Martial arts action guidance and digitalization dissemination method and system based on digital twinning

Abstract

This invention discloses a method and system for martial arts movement instruction and digital dissemination based on digital twins. It acquires three-dimensional joint coordinates and motion parameter data from the practitioner's limbs using motion sensors. A preset coordinate transformation algorithm processes the data to form a standardized input sequence, resulting in standardized limb joint data. This constructs a digital twin model of the practitioner. Simultaneously, a standard template model of the corresponding martial arts movement is retrieved from a preset database. A three-dimensional mapping algorithm is used to overlay and compare the two models to determine the initial joint position matching degree. If the matching degree is lower than a preset threshold, a deviation detection mechanism is triggered. Joint angle deviation and movement rhythm error indicators are extracted from the comparison results, and quantified values ​​are calculated using machine learning algorithms to obtain the deviation analysis results. This invention improves the accuracy and safety of training, significantly enhances the overall instruction effect, and provides an innovative technical path for the digital presentation, immersive dissemination, and cultural inheritance of traditional martial arts.

Description

Technical Field

[0001] This invention relates to the field of martial arts teaching technology, and in particular discloses a method and system for martial arts movement instruction and digital dissemination based on digital twins. Background Technology

[0002] Martial arts, as an important component of traditional Chinese culture, not only carries profound cultural value but also holds an irreplaceable position in the fields of martial arts, competition, and education. Its inheritance and promotion have far-reaching significance for promoting national spirit and improving the physical fitness of the entire population. However, how to achieve efficiency and standardization in martial arts teaching within the context of modern technology has become an important issue that urgently needs to be addressed.

[0003] In current martial arts teaching, traditional methods mainly rely on on-site instruction from coaches and repeated imitation by practitioners. However, this approach has significant shortcomings. Coaches' feedback is often based on subjective judgment, making it difficult to accurately assess subtle deviations in movements. At the same time, practitioners lack timely correction mechanisms during the learning process, which can easily lead to the formation of incorrect movement habits.

[0004] A deeper problem lies in the fact that current technologies often fail to comprehensively cover the dynamic changes of multiple joints in the human body when capturing and analyzing movements, resulting in insufficient targeted instruction. Furthermore, the complexity of martial arts movements lies in their extremely high demands on limb coordination and rhythm, and current technologies face significant challenges in handling these dynamic characteristics. In particular, the subtle changes in joint angles—a core factor—directly affect the standardization and aesthetics of the movements. Due to the inability to effectively capture and compare real-time changes in joint angles during movements, errors often accumulate in teaching. For example, in practicing a push-hands movement in Tai Chi, the practitioner's elbow angle may deviate from the standard by 5 degrees, affecting the force trajectory of the entire upper limb and even impairing the continuity of subsequent movements. If such subtle deviations are not detected and corrected in time, they will gradually amplify, ultimately affecting the overall quality of the movement.

[0005] Therefore, how to accurately capture and provide real-time feedback on changes in joint angles in martial arts teaching has become a key issue in improving teaching efficiency and the standardization of movements. Summary of the Invention

[0006] This invention provides a method and system for martial arts movement instruction and digital dissemination based on digital twins, aiming to solve at least one of the defects existing in the above-mentioned prior art.

[0007] One aspect of this invention relates to a method for martial arts movement instruction and digital dissemination based on digital twins, comprising the following steps:

[0008] S100: Acquire joint three-dimensional coordinate data and motion parameter data from the exerciser's limbs through motion sensor devices, process the joint three-dimensional coordinate data and motion parameter data using a preset coordinate transformation algorithm to form a uniform input sequence, and obtain standardized limb joint data;

[0009] S200. Construct a digital twin model of the practitioner based on standardized limb joint data, and simultaneously obtain the standard template model of the corresponding martial arts movement from the preset database. Use a three-dimensional mapping algorithm to overlay and compare the practitioner's digital twin model and the standard template model to determine the preliminary joint position matching degree.

[0010] S300. If the initial joint position matching degree is lower than the preset threshold, the deviation detection mechanism is triggered. The joint angle deviation and movement rhythm error index are extracted from the comparison results between the practitioner's digital twin model and the standard template model. The quantitative deviation values ​​of the joint angle deviation and movement rhythm error index are calculated to obtain the deviation analysis results.

[0011] S400. Obtain information on high-deviation joints from the deviation analysis results, where the quantified deviation value exceeds the preset deviation threshold. Generate a visual feedback report to mark the high-deviation joints. Use image rendering technology to overlay the standard motion trajectory onto the visual feedback report to determine the marked feedback structure.

