intelligent matching-dynamic feedback" in existing intelligent dumbbells, and provides a brand-new approach for the development of intelligent muscle relaxation devices.

[0024] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0025] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a schematic flowchart illustrating the steps of a method for dynamically generating training plans and scoring movement quality using smart dumbbells, provided in one embodiment of this application.

[0027] Figure 2 This is a schematic diagram of the structure of a smart dumbbell provided in one embodiment of this application;

[0028] Figure 3 This is a schematic block diagram of a training plan dynamic generation and motion quality scoring system for a smart dumbbell provided in one embodiment of this application;

[0029] Figure 4 This is a schematic block diagram of the structure of an intelligent dumbbell provided in one embodiment of this application.

[0030] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Detailed Implementation

[0031] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0033] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0034] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0035] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0036] Smart dumbbells, as devices that use high-frequency vibrations to strike muscles to relieve fatigue and promote blood circulation, have been widely used in sports rehabilitation, fitness relaxation, and other scenarios. The striking parameters of existing smart dumbbells (such as frequency, force, and amplitude) typically rely on manual adjustment by the user or operate based on fixed preset modes (such as "relaxation mode" or "deep striking mode"). However, these adjustment methods have significant drawbacks:

[0037] Lack of personalized adaptation: There are significant differences in muscle stiffness and basic body data (such as age, weight, and muscle mass) among different users, and the muscle state of the same user changes dynamically in different scenarios (such as before / after exercise, static / fatigue state). In the current technology, smart dumbbells cannot accurately sense real-time muscle stiffness and rely solely on the user's subjective experience to adjust parameters, which can easily lead to poor hitting effect (such as insufficient force to achieve deep relaxation, or excessive frequency causing muscle damage).

[0038] Limited detection methods: Traditional methods only obtain limited data through pressure sensors or simple contact detection, without integrating multi-dimensional signals such as pressure, vibration, and displacement to comprehensively characterize muscle hardness, elasticity, and viscosity characteristics, resulting in insufficient detection accuracy.

[0039] Lack of dynamic feedback mechanism: Existing smart dumbbells cannot collect feedback signals of muscle state changes in real time during the hitting process (such as electromyographic signals and vibration attenuation characteristics), making it difficult to dynamically adjust parameters according to the real-time response of the muscles. This results in a disconnect between parameters and muscle state during the hitting process, making it difficult to balance safety and comfort.

[0040] Although there are a few existing technologies that involve sensor-assisted adjustment, they all remain at the level of simple processing of a single signal (such as pressure), failing to form a complete closed loop of "multi-dimensional detection - intelligent modeling - dynamic matching". For example, some devices only control the upper limit of striking force through pressure sensors, without combining vibration, displacement and other signals to quantify muscle stiffness characteristics; or they rely on preset empirical formulas to match parameters, without building personalized models through machine learning, and thus cannot adapt to the complex differences in human muscles.

[0041] To resolve the above issues, please refer to... Figure 1 This application provides a method for dynamically generating training plans and scoring movement quality using smart dumbbells, applicable to, for example... Figure 2 The smart dumbbell shown is an example. It should also be noted that all information involved in the method provided in this application was extracted with the authorization of the relevant user and in accordance with relevant regulations, and will not infringe on user privacy.

[0042] The provided method for dynamically generating training plans and scoring movement quality using smart dumbbells includes steps S101 to S103. Details are as follows:

[0043] Step S101. The user's movement data during training is collected in real time through the accelerometer, angular velocity sensor and pressure sensor built into the smart dumbbell. The movement data includes the dumbbell's movement trajectory, movement speed, movement range, grip pressure and force angle. Personalized parameters input by the user are obtained. The personalized parameters include age, weight, body fat percentage, muscle strength level, training goals and historical training records.

[0044] Specifically, this step involves using multiple types of sensors built into the smart dumbbell to collect user training motion data in real time and obtain personalized basic information about the user, providing a data foundation for subsequent training plan generation and motion evaluation.

[0045] Motion data acquisition utilizes accelerometers and angular velocity sensors (such as MEMS inertial measurement units) to monitor the dumbbell's three-dimensional spatial trajectory (X / Y / Z axis displacement), motion speed (tangential velocity, angular velocity), and force angle (changes in joint motion angle) in real time. Grip pressure (contact force between the palm and the dumbbell) and force symmetry (pressure difference between the left and right hands) are collected using pressure sensors (such as piezoresistive or capacitive sensors).

[0046] Personalized parameters are obtained through static data actively input by the user: age, weight, body fat percentage, muscle strength level (e.g., through grip strength tests or annotation of historical training data), and training goals (e.g., muscle gain, fat loss, rehabilitation training). Historical training records include past training plans, exercise completion rates, fatigue recovery data, injury history, etc., and are stored locally on the device or in a cloud account.

[0047] Sensor data acquisition utilizes accelerometers / angular velocity sensors to sample at frequencies above 100Hz. Data is then fused using a Kalman filter algorithm to eliminate motion noise and generate a smooth motion trajectory curve. Pressure sensors are integrated into the dumbbell grip surface, employing an array layout (e.g., three pressure nodes on each side) to calculate the distribution and dynamic changes in grip pressure in real time.

[0048] User input interaction involves entering basic data through the dumbbell's companion app or the device's touchscreen. Historical training records are automatically synchronized from the user's account, and manual editing or integration with third-party health data platforms (such as Apple Health) is supported.

[0049] Data preprocessing normalizes and aligns the raw sensor data over time to generate motion sequences (such as time series arrays containing [trajectory coordinates, velocity, pressure, angle]) with timestamps.

[0050] Breaking through the limitations of traditional single pressure detection, this technology integrates acceleration, angular velocity, and pressure sensors to comprehensively capture movement details (such as trajectory deviation and uneven force application), providing a data foundation for accurate analysis. By combining user static body data (age, weight), dynamic training goals (rehabilitation / muscle building), and historical training habits, it avoids "one-size-fits-all" parameter settings and addresses the lack of individualized adaptation in existing technologies. Real-time collected grip pressure and force angle can indirectly reflect real-time muscle tension (e.g., a sudden increase in pressure may indicate muscle fatigue), providing real-time signals for subsequent fatigue assessment and program adjustments.

[0051] Step S102. Input the motion data and personalized parameters into the preset training plan generation algorithm module to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals. The weight adjustment strategy is adaptively adjusted according to the user's real-time force stability and historical strength growth curve.

[0052] Specifically, this step uses a training plan generation algorithm module to integrate real-time user motion data, personalized parameters, historical training results, and training goals to generate a dynamically adjustable training plan, and optimizes the weight adjustment strategy based on force stability and strength growth trends.

[0053] Core Algorithm Module: Input Parameters: Movement data (trajectory, speed, pressure), personalized parameters (age, muscle strength level), historical training fitness (completion rate of past plans, muscle recovery speed), training goals (e.g., "deep muscle stimulation" corresponding to a high-frequency striking strategy). Output Content: Phased plans include movement type (e.g., bicep curls, shoulder press), training duration per set (e.g., 30 seconds / set), rest interval (e.g., 60 seconds), and weight adjustment strategy (dynamically adjusting dumbbell resistance or vibration frequency based on real-time strength).