[0012] S500: Generate a correction suggestion sequence based on the labeled feedback structure, and use a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence. Simultaneously, encapsulate the standardized martial arts movement template, practitioner movement data, deviation analysis results, and the adjusted correction suggestion sequence into a standardized digital dissemination package. This package supports dissemination through multiple formats such as short videos, 3D interactive demonstrations, and online courses. At the same time, it extracts user interaction data from the visual feedback report and dynamically updates the practitioner's digital twin model and the dissemination content library.

[0013] Further, step S100 includes:

[0014] S110. Obtain joint three-dimensional coordinate data and motion parameter data from the exerciser's limbs through motion sensor devices;

[0015] S120. The preset coordinate transformation algorithm is used to process the three-dimensional coordinate data and motion parameter data of the joint, and the three-dimensional coordinate data and motion parameter data from different sources are converted into a unified format input sequence to obtain standardized limb joint data.

[0016] Further, step S200 includes:

[0017] S210. Based on standardized limb joint data, use graphics rendering tools to generate a digital twin model of the practitioner;

[0018] S220. Obtain the standard template model of the corresponding martial arts movement from the preset database, and use a data synchronization tool to align the practitioner's digital twin model with the standard template model on the timeline.

[0019] S230. Using a 3D mapping tool, the time-axis-aligned digital twin model of the trainee is spatially superimposed with the standard template model to determine the preliminary joint position matching degree.

[0020] Further, step S300 includes:

[0021] S310. If the joint position matching degree is lower than the preset threshold, the deviation detection mechanism is triggered to extract the joint angle deviation and movement rhythm error index from the comparison results between the practitioner's digital twin model and the standard template model.

[0022] S320. Based on the joint angle deviation and movement rhythm error index, calculate the quantitative deviation value of the joint angle deviation and movement rhythm error index to obtain the deviation analysis result.

[0023] Further, step S400 includes:

[0024] S410. Extract information on high-deviation joints whose quantitative deviation values ​​exceed a preset deviation threshold from the deviation analysis results. If the number of joints in the extracted high-deviation joint information is lower than a preset number threshold, perform secondary filtering using a data filtering tool to obtain a list of joints that meet the labeling requirements.

[0025] S420. Based on the list of joint parts, use image processing tools to generate an initial visualization feedback report, make preliminary annotations for high deviation parts, and determine the annotation content;

[0026] S430. Use an image rendering tool to overlay the pre-established standard motion trajectory data onto the initial visualization feedback report, obtain the overlaid trajectory comparison view, and determine whether it meets the display requirements.

[0027] S440. If the superimposed trajectory comparison view meets the display requirements, the report integration tool is used to adjust the format of the superimposed trajectory comparison view and the annotation content to obtain the annotation feedback structure.

[0028] Further, step S500 includes:

[0029] S510. Generate a correction suggestion sequence based on the labeled feedback structure, and use a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence;

[0030] S520 simultaneously encapsulates standardized martial arts movement templates, practitioner movement data, deviation analysis results, and adjusted correction suggestions into a standardized digital dissemination package, supporting output and dissemination through multiple formats such as short videos, 3D interactive demonstrations, and online courses.

[0031] S530: Simultaneously, user interaction data is extracted from the visual feedback report to dynamically update the practitioner's digital twin model and the content library, achieving synergistic optimization of precise movement guidance and efficient digital dissemination.

[0032] Another aspect of the present invention relates to a digital twin-based martial arts movement instruction and digital dissemination system, used to perform the aforementioned digital twin-based martial arts movement instruction and digital dissemination method, comprising:

[0033] The standardized limb joint data acquisition module is used to acquire three-dimensional coordinate data and motion parameter data of the joints from the exerciser's limbs through motion sensor devices. It processes the three-dimensional coordinate data and motion parameter data of the joints using a preset coordinate transformation algorithm to form a uniform input sequence and obtain standardized limb joint data.

[0034] The joint position matching degree determination module is used to construct a digital twin model of the practitioner based on standardized limb joint data, and simultaneously obtain the standard template model of the corresponding martial arts movement from the preset database. The digital twin model of the practitioner and the standard template model are superimposed and compared using a three-dimensional mapping algorithm to determine the preliminary joint position matching degree.

[0035] The deviation analysis result acquisition module is used to trigger the deviation detection mechanism if the initial joint position matching degree is lower than the preset threshold. It extracts joint angle deviation and movement rhythm error indicators from the comparison results between the practitioner's digital twin model and the standard template model, calculates the quantitative deviation value of the joint angle deviation and movement rhythm error indicators, and obtains the deviation analysis result.

[0036] The feedback structure determination module is used to obtain information on high-deviation joints whose quantitative deviation values ​​exceed a preset deviation threshold from the deviation analysis results, generate a visual feedback report to mark the high-deviation joints, and use image rendering technology to overlay the standard motion trajectory onto the visual feedback report to determine the marked feedback structure.