[0054] The adaptive adjustment mechanism determines the current strength state based on the user's real-time force stability (such as trajectory fluctuation amplitude and pressure uniformity). If the stability of force exertion decreases by 20% for three consecutive sets, the weight or striking frequency of the next set is automatically reduced. Combining historical strength growth curves (such as the strength improvement rate over the past 4 weeks), the machine learning model predicts the current optimal load to avoid overtraining or undertraining.

[0055] The training plan generation algorithm employs a hierarchical planning strategy: First, a basic framework is determined based on the training objectives (e.g., rehabilitation training prioritizing low loads and long intervals). Then, a reinforcement learning algorithm (e.g., Q-learning) is used to optimize the movement types and duration allocation in conjunction with historical training fitness. Weight adjustment strategy: The dumbbells have built-in adjustable resistance modules (e.g., electromagnetic damping or spring structures). The algorithm dynamically calculates the resistance coefficient based on real-time pressure data and historical strength levels (e.g., if the safe load limit corresponding to the current strength level is 15kg, it is initially set to 80%, i.e., 12kg, and then gradually adjusted based on strength stability).

[0056] Historical training fitness is calculated by defining fitness indicators: completion rate (number of planned sets completed / preset number of sets), recovery index (heart rate recovery speed 24 hours after training), and injury risk coefficient (based on historical movement trajectory deviation data), and generating a comprehensive fitness score (0-100 points) through weighted averaging.

[0057] Real-time feedback and adjustment: After each training session, the algorithm module automatically analyzes the current action data. If it detects an abnormal force angle (such as exceeding the standard value ±15°) or excessive pressure fluctuation, it triggers the weight adjustment logic (such as reducing resistance by 5%) and records it as historical training data.

[0058] Breaking away from the limitations of traditional fixed preset modes, this technology integrates real-time motion data and historical training habits to generate personalized training plans, addressing the issue of parameter adjustments relying on subjective experience in existing technologies. It dynamically adjusts weight based on force stability and historical strength growth, avoiding insufficient force (poor results) or overload (muscle injury), thus improving training safety and efficiency. Through a closed loop of "data acquisition - plan generation - real-time adjustment," the training plan is dynamically optimized according to changes in the user's muscle state (such as post-exercise fatigue), achieving a complete "detection - modeling - matching" process and overcoming the lack of dynamic feedback in existing technologies.

[0059] Step S103. Construct a motion quality scoring model. The scoring model pre-stores standard motion templates, which include standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters corresponding to different training movements. Perform multi-dimensional matching analysis between the motion data and the standard motion templates, and calculate the motion standardization score from four dimensions: motion trajectory matching degree, speed uniformity, amplitude compliance rate, and left-right force symmetry. Obtain the user's current fatigue parameter based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude, and dynamically correct the score based on the user's current fatigue parameter.

[0060] Specifically, this step constructs a motion quality scoring model, which uses multi-dimensional matching of standard motion templates and combines real-time user fatigue correction scores to achieve motion standardization assessment and fatigue state perception.

[0061] The standard movement template uses preset standard trajectory curves (based on data from exercise physiology experts or generated from a large number of sample training) for different training movements (such as bicep curls and presses), speed threshold ranges (such as concentric contraction speed of 2-4° / s), amplitude benchmark values ​​(lower limit of joint range of motion, such as elbow flexion of 90°), and force symmetry parameters (grip pressure difference between left and right hands ≤10%).

[0062] The scoring dimensions and calculations include: Movement trajectory conformity: Calculate the average Euclidean distance between the real-time trajectory and the standard curve, and score after normalization (0-100 points). Speed ​​uniformity: Statistically calculate the percentage of time the movement speed deviates from the threshold range, and inversely map this to the score (e.g., exceeding the limit for ≤10% earns 90 points or more). Range of motion compliance rate: Calculate the percentage of times the joint range of motion reaches the benchmark value in each set of movements (e.g., 8 out of 10 bicep curls achieve the benchmark, earning 80 points). Left-right force symmetry: Calculate the percentage of left-right pressure difference using pressure sensor data; deduct points if it exceeds the threshold (e.g., 15%).

[0063] Fatigue parameter correction includes: fatigue index: combining continuous training duration (fatigue warning triggered when exceeding 45 minutes), historical heart rate data (obtained through an external heart rate belt or the device's built-in PPG sensor; heart rate variability ≥20% indicates fatigue), and grip pressure fluctuation amplitude (pressure standard deviation exceeding 30% of the baseline value is considered unstable). A fatigue coefficient (0-1) is generated through a fuzzy logic algorithm, and the score is dynamically corrected (e.g., when fatigued, the trajectory consistency tolerance is allowed to be relaxed by 5%).

[0064] The standard template construction involves demonstrating standard movements by sports rehabilitation experts, collecting high-precision trajectory data (error ≤ 1mm), and combining this with sports biomechanical models (such as the Newton-Euler equation) to calculate speed and amplitude benchmark values, thus forming a multi-movement template library.

[0065] The real-time scoring algorithm uses the Dynamic Time Warping (DTW) algorithm to match real-time action sequences with standard templates, solving the alignment problem of different users' action rhythm differences; and calculates scores for each dimension in real time through a sliding window (such as a 1-second window).

[0066] Fatigue fusion correction establishes a fatigue assessment model: inputting continuous training duration (T), heart rate variability (HRV), and stress fluctuation coefficient (σ), and outputting a correction factor (e.g., when fatigue = 0.8, the scoring baseline value is reduced by 10%), avoiding excessive demands on the standardization of movements under fatigue conditions and balancing safety and training effectiveness.

[0067] By employing multi-dimensional quantitative scoring (trajectory, speed, amplitude, symmetry), this approach addresses the reliance on subjective judgment in traditional motion accuracy assessments, helping users correct incorrect postures in real time (such as muscle compensation due to trajectory deviation). It dynamically adjusts scores by combining physiological signals (heart rate) and motion signals (pressure fluctuations), avoiding injury risks caused by forcibly pursuing perfect motion accuracy under fatigue, thus improving training comfort and safety. Overcoming the limitations of single pressure detection, it integrates multi-modal data such as trajectory, speed, and pressure to comprehensively characterize muscle movement status (e.g., vibration attenuation characteristics can be indirectly reflected through trajectory fluctuations), overcoming the shortcomings of single detection methods in existing technologies.

[0068] Through the closed-loop design of S101-S103, this method realizes a complete process of "multi-dimensional data acquisition → intelligent model planning → dynamic feedback adjustment," specifically addressing the problems of existing smart dumbbells lacking personalized adaptation, single detection, and lack of feedback. Compared with traditional solutions, its core innovation lies in: constructing a three-dimensional perception system of muscle state and movement details through the collaboration of acceleration, angular velocity, and pressure sensors; generating adaptive training plans by combining user static vital signs, dynamic goals, and historical training data, rather than relying on fixed preset modes; and achieving dynamic matching of parameters and muscle state during training through movement scoring and fatigue correction, balancing effectiveness, safety, and comfort. This solution, through technological innovation, promotes the upgrade of smart dumbbells from "manual adjustment tools" to "intelligent adaptive training systems," significantly improving the scientific rigor and safety of sports rehabilitation and fitness relaxation.

[0069] In some embodiments, after dynamically correcting the score by combining the user's current fatigue level parameters, the method further includes: during the training process, feeding back the action quality score results to the training plan generation algorithm module in real time; if the action standard score is lower than a preset threshold multiple times in a row, triggering a dynamic adjustment mechanism for the training plan, automatically reducing the training weight or shortening the training time of a single set until the action standard score rises back to a reasonable range, thus forming a closed-loop optimization control between the training plan and the action quality.

[0070] After dynamic correction of the movement quality score, a closed-loop feedback mechanism between the training plan and movement quality is established. The movement standardization score is fed back to the training plan generation module in real time. When the score falls below the preset threshold for multiple consecutive times, the plan is automatically adjusted to reduce the training load (weight / duration) until the score recovers, forming a closed-loop control of "scoring-adjustment-optimization".

[0071] The feedback trigger condition is set by setting a threshold for the action standard score (e.g., 70 points). When the score of a single group is lower than the threshold for three consecutive times, it is determined that the current training load exceeds the user's ability.

[0072] The adjustment strategies include: Weight adjustment: reducing the weight for the next training session according to a preset reduction rule (e.g., 10% of the current weight), or switching to a low vibration frequency mode (for smart dumbbell hitting scenarios). Duration adjustment: shortening the duration of a single set from 40 seconds to 30 seconds, while extending the rest interval (e.g., from 60 seconds to 90 seconds) to give muscles more recovery time. Closed-loop mechanism: After each adjustment, the system continues to monitor the score of subsequent movements. If the score rises above 80 points for two consecutive sets, the adjustment stops; if it remains below the threshold, the load reduction strategy is triggered again (at least 50% of the initial load).

[0073] This system avoids movement distortion due to excessive load, reduces the risk of muscle strain and joint injury, and is especially suitable for safety protection for beginners or those in a fatigued state. Real-time score feedback dynamically calibrates the training plan, solving the problem of disconnect between plan and actual execution in traditional methods, and improving the adaptability of plan implementation. It forms a complete closed loop of "data collection - quality assessment - plan adjustment," enabling the training system to have self-optimization capabilities, rather than simply outputting a plan unidirectionally.

[0074] In some embodiments, the real-time acquisition of user movement data during training via the built-in accelerometer, angular velocity sensor, and pressure sensor of the smart dumbbell includes: real-time acquisition of the dumbbell's acceleration data in three-dimensional space using the accelerometer; real-time acquisition of the dumbbell's angular velocity data around three rotation axes using the angular velocity sensor; and real-time acquisition of the user's grip pressure data when holding the dumbbell using the pressure sensor; calculating the dumbbell's trajectory, speed, and amplitude of motion based on the acceleration and angular velocity data; and calculating the user's force angle when exerting force based on the acceleration, angular velocity, and grip pressure data.

[0075] By clarifying the specific implementation method of multi-sensor data acquisition: the accelerometer collects three-dimensional acceleration, the angular velocity sensor collects three-axis angular velocity, and the pressure sensor collects grip pressure; by fusing acceleration and angular velocity data, the motion trajectory, velocity, and amplitude are calculated, and the force angle is calculated by combining the three.

[0076] Sensor data acquisition: An accelerometer collects X / Y / Z-axis acceleration data at a frequency of 200Hz, including the gravitational acceleration component. An angular velocity sensor collects the angular velocity (° / s) around the X / Y / Z axes to calculate the dumbbell's rotational posture. A pressure sensor (such as an FSR thin-film sensor) is embedded in the grip, with two contacts on each side, to collect grip force (N) in real time.

[0077] Data processing algorithm: Motion trajectory: By fusing acceleration and angular velocity data through complementary filtering to eliminate gravity interference, the position coordinates (x, y, z) of the dumbbell in three-dimensional space are obtained, and the trajectory curve is generated by integrating the velocity data. Force angle: Based on the dumbbell posture matrix (obtained by integrating the angular velocity) and the grip pressure distribution, the angle between the force direction and the coronal / sagittal plane of the body (such as the angle between the forearm and upper arm during a bicep curl) is calculated.

[0078] By fusing three-axis sensors, millimeter-level trajectory accuracy and angle measurement (error ≤2°) are achieved, providing richer motion details (such as trajectory offset direction and force tilt angle) compared to traditional single pressure sensors. This provides high-precision input for subsequent motion scoring (trajectory consistency, force symmetry) and planning adjustments (weight strategy), solving the problem of coarse parameter adjustment caused by single detection methods.

[0079] In some embodiments, the step of dynamically generating a phased training plan based on user motion data, personalized parameters, historical training fitness, and training goals, including motion type, training duration per set, rest interval, and weight adjustment strategy, includes: the training plan generation algorithm module first analyzes the motion completion rate, muscle recovery cycle, and training effect data in the user's historical training records to determine historical training fitness; combining the user's current motion data reflecting real-time physical condition, training goals in personalized parameters, and muscle strength level, it matches suitable motion types from a preset motion library; it sets the training duration per set based on the fatigue tolerance time of similar motions in the user's historical training; it sets the rest interval between sets based on the recovery law of exercise physiology; and it uses the weight adjustment strategy as a variable parameter of the training plan, so that the generated phased training plan includes a weight change strategy that is dynamically adjusted with the training stage.

[0080] By refining the training plan generation logic: matching movement types based on historical training fitness (movement completion rate, recovery cycle, effect data), setting set durations according to fatigue tolerance time, setting rest intervals according to exercise physiology principles, and embedding weight strategies as dynamic parameters into the plan.

[0081] Historical training fitness analysis extracts "movement completion rate" (such as the percentage of sets completed for bicep curls in the past 10 training sessions), "recovery time" (time for muscle soreness to disappear after training), and "strength growth rate" (weekly maximum load increase) from historical records, and calculates fitness scores (0-100 points) using the analytic hierarchy process.

[0082] Movement type matching: The preset movement library contains 20+ basic movements (such as bicep curls, overhead presses, and lateral raises), each marked with a difficulty level (1-5) and the main muscle groups trained (such as biceps and deltoids). Based on the user's muscle strength level (e.g., beginner ≤10kg) and training goals (prioritizing compound movements for muscle growth), 3-5 suitable movements are selected.

[0083] Duration and rest interval settings: Set duration: Refer to the historical endurance duration of similar exercises (e.g., if the user has previously averaged 35 seconds per set for bicep curls, set the current set duration to 30 seconds, with a 5-second safety margin). Rest interval: Automatically matched based on the excess post-exercise oxygen consumption (EPOC) model, following the principle of "long rest intervals (90 seconds) for high-intensity exercises and short rest intervals (60 seconds) for low-intensity exercises".