[0037] The motion guidance and digital dissemination collaborative optimization module is used to generate a sequence of correction suggestions based on the annotated feedback structure. It uses a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence and simultaneously encapsulates the standardized martial arts motion template, practitioner motion data, deviation analysis results, and the adjusted correction suggestion sequence into a standardized digital dissemination package. This package supports dissemination through various formats such as short videos, 3D interactive demonstrations, and online courses. At the same time, it extracts user interaction data from the visual feedback report and dynamically updates the practitioner's digital twin model and the dissemination content library.

[0038] The beneficial effects achieved by this invention are as follows:

[0039] This invention provides a method and system for martial arts movement instruction and digital dissemination based on digital twins. Addressing the problem of difficulty in real-time and accurate detection and correction of movement deviations in sports training, leading to low training efficiency and potential injury risks, the method acquires three-dimensional coordinates and motion parameters of the practitioner's joints using motion sensors. A preset coordinate transformation algorithm processes the data to form a standardized input sequence, resulting in standardized limb joint data. This constructs a digital twin model of the practitioner. Simultaneously, a standard template model of the corresponding martial arts movement is retrieved from a preset database. A three-dimensional mapping algorithm is used to overlay and compare the two models to determine the initial joint position matching degree. If the matching degree is lower than a preset threshold, a deviation detection mechanism is triggered. Joint angle deviation and movement rhythm error indicators are extracted from the comparison results. Machine learning algorithms are used to calculate quantified values ​​and obtain deviation analysis results. Information on high-deviation joints is then extracted from these results, and a visual feedback report is generated to mark these high-deviation areas. Image rendering technology is used to overlay standard movement trajectories to determine the marked feedback structure. Based on this feedback structure, a correction suggestion sequence is generated, and a sequence optimization algorithm is used to adjust the dynamic presentation order. Simultaneously, standardized martial arts movement templates, practitioner movement data, deviation analysis results, and the adjusted correction suggestion sequence are packaged into a standardized digital dissemination package. This package supports dissemination through short videos, 3D interactive demonstrations, and online courses. Furthermore, user interaction data is extracted from the visual feedback report to dynamically update the practitioner's digital twin model and the dissemination content library, achieving synergistic optimization of precise movement guidance and efficient digital dissemination.

[0040] The beneficial effects achieved by this invention are specifically as follows:

[0041] 1. This invention synchronously retrieves standard template models of corresponding martial arts movements from a preset database, uses a 3D mapping algorithm to overlay and compare the two to determine the initial joint position matching degree. If the matching degree is lower than a preset threshold, a deviation detection mechanism is triggered. Joint angle deviation and movement rhythm error indicators are extracted from the comparison results, and quantified values ​​are calculated using machine learning algorithms to obtain deviation analysis results. Further, information on high-deviation joints is obtained from the results, and a visual feedback report is generated to mark the high-deviation parts. Standard movement trajectories are overlaid using image rendering technology to determine the marked feedback structure. A correction suggestion sequence is generated based on this feedback structure, and a sequence optimization algorithm is used to adjust the dynamic presentation order. This achieves real-time deviation detection, accurate feedback, and dynamic optimization of martial arts movements, improving the accuracy and safety of training and significantly enhancing the overall guidance effect.

[0042] 2. By acquiring three-dimensional coordinates and motion parameter data of the joints from the practitioner's limbs through motion sensors, and processing the data using a preset coordinate transformation algorithm to form a unified input sequence, standardized limb joint data is obtained, thereby constructing a digital twin model of the practitioner, which can be used for three-dimensional reconstruction and virtual display of martial arts movements.

[0043] 3. Simultaneously, standardized martial arts movement templates, practitioner movement data, deviation analysis results, and adjusted correction suggestions are packaged into a standardized digital dissemination package. This package supports dissemination through various formats, including short videos, 3D interactive demonstrations, and online courses. Furthermore, user interaction data is extracted from visual feedback reports to dynamically update the practitioner's digital twin model and the dissemination content library. It supports recording, playback, and immersive interactive experiences, creating a visualized digital martial arts performance. This not only improves the standardization of martial arts teaching and training but also provides an innovative technological path for the digital presentation, immersive dissemination, and cultural inheritance of traditional martial arts, possessing significant educational and cultural promotion value. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating an embodiment of the martial arts movement instruction and digital dissemination method based on digital twins according to the present invention.

[0045] Figure 2 This is a functional block diagram of an embodiment of the martial arts movement guidance and digital dissemination system based on digital twins of the present invention.