[0084] To avoid the blind application of "generic templates," historical data should be used to accurately match the user's ability boundaries (e.g., beginners should avoid high-difficulty movements) to improve the feasibility of the training plan. Rest intervals should be set based on the principles of exercise physiology (e.g., to fully restore the ATP-CP system) to ensure training effectiveness (e.g., focusing on recovery between sets for muscle building) and reduce the risk of over-fatigue.

[0085] In some embodiments, the adaptive weight adjustment strategy based on the user's real-time force stability and historical strength growth curve includes: judging the user's real-time force stability by analyzing the fluctuation range of real-time collected grip pressure data and force angle data; if the fluctuation range is within a preset stability threshold, the force is determined to be stable; otherwise, the force is determined to be unstable; retrieving weight increase data of the same movement type from the user's historical training records to form a historical strength growth curve; when the force is stable and the historical strength growth curve shows an upward trend, gradually increasing the training weight according to a preset incremental rule; when the force is unstable or the historical strength growth curve shows a plateau, maintaining the current training weight or reducing the training weight according to a preset reduction rule.

[0086] The adaptive logic of the weight adjustment strategy is clearly defined: the stability of force exertion is judged by the fluctuation of grip pressure and force angle, and the load is dynamically adjusted in combination with the historical strength growth curve (weight increase data), which is divided into three modes: "stable increase", "unstable maintenance" and "plateau period adjustment".

[0087] Stability of force exertion is assessed by calculating the standard deviation of grip pressure (σ_pressure) and the range of force exertion angle fluctuation (Δangle). If σ_pressure < 5N and Δangle < 5°, the force exertion is considered stable; otherwise, it is considered unstable. Historical strength growth curve analysis: Extract weight data for the same movement over the past 8 weeks and fit a growth curve (e.g., a linear regression model). If the slope > 0 and persists for 4 weeks, it is considered an upward trend; if the slope ≤ 0 and persists for 2 weeks, it is considered a plateau.

[0088] The dynamic adjustment rules include: Stable + Increase: Increase weight in 5% increments (e.g., from 20kg to 21kg), with no more than 2 increases per week. Unstable / Plateau Period: Maintain the current weight for 2 weeks. If the weight remains unstable, reduce it by 10% to avoid forcibly increasing the load and causing movement errors.

[0089] Based on real-time ability growth, the load is dynamically increased, conforming to the principle of progressive muscle adaptation (a key principle for muscle growth), avoiding the blind application of fixed increments in traditional programs. Through real-time stability monitoring, the weight is automatically reduced before the user reaches exhaustion, preventing sports injuries caused by "loss of control in the last set of movements," making it especially suitable for home training scenarios without supervision.

[0090] In some embodiments, constructing the motion quality scoring model includes: collecting standard motion data from professional trainees performing different training movements through sports biomechanics experiments; the standard motion data includes standard motion trajectory, standard motion speed range, standard motion amplitude, and left-right force symmetry benchmark values; classifying the standard motion data according to motion type to generate standard motion templates containing corresponding standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters; and storing the standard motion templates in the database of the motion quality scoring model.

[0091] The process of constructing standardized movement templates involves collecting standard data (trajectory, speed, amplitude, symmetry) from professional trainees through sports biomechanics experiments, generating templates by movement type, and storing them in the scoring model database.

[0092] The data acquisition experiment recruited 10 professional athletes / rehabilitation therapists and used a high-precision motion capture system (such as Vicon) to collect standard movement data, while simultaneously recording sensor data (accuracy: trajectory ±1mm, angle ±1°). Fifty samples were collected for each movement, and after removing outliers, the average value was taken as the standard trajectory. The speed range was set to ±2σ (standard deviation) as a threshold (e.g., standard speed 2° / s, threshold range 1.6-2.4° / s).

[0093] Template generation and storage: Create folders according to the type of movement (such as "dumbbell curl" and "shoulder press"). Each template contains: standard trajectory curve (.csv format, time-coordinate sequence); speed threshold range (min_speed, max_speed); amplitude reference value (such as elbow flexion angle ≥90° during curl); force symmetry parameters (left-right pressure difference ≤15N).

[0094] The template data is sourced from professional sports biomechanics experiments, avoiding biases from subjective experience and ensuring that the scoring criteria conform to the laws of human movement (e.g., trajectory conformity directly reflects the correctness of muscle force application). The template library is expandable (supporting the import of user-defined movements), adapting to diverse training needs (rehabilitation training / strength training / functional training), and improving the system's versatility.

[0095] In some embodiments, the multi-dimensional matching analysis of motion data with a standard motion template, calculating a motion standardization score from four dimensions—motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry—includes: calculating the coordinate point fitting degree between the real-time collected dumbbell motion trajectory and the standard trajectory curve in the standard motion template to obtain a motion trajectory conformity score; comparing the real-time motion speed with the speed threshold range in the standard motion template, calculating the percentage of time the speed value is within the threshold range as a speed uniformity score; comparing the real-time motion amplitude with the amplitude benchmark value in the standard motion template, calculating the percentage of movements that reach or exceed the amplitude benchmark value as an amplitude compliance rate score; acquiring grip pressure data and force angle data of the smart dumbbells on both sides, calculating the difference between the data on both sides and the deviation from the symmetry benchmark value as a left-right force symmetry score; and calculating the total motion standardization score by combining the scores of the four dimensions.

[0096] The specific calculation methods for the four scoring dimensions are clarified: trajectory matching degree is calculated by coordinate fitting degree, velocity uniformity is the proportion of time within the threshold, amplitude compliance rate is the proportion of the number of times compliance is achieved, and symmetry score is based on the deviation of the left and right data from the benchmark value.

[0097] The trajectory matching score (0-100 points) is calculated by using the Dynamic Time Warping (DTW) algorithm to align the real-time trajectory with the time series of the standard template, calculating the Euclidean distance of each frame coordinate, taking the average value and normalizing it (the smaller the distance, the higher the score; the baseline distance corresponds to 70 points, and 1 point is added for every 1mm reduction).

[0098] Speed ​​uniformity (0-100 points) is determined by statistically analyzing the percentage of time that the speed is within the standard threshold range in a single set of movements (e.g., if the total duration is 30 seconds and the time to meet the standard is 25 seconds, the score = 25 / 30 × 100 ≈ 83 points).

[0099] Range of motion attainment rate (0-100 points) is determined by judging whether the range of motion of the joint is ≥ the benchmark value (e.g., ≥90° for bicep curl) after each exercise is completed. The score is calculated as: (Number of times that meet the benchmark / Total number of times) × 100 (e.g., 8 out of 10 times that meet the benchmark, the score is 80 points).

[0100] Left-right symmetry (0-100 points) is determined by calculating the ratio of the absolute value of the difference in grip pressure between the left and right hands to the symmetry reference value (e.g., reference difference is 15N, actual difference is 20N, score = 100 - [(20-15) / 15×10] ≈ 97 points, the greater the difference, the more points are deducted).

[0101] The system breaks down movement specifications into calculable objective indicators, avoiding the subjectivity of human scoring (such as errors in coach's visual judgment). Users can view shortcomings in each dimension in real time (such as trajectory deviation mainly in the X-axis direction). Through scores in four dimensions, movement problems are accurately located (such as uneven speed may be due to poor force application habits, and poor symmetry may indicate left and right muscle imbalance), guiding targeted corrections.

[0102] In some embodiments, obtaining the user's current fatigue parameters based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude includes: calculating the uninterrupted continuous training duration during the user's current training phase, and marking fatigue risk when the continuous training duration exceeds a preset fatigue warning duration; retrieving heart rate data from the user's historical training records, analyzing the deviation between the current training heart rate and the historical average heart rate to determine the degree of cardiopulmonary fatigue; calculating the fluctuation amplitude of real-time grip pressure data, and determining a decline in muscle control ability when the fluctuation amplitude exceeds a preset stability threshold; and determining the user's current fatigue parameters by comprehensively considering the continuous training duration, heart rate deviation amplitude, and grip pressure fluctuation amplitude through preset fatigue assessment rules.

[0103] The fatigue level parameter is calculated by defining the following methods: continuous training duration (risk is marked by exceeding the warning duration), heart rate deviation (cardiopulmonary fatigue), grip pressure fluctuation (muscle control ability), and fatigue level (0-1, the higher the value, the deeper the fatigue) is generated by preset rules.

[0104] Continuous training duration: Set a fatigue warning duration (e.g., 45 minutes). For every 10 minutes exceeding this duration, the fatigue level increases by 0.1 (e.g., 55 minutes → 0.2).

[0105] Heart rate deviation analysis: retrieve the historical average heart rate (HR_avg) and calculate the percentage difference between the current heart rate (HR_current) and HR_avg (ΔHR% = (HR_current-HR_avg) / HR_avg). When ΔHR% > 20%, increase fatigue by 0.1 for every 5% exceeding 20%.

[0106] Pressure fluctuation calculation is performed by calculating the standard deviation (σ_pressure) of grip pressure in a single set of actions and comparing it with the baseline value (σ0 in the absence of fatigue). When σ_pressure / σ0 > 1.5, fatigue degree increases by 0.1 for every 0.1 increase.

[0107] Comprehensive evaluation rule: Fatigue level = Duration weight × 0.4 + Heart rate weight × 0.3 + Stress fluctuation weight × 0.3. The three factors are fused through a fuzzy logic algorithm to output a continuous value between 0 and 1 (e.g., mild fatigue 0.3-0.5, moderate fatigue 0.5-0.7, severe fatigue > 0.7).

[0108] Combining physiological signals (heart rate) and action signals (stress fluctuations) provides a more comprehensive approach than simply monitoring heart rate (e.g., heart rate may be normal during resting fatigue, but muscle control may have already declined). It dynamically captures early signs of fatigue (e.g., muscles may begin to compensate when stress fluctuations increase), avoiding overtraining caused by relying solely on subjective feelings of fatigue in traditional programs (which may lead to user misjudgment).

[0109] In some embodiments, the dynamic correction of the score based on the user's current fatigue level parameters includes: setting a correction coefficient for the action standardization score based on the user's current fatigue level parameters; when the fatigue level parameters indicate that the user is in a state of mild fatigue, the score thresholds for the consistency of the action trajectory and the uniformity of speed are adaptively relaxed; when the fatigue level parameters indicate that the user is in a state of moderate or severe fatigue, in addition to relaxing the score thresholds, an additional judgment on the safety of the action trajectory is added, and if a force deviation that may lead to sports injury occurs, the action standardization score is directly reduced; through the correspondence between the fatigue level parameters and the correction rules, the dynamic adjustment of the action quality score is realized.

[0110] By clearly defining the dynamic correction rules for the score based on fatigue level: different correction strategies are set according to the degree of fatigue (mild / moderate / severe). For mild fatigue, the trajectory / speed threshold is relaxed, and for moderate and above fatigue, safety judgment is added. The score is adaptively adjusted through the correction coefficient.

[0111] Correction coefficient settings: Mild fatigue (0.3 ≤ fatigue level < 0.5): The trajectory matching threshold is relaxed by 5% (e.g., the passing score is reduced from 70 to 65), and the speed uniformity threshold is relaxed by 3% of the time percentage. Moderate fatigue (0.5 ≤ fatigue level < 0.7): In addition to relaxing the threshold, if joint hyperextension occurs in the trajectory (e.g., elbow extension angle > 185°), 10 points will be deducted per instance, triggering a safety warning. Severe fatigue (≥ 0.7): The current set of training will be forcibly terminated, a rest prompt will be given, and the score for proper movement will be calculated at 50% (to avoid continuing training in a fatigued state).

[0112] The dynamic adjustment mechanism automatically loads the corresponding correction rules based on the real-time fatigue parameters after each set of actions is completed, and generates a corrected score (e.g., the original trajectory score is 68 points, which is corrected to 72 points under mild fatigue, thus meeting the standard).

[0113] To avoid the frustration or forced persistence caused by "strict scoring even when fatigued," this approach balances training intensity and safety (e.g., allowing some movement deviation for mild fatigue to ensure training continuity). For moderate to severe fatigue, it adds safety assessments (e.g., hyperextension detection), directly deducting points and issuing warnings. It also provides protection against dangerous movements prone to occur under fatigue (e.g., joint locking), making it more intelligent than traditional methods.

[0114] Please see Figure 3 As shown, Figure 3This is a schematic diagram of the structure of the intelligent dumbbell training plan dynamic generation and movement quality scoring system 200 provided in this application embodiment. The intelligent dumbbell training plan dynamic generation and movement quality scoring system 200 is used to execute the steps of the intelligent dumbbell training plan dynamic generation and movement quality scoring methods shown in the above embodiments. The intelligent dumbbell training plan dynamic generation and movement quality scoring system 200 can be a single server or a server cluster, or it can be a terminal, such as a handheld terminal, a laptop computer, a wearable device, or a robot.

[0115] like Figure 3 As shown, the intelligent dumbbell training plan dynamic generation and movement quality scoring system 200 includes:

[0116] The parameter acquisition unit 201 is used to collect the user's movement data during training in real time through the accelerometer, angular velocity sensor and pressure sensor built into the smart dumbbell. The movement data includes the dumbbell's movement trajectory, movement speed, movement amplitude, grip pressure and force angle. It also acquires personalized parameters input by the user, including age, weight, body fat percentage, muscle strength level, training goals and historical training records.

[0117] The strategy acquisition unit 202 is used to input motion data and personalized parameters into the preset training plan generation algorithm module, so as to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals, and adaptively adjust the weight adjustment strategy according to the user's real-time force stability and historical strength growth curve.

[0118] The dynamic correction unit 203 is used to construct a motion quality scoring model. The scoring model pre-stores standard motion templates, which include standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters corresponding to different training movements. The motion data is matched and analyzed with the standard motion templates in multiple dimensions. The motion standard score is calculated from four dimensions: motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry. The user's current fatigue parameter is obtained based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude. The score is dynamically corrected based on the user's current fatigue parameter.