[0046] Explanation of icon numbers:

[0047] 10. Standardized limb joint data acquisition module; 20. Joint position matching degree determination module; 30. Deviation analysis result acquisition module; 40. Feedback structure determination module; 50. Collaborative optimization module for motion guidance and digital dissemination. Detailed Implementation

[0048] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0049] like Figure 1 As shown, the first embodiment of the present invention proposes a method for martial arts movement instruction and digital dissemination based on digital twins, including the following steps:

[0050] Step S100: Obtain joint three-dimensional coordinate data and motion parameter data from the exerciser's limbs through motion sensor devices, process the joint three-dimensional coordinate data and motion parameter data using a preset coordinate transformation algorithm to form a uniform input sequence, and obtain standardized limb joint data.

[0051] Motion sensor devices (such as inertial measurement units, optical motion capture sensors, and electromagnetic tracking sensors) deployed on key joints of the practitioner's limbs (such as shoulders, elbows, wrists, hips, knees, and ankles) are used to collect two types of core data in real time: first, joint three-dimensional coordinate data (i.e., the spatial position information of each joint in three-dimensional space, such as X / Y / Z coordinate values ​​in the Cartesian coordinate system); second, motion parameter data (including dynamic characteristic data such as joint motion velocity, acceleration, angular velocity, angular acceleration, and motion trajectory curvature). Pre-set coordinate transformation algorithms (such as coordinate system calibration algorithms, heterogeneous data normalization processing, and coordinate origin...) are employed. The unified mapping algorithm eliminates format differences, coordinate system inconsistencies, and measurement errors (such as data offsets caused by sensor installation deviations) in data from different sensors. It transforms two types of raw data—joint 3D coordinate data and motion parameter data—into a time-series input sequence with a unified dimension and format (such as a standardized data array sorted by timestamps). The result is standardized limb joint data with a well-structured structure that can be directly used for modeling. The data format consistency is required to be ≥99%, the coordinate transformation error ≤0.5cm, and the motion parameter quantization accuracy ≤0.1rad / s, providing accurate data support for the subsequent construction of digital twin models.

[0052] Step S200: Construct a digital twin model of the practitioner based on standardized limb joint data, and simultaneously obtain the standard template model of the corresponding martial arts movement from the preset database. Use a three-dimensional mapping algorithm to overlay and compare the practitioner's digital twin model and the standard template model to determine the preliminary joint position matching degree.

[0053] Using the standardized limb joint data generated in step S100 as the core input, a digital twin model that is synchronized with the practitioner's limb movements in real time is constructed through 3D modeling and skeletal binding technology (such as a real-time rendering algorithm based on skeletal skinning). The digital twin model accurately replicates the spatial position, movement trajectory, and limb connection relationships of the practitioner's key joints, restoring the dynamic process of the movement. Simultaneously, a standard template model of martial arts movements corresponding to the practitioner's current performance content is retrieved from a preset database (this digital twin model is recorded by professional martial arts athletes and generated after data calibration, containing benchmark information such as joint position, range of motion, and limb posture of the standard movements). Using 3D mapping algorithms that adapt to the action comparison scene (such as point cloud registration algorithm and one-to-one correspondence mapping technology for bones and joints), the digital twin model of the practitioner and the standard template model are superimposed and aligned in the same 3D coordinate system (ensuring that the central axis of the torso and the origin of the core joints coincide). By calculating the spatial position distance and limb angle deviation of corresponding joints (such as left shoulder to left shoulder, right knee to right knee), the preliminary joint position matching degree is quantified (such as a matching degree of 80% indicates the degree of fit between the overall joint position and the standard model). The model superposition alignment error is required to be ≤1cm, and the joint position matching degree calculation accuracy is required to be ≤±2%, providing a preliminary judgment basis for subsequent deviation detection.

[0054] Step S300: If the initial joint position matching degree is lower than the preset threshold, the deviation detection mechanism is triggered. The joint angle deviation and movement rhythm error index are extracted from the comparison results between the practitioner's digital twin model and the standard template model. The quantitative deviation values ​​of the joint angle deviation and movement rhythm error index are calculated to obtain the deviation analysis results.