[0119] In some embodiments, after dynamically correcting the score by combining the user's current fatigue level parameters, the method further includes: during the training process, feeding back the action quality score results to the training plan generation algorithm module in real time; if the action standard score is lower than a preset threshold multiple times in a row, triggering a dynamic adjustment mechanism for the training plan, automatically reducing the training weight or shortening the training time of a single set until the action standard score rises back to a reasonable range, thus forming a closed-loop optimization control between the training plan and the action quality.

[0120] In some embodiments, the real-time acquisition of user movement data during training via the built-in accelerometer, angular velocity sensor, and pressure sensor of the smart dumbbell includes: real-time acquisition of the dumbbell's acceleration data in three-dimensional space using the accelerometer; real-time acquisition of the dumbbell's angular velocity data around three rotation axes using the angular velocity sensor; and real-time acquisition of the user's grip pressure data when holding the dumbbell using the pressure sensor; calculating the dumbbell's trajectory, speed, and amplitude of motion based on the acceleration and angular velocity data; and calculating the user's force angle when exerting force based on the acceleration, angular velocity, and grip pressure data.

[0121] In some embodiments, the step of dynamically generating a phased training plan based on user motion data, personalized parameters, historical training fitness, and training goals, including motion type, training duration per set, rest interval, and weight adjustment strategy, includes: the training plan generation algorithm module first analyzes the motion completion rate, muscle recovery cycle, and training effect data in the user's historical training records to determine historical training fitness; combining the user's current motion data reflecting real-time physical condition, training goals in personalized parameters, and muscle strength level, it matches suitable motion types from a preset motion library; it sets the training duration per set based on the fatigue tolerance time of similar motions in the user's historical training; it sets the rest interval between sets based on the recovery law of exercise physiology; and it uses the weight adjustment strategy as a variable parameter of the training plan, so that the generated phased training plan includes a weight change strategy that is dynamically adjusted with the training stage.

[0122] In some embodiments, the adaptive weight adjustment strategy based on the user's real-time force stability and historical strength growth curve includes: judging the user's real-time force stability by analyzing the fluctuation range of real-time collected grip pressure data and force angle data; if the fluctuation range is within a preset stability threshold, the force is determined to be stable; otherwise, the force is determined to be unstable; retrieving weight increase data of the same movement type from the user's historical training records to form a historical strength growth curve; when the force is stable and the historical strength growth curve shows an upward trend, gradually increasing the training weight according to a preset incremental rule; when the force is unstable or the historical strength growth curve shows a plateau, maintaining the current training weight or reducing the training weight according to a preset reduction rule.

[0123] In some embodiments, constructing the motion quality scoring model includes: collecting standard motion data from professional trainees performing different training movements through sports biomechanics experiments; the standard motion data includes standard motion trajectory, standard motion speed range, standard motion amplitude, and left-right force symmetry benchmark values; classifying the standard motion data according to motion type to generate standard motion templates containing corresponding standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters; and storing the standard motion templates in the database of the motion quality scoring model.

[0124] In some embodiments, the multi-dimensional matching analysis of motion data with a standard motion template, calculating a motion standardization score from four dimensions—motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry—includes: calculating the coordinate point fitting degree between the real-time collected dumbbell motion trajectory and the standard trajectory curve in the standard motion template to obtain a motion trajectory conformity score; comparing the real-time motion speed with the speed threshold range in the standard motion template, calculating the percentage of time the speed value is within the threshold range as a speed uniformity score; comparing the real-time motion amplitude with the amplitude benchmark value in the standard motion template, calculating the percentage of movements that reach or exceed the amplitude benchmark value as an amplitude compliance rate score; acquiring grip pressure data and force angle data of the smart dumbbells on both sides, calculating the difference between the data on both sides and the deviation from the symmetry benchmark value as a left-right force symmetry score; and calculating the total motion standardization score by combining the scores of the four dimensions.

[0125] In some embodiments, obtaining the user's current fatigue parameters based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude includes: calculating the uninterrupted continuous training duration during the user's current training phase, and marking fatigue risk when the continuous training duration exceeds a preset fatigue warning duration; retrieving heart rate data from the user's historical training records, analyzing the deviation between the current training heart rate and the historical average heart rate to determine the degree of cardiopulmonary fatigue; calculating the fluctuation amplitude of real-time grip pressure data, and determining a decline in muscle control ability when the fluctuation amplitude exceeds a preset stability threshold; and determining the user's current fatigue parameters by comprehensively considering the continuous training duration, heart rate deviation amplitude, and grip pressure fluctuation amplitude through preset fatigue assessment rules.

[0126] In some embodiments, the dynamic correction of the score based on the user's current fatigue level parameters includes: setting a correction coefficient for the action standardization score based on the user's current fatigue level parameters; when the fatigue level parameters indicate that the user is in a state of mild fatigue, the score thresholds for the consistency of the action trajectory and the uniformity of speed are adaptively relaxed; when the fatigue level parameters indicate that the user is in a state of moderate or severe fatigue, in addition to relaxing the score thresholds, an additional judgment on the safety of the action trajectory is added, and if a force deviation that may lead to sports injury occurs, the action standardization score is directly reduced; through the correspondence between the fatigue level parameters and the correction rules, the dynamic adjustment of the action quality score is realized.

[0127] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the above-described intelligent dumbbell training plan dynamic generation and movement quality scoring system and its modules can be referred to the corresponding content in the various embodiments of the above-described intelligent dumbbell training plan dynamic generation and movement quality scoring method, and will not be repeated here.

[0128] The aforementioned method for dynamically generating training plans and scoring movement quality using smart dumbbells can be implemented as a computer program, which can, for example... Figure 3 It runs on the device shown.

[0129] Please see Figure 4 , Figure 4 This is a schematic block diagram of the structure of a smart dumbbell provided in an embodiment of this application. The smart dumbbell includes a processor, a memory, and a network interface connected via a device bus, wherein the memory may include a storage medium and internal memory.

[0130] The storage medium can store operating devices and computer programs. The computer program includes program instructions that, when executed, cause the processor to perform any method for dynamically generating training plans and scoring the quality of movements using smart dumbbells.

[0131] The processor provides computing and control capabilities to support the operation of the entire smart dumbbell.

[0132] The internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to perform any method for dynamically generating training plans and scoring the quality of movements for smart dumbbells.

[0133] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 4The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the terminal to which the present application is applied. A specific smart dumbbell may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0134] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0135] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps:

[0136] The smart dumbbells use built-in accelerometers, angular velocity sensors, and pressure sensors to collect real-time motion data from users during training. This motion data includes dumbbell movement trajectory, movement speed, range of motion, grip pressure, and force angle. The system also acquires personalized parameters input by the user, including age, weight, body fat percentage, muscle strength level, training goals, and historical training records.

[0137] The motion data and personalized parameters are input into the preset training plan generation algorithm module to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals. The weight adjustment strategy is also adaptively adjusted according to the user's real-time force stability and historical strength growth curve.

[0138] A motion quality scoring model is constructed, in which standard motion templates are pre-stored. These templates include standard trajectory curves, speed threshold ranges, amplitude benchmarks, and force symmetry parameters corresponding to different training movements. Motion data is matched and analyzed against the standard motion templates in multiple dimensions. Motion compliance scores are calculated from four dimensions: motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry. The user's current fatigue level is obtained based on continuous training duration, historical heart rate data, and grip pressure fluctuations. The scores are then dynamically adjusted based on these fatigue parameters.

[0139] In some embodiments, after dynamically correcting the score by combining the user's current fatigue level parameters, the method further includes: during the training process, feeding back the action quality score results to the training plan generation algorithm module in real time; if the action standard score is lower than a preset threshold multiple times in a row, triggering a dynamic adjustment mechanism for the training plan, automatically reducing the training weight or shortening the training time of a single set until the action standard score rises back to a reasonable range, thus forming a closed-loop optimization control between the training plan and the action quality.

[0140] In some embodiments, the real-time acquisition of user movement data during training via the built-in accelerometer, angular velocity sensor, and pressure sensor of the smart dumbbell includes: real-time acquisition of the dumbbell's acceleration data in three-dimensional space using the accelerometer; real-time acquisition of the dumbbell's angular velocity data around three rotation axes using the angular velocity sensor; and real-time acquisition of the user's grip pressure data when holding the dumbbell using the pressure sensor; calculating the dumbbell's trajectory, speed, and amplitude of motion based on the acceleration and angular velocity data; and calculating the user's force angle when exerting force based on the acceleration, angular velocity, and grip pressure data.

[0141] In some embodiments, the step of dynamically generating a phased training plan based on user motion data, personalized parameters, historical training fitness, and training goals, including motion type, training duration per set, rest interval, and weight adjustment strategy, includes: the training plan generation algorithm module first analyzes the motion completion rate, muscle recovery cycle, and training effect data in the user's historical training records to determine historical training fitness; combining the user's current motion data reflecting real-time physical condition, training goals in personalized parameters, and muscle strength level, it matches suitable motion types from a preset motion library; it sets the training duration per set based on the fatigue tolerance time of similar motions in the user's historical training; it sets the rest interval between sets based on the recovery law of exercise physiology; and it uses the weight adjustment strategy as a variable parameter of the training plan, so that the generated phased training plan includes a weight change strategy that is dynamically adjusted with the training stage.

[0142] In some embodiments, the adaptive weight adjustment strategy based on the user's real-time force stability and historical strength growth curve includes: judging the user's real-time force stability by analyzing the fluctuation range of real-time collected grip pressure data and force angle data; if the fluctuation range is within a preset stability threshold, the force is determined to be stable; otherwise, the force is determined to be unstable; retrieving weight increase data of the same movement type from the user's historical training records to form a historical strength growth curve; when the force is stable and the historical strength growth curve shows an upward trend, gradually increasing the training weight according to a preset incremental rule; when the force is unstable or the historical strength growth curve shows a plateau, maintaining the current training weight or reducing the training weight according to a preset reduction rule.

[0143] In some embodiments, constructing the motion quality scoring model includes: collecting standard motion data from professional trainees performing different training movements through sports biomechanics experiments; the standard motion data includes standard motion trajectory, standard motion speed range, standard motion amplitude, and left-right force symmetry benchmark values; classifying the standard motion data according to motion type to generate standard motion templates containing corresponding standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters; and storing the standard motion templates in the database of the motion quality scoring model.

[0144] In some embodiments, the multi-dimensional matching analysis of motion data with a standard motion template, calculating a motion standardization score from four dimensions—motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry—includes: calculating the coordinate point fitting degree between the real-time collected dumbbell motion trajectory and the standard trajectory curve in the standard motion template to obtain a motion trajectory conformity score; comparing the real-time motion speed with the speed threshold range in the standard motion template, calculating the percentage of time the speed value is within the threshold range as a speed uniformity score; comparing the real-time motion amplitude with the amplitude benchmark value in the standard motion template, calculating the percentage of movements that reach or exceed the amplitude benchmark value as an amplitude compliance rate score; acquiring grip pressure data and force angle data of the smart dumbbells on both sides, calculating the difference between the data on both sides and the deviation from the symmetry benchmark value as a left-right force symmetry score; and calculating the total motion standardization score by combining the scores of the four dimensions.

[0145] In some embodiments, obtaining the user's current fatigue parameters based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude includes: calculating the uninterrupted continuous training duration during the user's current training phase, and marking fatigue risk when the continuous training duration exceeds a preset fatigue warning duration; retrieving heart rate data from the user's historical training records, analyzing the deviation between the current training heart rate and the historical average heart rate to determine the degree of cardiopulmonary fatigue; calculating the fluctuation amplitude of real-time grip pressure data, and determining a decline in muscle control ability when the fluctuation amplitude exceeds a preset stability threshold; and determining the user's current fatigue parameters by comprehensively considering the continuous training duration, heart rate deviation amplitude, and grip pressure fluctuation amplitude through preset fatigue assessment rules.

[0146] In some embodiments, the dynamic correction of the score based on the user's current fatigue level parameters includes: setting a correction coefficient for the action standardization score based on the user's current fatigue level parameters; when the fatigue level parameters indicate that the user is in a state of mild fatigue, the score thresholds for the consistency of the action trajectory and the uniformity of speed are adaptively relaxed; when the fatigue level parameters indicate that the user is in a state of moderate or severe fatigue, in addition to relaxing the score thresholds, an additional judgment on the safety of the action trajectory is added, and if a force deviation that may lead to sports injury occurs, the action standardization score is directly reduced; through the correspondence between the fatigue level parameters and the correction rules, the dynamic adjustment of the action quality score is realized.

[0147] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the steps of the method for dynamically generating training plans and scoring motion quality of a smart dumbbell as provided in any embodiment of this application.

[0148] The computer-readable storage medium can be the internal storage unit of the smart dumbbell described in the foregoing embodiments, such as the hard drive or memory of the smart dumbbell. Alternatively, the computer-readable storage medium can be an external storage device of the smart dumbbell, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the smart dumbbell.

[0149] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for dynamically generating training plans and scoring movement quality using smart dumbbells, characterized in that, Applied to smart dumbbells; the method includes: The smart dumbbells use built-in accelerometers, angular velocity sensors, and pressure sensors to collect real-time motion data from users during training. This motion data includes dumbbell movement trajectory, movement speed, range of motion, grip pressure, and force angle. The system also acquires personalized parameters input by the user, including age, weight, body fat percentage, muscle strength level, training goals, and historical training records. The motion data and personalized parameters are input into the preset training plan generation algorithm module to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals. The weight adjustment strategy is also adaptively adjusted according to the user's real-time force stability and historical strength growth curve. A motion quality scoring model is constructed, in which standard motion templates are pre-stored. These templates include standard trajectory curves, speed threshold ranges, amplitude benchmarks, and force symmetry parameters corresponding to different training movements. Motion data is matched and analyzed against the standard motion templates in multiple dimensions. Motion compliance scores are calculated from four dimensions: motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry. The user's current fatigue level is obtained based on continuous training duration, historical heart rate data, and grip pressure fluctuations. The scores are then dynamically adjusted based on these fatigue parameters.