[0055] First, determine whether the preliminary joint position matching degree obtained in step S200 is lower than a preset threshold (e.g., matching degree < 85%, the threshold can be dynamically adjusted according to the training difficulty). If the condition is met, the deviation detection mechanism is automatically triggered. From the 3D superposition comparison results of the practitioner's digital twin model and the standard template model, extract two types of core error indicators: joint angle deviation (e.g., the difference between the actual angle and the standard angle of key joints such as the shoulder-elbow angle and hip-knee bending angle) and movement rhythm error indicators (e.g., the completion time of a single movement segment, the sequence of joint movements, and the difference between the overall movement cycle and the standard rhythm). Input the two types of indicators, joint angle deviation and movement rhythm error indicators, into a preset machine learning algorithm (e.g., random forest regression model, BP neural network). The machine learning algorithm performs weight allocation and quantification calculation on the error features, and outputs specific numerical deviation results (e.g., elbow joint angle deviation of 12°, punching rhythm lag of 0.4 seconds). Finally, a structured deviation analysis result containing high deviation joint identifiers, error values, and influencing weights is formed. The deviation quantification accuracy is required to be ≥ 93%, and the error calculation error is ≤ ± 1%, providing a precise basis for subsequent feedback report generation and correction suggestion formulation.

[0056] Step S400: Obtain information on high-deviation joints from the deviation analysis results where the quantified deviation value exceeds the preset deviation threshold, generate a visualization feedback report to mark the high-deviation joints, and use image rendering technology to overlay the standard motion trajectory onto the visualization feedback report to determine the marked feedback structure.

[0057] From the deviation analysis results in step S300, information on high-deviation joints (including joint name, deviation type, and specific error value) with quantified deviation values ​​exceeding preset deviation thresholds (e.g., angle deviation ≥10°, rhythm error ≥0.3 seconds; thresholds can be set according to the difficulty level of martial arts movements) is selected. Based on this information, a structured visual feedback report is generated. The visual feedback report uses the practitioner's digital twin model as a carrier, clearly identifying high-deviation joints and their corresponding degree of deviation through highlighting, color differentiation (e.g., red marking of high-deviation joints), and numerical pop-ups. Simultaneously, real-time image rendering technology (e.g., dynamic trajectory line drawing, semi-transparent motion shadow overlay, and 3D arrow guidance) is used to overlay the joint movement trajectory and limb posture path of the standard movement onto the digital twin model of the feedback report, forming a comparative visualization effect of "practitioner's movement + standard trajectory". Finally, a post-annotation feedback structure is determined, including high-deviation annotations, standard trajectory overlay, and deviation value explanations. The annotation accuracy is required to be ≥99%, and the trajectory overlay synchronization error is ≤0.1 seconds, allowing the practitioner to intuitively perceive the location of the movement deviation and the direction of correction, providing clear guidance for subsequent correction suggestions.

[0058] Step S500: Generate a correction suggestion sequence based on the labeled feedback structure, and use a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence. Simultaneously, encapsulate the standardized martial arts movement template, practitioner movement data, deviation analysis results, and the adjusted correction suggestion sequence into a standardized digital dissemination package, which supports output and dissemination through multiple forms such as short videos, 3D interactive demonstrations, and online courses. At the same time, extract user interaction data from the visualization feedback report and dynamically update the practitioner digital twin model and dissemination content library.

[0059] Based on the feedback structure annotated in step S400 (including high-deviation joint markers and overlay of standard movement trajectories), a targeted and actionable sequence of movement correction suggestions is generated. Each suggestion clearly points to a specific deviation (e.g., "shoulder abduction angle is less than 20°, range of motion needs to be appropriately increased," "waist twisting rhythm is delayed by 0.5 seconds, leg force needs to be synchronized"). Sequence optimization algorithms (such as greedy algorithms, dynamic programming algorithms, and priority sorting algorithms) are used, combined with deviation impact weights (core joints prioritized, basic movements prioritized), practice difficulty gradients (easy to difficult), and movement connection logic, to adjust the dynamic presentation order of the correction suggestion sequence, ensuring that the practitioner can gradually correct their movements along the optimal path. Simultaneously, standardized martial arts movement templates (3D standardized models), practitioners' original movement data (standardized limb joint data), deviation analysis results (quantified deviation values, error indicators), and adjusted correction suggestion sequences are packaged into a standardized digital dissemination package in a unified data format. This standardized digital dissemination package is compatible with multiple output scenarios and supports digital dissemination through various forms such as short videos (simplified movement comparison clips), 3D interactive demonstrations (with drag-and-rotate viewing of standard and deviation movements), and online courses (including theoretical explanations and practical guidance). User interaction data (such as the frequency of correction suggestion views, 3D demonstration operation duration, and repeated practice movement data) is extracted from the visual feedback report. Based on this user interaction data, the practitioner's digital twin model (optimizing movement restoration accuracy and adapting to the practitioner's movement improvement trends) and the dissemination content library (supplementing in-depth guidance content corresponding to high-frequency deviations and adapted versions of popular dissemination formats) are dynamically updated. Ultimately, this achieves a collaborative closed loop of "precise movement guidance (targeted correction) - efficient digital dissemination (multi-channel coverage) - data-driven iteration (dynamic optimization)," requiring a correction suggestion adoption rate ≥75%, a dissemination package adaptation rate 100%, and a model update response delay ≤10 minutes.

[0060] Furthermore, the method for martial arts movement instruction and digital dissemination based on digital twins proposed in this embodiment includes step S100:

[0061] Step S110: Obtain joint three-dimensional coordinate data and motion parameter data from the exerciser's limbs using a motion sensor device.

[0062] The following formula is used to describe the three-dimensional joint coordinate data acquired by the motion sensor:

[0063] (1)

[0064] In formula (1), Indicates the first Each joint at any time The three-dimensional coordinate vector, Indicates the first Each joint The coordinate values ​​along the axis. This indicates that the joint is in The coordinate values ​​along the axis. Indicates the first Each joint The coordinate values ​​in the axial direction. The control logic of formula (1) is to encapsulate the structured data of the "three-dimensional spatial position of a single joint at a single moment". Its core function is to integrate the discrete coordinate data (x, y, z axes) collected by the motion sensor into a standardized vector form, which is convenient for subsequent data processing and analysis.

[0065] The following formula is used to obtain motion parameter data measured by motion sensor devices:

[0066] (2)

[0067] In formula (2), Indicates the first Each sensor at time The average acceleration amplitude, Indicates the integration time window. , , These represent the acceleration components of the sensor along the three coordinate axes. The integral variable is represented by the control logic of formula (2), which is to calculate the time average of the amplitude of the triaxial acceleration of a single sensor within a local time window. The core is to obtain the stable characteristic value of the acceleration intensity of the joint corresponding to the sensor near the current moment by first calculating the instantaneous acceleration amplitude and then performing the integral average within the time window.

[0068] In martial arts practice, wearable motion sensor devices worn on the practitioner's arms, legs, and torso capture real-time three-dimensional coordinate data of the joints. This data includes the x, y, and z-axis positions of the elbow, knee, and shoulder joints in space, as well as motion parameters such as joint rotation angles, acceleration, and velocity. These devices typically operate based on the principle of inertial measurement units (IMUs), combining gyroscopes and accelerometers to sense changes in limb movement, thus providing the foundational data input for subsequent posture analysis. In this way, when a practitioner performs the "Warrior II" stance, the sensors record data from the initial posture to the completed posture, ensuring data accuracy and real-time performance. This facilitates personalized guidance in martial arts applications.

[0069] Step S120: Process the joint three-dimensional coordinate data and motion parameter data using a preset coordinate transformation algorithm, convert the three-dimensional coordinate data and motion parameter data from different sources into a unified format input sequence to obtain standardized limb joint data.

[0070] The following formula enables the transformation of joint coordinate data from different sources to a unified coordinate system via rigid body transformation:

[0071] (3)

[0072] In formula (3), This represents the standardized 3D joint coordinate vector. Represents the rotation transformation matrix. This represents the original input joint 3D coordinate data. The translation transformation vector is represented. The control logic of formula (3) is to map the joint coordinates of different sources / different initial postures to the same target coordinate system through rigid body transformation of "rotation + translation". The core is to eliminate the coordinate system differences between multiple sensors and multiple joints and realize the spatial alignment of data.

[0073] The following formula is used to normalize motion parameter data with different dimensions and numerical ranges into a unified format:

[0074] (4)

[0075] In formula (4), Indicates the first The standardized results of each motion parameter Indicates the first One set of original motion parameter data, Indicates the first The mean of each motion parameter, Indicates the first The standard deviation of each motion parameter. The control logic of formula (4) is to perform "mean-standard deviation standardization (Z-score standardization)" on motion parameters with different dimensions and different numerical ranges. The core is to eliminate the differences in dimensions and numerical scales between parameters and to uniformly map all parameters to a standard distribution with a mean of 0 and a standard deviation of 1.

[0076] The processed joint coordinates and motion parameters are then concatenated into a standardized limb joint data sequence in a predetermined order using the following formula:

[0077] (5)

[0078] In formula (5), This represents an input sequence in a uniform format. arrive express Standardized three-dimensional coordinate data of each joint, arrive express Standardized motion parameter data. The control logic of formula (5) is to combine the "standardized joint coordinate data" and the "standardized motion parameter data" into a unified one-dimensional sequence in a preset fixed order. The core is to package the scattered multi-dimensional data (different joints, different parameters) into a structured "data unit" to adapt to the input format requirements of subsequent algorithms (such as action recognition, posture evaluation model).