2. The method according to claim 1, characterized in that, After dynamically correcting the score based on the user's current fatigue level parameters, the method further includes: During training, the motion quality score is fed back to the training plan generation algorithm module in real time. If the motion standardization score is lower than the preset threshold for several consecutive times, the dynamic adjustment mechanism of the training plan is triggered, which automatically reduces the training weight or shortens the training time of a single set until the motion standardization score rises back to a reasonable range, thus forming a closed-loop optimization control between the training plan and motion quality.

3. The method according to claim 1, characterized in that, The system uses the built-in accelerometer, angular velocity sensor, and pressure sensor of the smart dumbbell to collect real-time motion data from the user during training, including: Accelerometers are used to collect real-time acceleration data of dumbbells in three-dimensional space, angular velocity sensors are used to collect real-time angular velocity data of dumbbells around three rotation axes, and pressure sensors are used to collect real-time grip pressure data of users holding dumbbells. Based on the acceleration and angular velocity data, the motion trajectory, speed, and amplitude of the dumbbell are calculated. Based on the acceleration, angular velocity, and grip pressure data, the angle of force exerted by the user is calculated.

4. The method according to claim 1, characterized in that, The method, based on user action data, personalized parameters, historical training fitness, and training objectives, dynamically generates phased training plans that include action types, training duration for each set, rest intervals, and weight adjustment strategies. The training plan generation algorithm module first analyzes the action completion rate, muscle recovery cycle and training effect data in the user's historical training records to determine the historical training adaptability. It then combines the user's current action data to reflect the real-time physical condition, training goals and muscle strength levels in the personalized parameters, and matches suitable action types from the preset action library. The training duration for each set is set based on the fatigue tolerance time of similar movements in the user's historical training, and the rest interval between sets is set according to the recovery law of exercise physiology; the weight adjustment strategy is used as a variable parameter of the training plan, so that the generated phased training plan includes a weight change strategy that is dynamically adjusted with the training phase.

5. The method according to claim 1, characterized in that, The adaptive weight adjustment strategy based on the user's real-time force stability and historical force growth curve includes: The stability of the user's real-time force application is determined by analyzing the fluctuation range of the real-time collected grip pressure data and force angle data. If the fluctuation range is within the preset stability threshold, the force application is determined to be stable; otherwise, the force application is determined to be unstable. The system retrieves weight increase data for the same type of exercise from the user's historical training records to form a historical strength growth curve. When the force exertion is stable and the historical strength growth curve shows an upward trend, the training weight is gradually increased according to the preset increment rules. When the force exertion is unstable or the historical strength growth curve shows a plateau, the current training weight is maintained or the training weight is reduced according to the preset reduction rules.

6. The method according to claim 1, characterized in that, The construction of the action quality scoring model includes: Standard movement data of professional trainees performing different training movements were collected through sports biomechanics experiments. The standard movement data included standard movement trajectory, standard movement speed range, standard movement amplitude, and left-right force symmetry benchmark value. The standard motion data is classified according to motion type, and a standard motion template containing the corresponding standard trajectory curve, speed threshold range, amplitude benchmark value and force symmetry parameter is generated. The standard motion template is then stored in the database of the motion quality scoring model.

7. The method according to claim 1, characterized in that, The process involves multi-dimensional matching analysis of motion data with standard motion templates, calculating a motion standardization score based on four dimensions: motion trajectory consistency, speed uniformity, amplitude compliance rate, and left-right force symmetry. The coordinate point fitting degree of the dumbbell movement trajectory collected in real time is calculated by comparing it with the standard trajectory curve in the standard movement template to obtain the movement trajectory matching score. The real-time motion speed is compared with the speed threshold range in the standard motion template, and the percentage of time the speed value is within the threshold range is calculated as the speed uniformity score; the real-time motion amplitude is compared with the amplitude benchmark value in the standard motion template, and the percentage of movements that reach or exceed the amplitude benchmark value is calculated as the amplitude compliance rate score. The grip pressure data and force angle data of the smart dumbbells on both sides are obtained. The difference between the data on both sides and the degree of deviation from the symmetrical benchmark value are calculated as the left and right force symmetry score. The total score of action standardization is calculated by combining the scores of the four dimensions.

8. The method according to claim 1, characterized in that, The method of obtaining the user's current fatigue level parameters based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude includes: The system calculates the duration of uninterrupted continuous training during the current training phase and marks fatigue risk when the continuous training duration exceeds the preset fatigue warning duration. It also retrieves heart rate data from the user's historical training records and analyzes the deviation between the current training heart rate and the historical average heart rate to determine the degree of cardiopulmonary fatigue. Calculate the fluctuation range of real-time grip pressure data, and determine that muscle control ability has declined when the fluctuation range exceeds a preset stability threshold. By combining continuous training duration, heart rate deviation, and grip pressure fluctuation, the user's current fatigue level parameters are determined through preset fatigue assessment rules.

9. The method according to claim 1, characterized in that, The dynamic correction of the score based on the user's current fatigue level parameters includes: The correction coefficient for the action standardization score is set according to the user's current fatigue level parameter. When the fatigue level parameter indicates that the user is in a state of mild fatigue, the score thresholds for action trajectory consistency and speed uniformity are adaptively relaxed. When the fatigue parameter indicates that the user is in a state of moderate or higher fatigue, in addition to relaxing the scoring threshold, an additional judgment on the safety of the movement trajectory is added. If there is a force deviation that may lead to sports injury, the movement standardization score will be directly reduced. By establishing a correspondence between fatigue parameters and correction rules, dynamic adjustments to the motion quality score can be achieved.

10. A system for dynamically generating training plans and scoring movement quality using intelligent dumbbells, characterized in that, Applied to smart dumbbells; the system includes: The parameter acquisition unit is used to collect the user's movement data in real time during training through the accelerometer, angular velocity sensor and pressure sensor built into the smart dumbbell. The movement data includes the dumbbell's movement trajectory, movement speed, movement range, grip pressure and force angle. It also acquires personalized parameters input by the user, including age, weight, body fat percentage, muscle strength level, training goals and historical training records. The strategy acquisition unit is used to input motion data and personalized parameters into the preset training plan generation algorithm module, so as to dynamically generate a phased training plan that includes motion type, training duration of each set, rest time and weight adjustment strategy based on user motion data, personalized parameters, historical training fitness and training goals, and adaptively adjust the weight adjustment strategy according to the user's real-time force stability and historical strength growth curve. A dynamic correction unit is used to construct a motion quality scoring model. The scoring model pre-stores standard motion templates, which include standard trajectory curves, speed threshold ranges, amplitude benchmark values, and force symmetry parameters corresponding to different training movements. The unit performs multi-dimensional matching analysis between the motion data and the standard motion templates, and calculates the motion standardization score from four dimensions: motion trajectory conformity, speed uniformity, amplitude compliance rate, and left-right force symmetry. The unit obtains the user's current fatigue parameter based on continuous training duration, historical heart rate data, and grip pressure fluctuation amplitude, and dynamically corrects the score based on the user's current fatigue parameter.

Images