[0079] When data is acquired from different sources, such as some data from the sensors of a smart bracelet and some from visual data captured by the camera built into a mobile phone, these sources may use different coordinate systems. For example, the bracelet uses local device coordinates, while the camera uses world coordinates. This necessitates the use of a pre-defined coordinate transformation algorithm. This algorithm first identifies the differences in data sources, and then maps all data to a unified Cartesian coordinate system through matrix transformation methods. For example, it converts the bracelet's local coordinates to global coordinates, ensuring that the three-dimensional coordinate data of all joints and motion parameter data such as velocity vectors conform to the same unit and format requirements, thereby generating a standardized input sequence. Suppose a practitioner is practicing Tai Chi. Sensor devices collect raw data from multiple joint points, including the coordinates of the hip joint (2.5, 1.0, 0.5) and the corresponding angular velocity parameter 0.8 rad / s. However, if the data comes from different devices, the coordinates may be in meters or centimeters. The coordinate transformation algorithm will first normalize the units, and then apply a rotation matrix to convert the local rotation into a unified Euler angle representation. Finally, it outputs a serialized data stream, where each timestamp corresponds to standardized joint data, such as [time t, joint 1: (x1, y1, z1, vel1), joint 2: (x2, y2, z2, vel2)]. The standardized limb joint data obtained in this way can be directly input into the posture recognition model, improving the system's compatibility and analysis efficiency. The core of this coordinate transformation algorithm lies in handling data heterogeneity. It eliminates source differences through preset transformation rules such as affine transformation. For example, in martial arts training, if one sensor provides the coordinate data of the right wrist as (100, 200, 50) in pixels, and another provides (1.0, 2.0, 0.5) in centimeters, the coordinate transformation algorithm will calculate the scaling factor and apply the translation vector to unify the two into a standard sequence of meters. This process ensures data consistency, thereby avoiding error accumulation in subsequent motion assessments and providing more accurate feedback, such as prompting the practitioner to adjust the arm angle to optimize the smoothness of the movement.

[0080] Furthermore, the method for martial arts movement instruction and digital dissemination based on digital twins proposed in this embodiment includes step S200:

[0081] Step S210: Based on standardized limb joint data, generate a digital twin model of the practitioner using a graphics rendering tool;

[0082] The following formula is used to generate a personalized digital twin 3D model through geometric transformation:

[0083] (6)

[0084] In formula (6), The three-dimensional vertex matrix representing the digital twin model. Represents the rotation transformation matrix. Represents the scaling transformation matrix. Represents the translation transformation matrix. The vertex of the basic human body model is represented. The control logic of formula (6) is to generate a personalized digital twin three-dimensional model adapted to the practitioner from the basic human body model based on standardized limb joint data through a combination of geometric transformations of "rotation → scaling → translation". The core is to adjust the basic model according to the physical characteristics of the martial arts practitioner (such as limb length and joint ratio) to achieve the matching of the form and posture of the twin with the real practitioner.

[0085] When generating a digital twin model of a practitioner using standardized limb joint data, a graphics rendering tool such as the Unity engine is used for processing. This graphics rendering tool first reads the joint coordinates and motion parameters from the standardized data. For example, in a martial arts training scenario, the data includes the x, y, and z positions and velocity values ​​of the spine, hip joint, and ankle joint.

[0086] The graphics rendering tool constructs a 3D mesh model and maps this data onto a virtual skeletal structure. This works by using skeletal animation technology, where each joint acts as a bone node. The rendering tool calculates the relative positions and rotational quaternions between these nodes to simulate limb movement, thus creating a dynamic digital twin. This process involves inputting a standardized sequence into the rendering pipeline and updating the model's pose frame-by-frame to ensure the model reflects the practitioner's actual movements in real time.

[0087] Step S220: Obtain the standard template model of the corresponding martial arts movement from the preset database, and use a data synchronization tool to align the practitioner's digital twin model with the standard template model on the timeline.

[0088] The following formula is used to retrieve the standard template model of the corresponding martial arts movement from a preset database:

[0089] (7)

[0090] In formula (7), This indicates a standard template model retrieved from a pre-defined database. Representing candidate models, This represents the set of all available standard templates in the preset database. This indicates the number of key feature points in a martial arts movement. Indicates the first The weight coefficients of each feature point Indicates the characteristic points of the practitioner's movements With template feature points Distance metric between The first candidate template The control logic of formula (7) is to select the template model that best matches the current movement of the practitioner from the standard template database. The core is to achieve the optimal matching of the movement template through "weighted distance measurement + minimizing error", so as to provide a reference benchmark for the standardized evaluation of martial arts movements.

[0091] When retrieving a standard template model for a corresponding martial arts movement from a pre-set database, this database typically stores various pre-recorded professional movement data. For example, in martial arts training, the template model includes sequences of martial arts movements performed by professional athletes, covering the complete joint trajectory from the start to the end of a movement. The retrieval process is implemented through a query interface. The user inputs the movement type, such as "Warrior II," and the system retrieves and loads the matching template. This template is saved in a standardized format, including timestamps and joint parameters. The principle is based on keyframe animation storage, ensuring that the template is compatible with the data of any practitioner.

[0092] When using data synchronization tools for timeline alignment, the sequence of the practitioner's digital twin model is aligned with that of the standard template model. For example, in martial arts training, if the practitioner's model sequence is 30 frames per second, while the template is 60 frames per second, the data synchronization tool first identifies the starting point, such as the moment the foot touches the ground. Then, it uses a dynamic time warping algorithm to adjust the sequence length and calculates the time offset, such as aligning the practitioner's sequence at t=0 to the template's t=0.5, ensuring synchronization of the movement phases. The principle behind this data synchronization tool is to minimize the distance between time series and fill in missing frames through interpolation, thereby making the two models coincide in the time dimension.

[0093] Step S230: Use a 3D mapping tool to spatially overlay the time-axis aligned digital twin model of the trainee with the standard template model to determine the preliminary joint position matching degree.

[0094] The initial joint position matching degree is obtained by the following formula:

[0095] (8)

[0096] In formula (8), Indicates the number of trainees The first joint and standard template The positional matching degree of each joint This represents the number of coordinate dimensions in three-dimensional space. Indicates the first Weighting coefficients for dimensional coordinates, The digital twin model of the trainee is represented by the first... The joint in the first dimensional coordinates, The standard template model is represented by the first one. The joint in the first dimensional coordinates, The standard deviation parameter represents the spatial distance. The control logic of formula (8) is to quantify the degree of spatial position matching between the practitioner's joints and the standard template joints by using "weighted Gaussian distance". The core is to use the method of "weakening small distance differences and strengthening large deviations" to more accurately assess the fit of the joint positions.

[0097] During spatial overlay processing, the 3D mapping tool overlays the aligned models. For example, in martial arts practice, the tool compares the coordinates of the practitioner's joints, such as the shoulder (1.2, 0.8, 0.3), with the corresponding points on the template (1.1, 0.9, 0.4) using vectors, calculating the Euclidean distance to determine the matching degree. The principle is to apply transformation matrices, such as translation and rotation matrices, to place the practitioner's digital twin model and the standard template model in the same coordinate space. Then, the differences are visualized by overlaying each joint. A preliminary matching degree is determined by an average distance threshold, such as less than 0.1 meters, indicating a high match. This process ensures spatial consistency, providing a foundation for subsequent analysis.

[0098] Furthermore, the method for martial arts movement instruction and digital dissemination based on digital twins proposed in this embodiment includes step S300:

[0099] Step S310: If the joint position matching degree is lower than the preset threshold, the deviation detection mechanism is triggered to extract the joint angle deviation and movement rhythm error index from the comparison results between the practitioner's digital twin model and the standard template model.

[0100] Joint angle deviation is calculated using the following formula:

[0101] (9)

[0102] In formula (9), Indicates the first Angle deviation of each joint and Indicates the first test action Two bone vectors connected by a joint. and Indicates the first reference action The joint angle deviation is quantified by calculating the difference between the corresponding joint angles in the two movements. The control logic of formula (9) is to quantify the degree of deviation between the joint angle of the trainee and the reference movement by "calculating the angle between the bone vectors". The core is to first calculate the joint angles of the incident movement and the reference movement respectively, and then calculate the difference between the two.

[0103] The movement rhythm error index is derived using the following formula:

[0104] (10)

[0105] In formula (10), Indicator of movement rhythm error. This represents the total number of frames in the action sequence. Indicates the test action in the first... The amount of frame displacement. Indicates the reference action in the first position. The amount of frame displacement. Indicates the total duration of the test action. The total duration of the reference action is represented by the normalized inter-frame displacement difference to evaluate the rhythm error of the action. The control logic of formula (10) is to quantify the rhythm matching degree between the trainee's action and the reference action by the frame-by-frame difference of the "normalized inter-frame displacement". The core is to first unify the "time scale" and then compare the displacement changes of each frame.

[0106] When the joint position matching accuracy falls below a preset threshold, such as 0.8, the system automatically triggers a deviation detection mechanism. This mechanism is essentially an automated process used to deeply analyze the differences after model overlay. The deviation detection mechanism first scans all joint data in the comparison results. For example, in martial arts practice, if the spinal position of the practitioner's digital twin model deviates from the standard template by more than a threshold, the deviation detection mechanism activates a dedicated module. This module extracts key indicators from the overlay data, including joint angle deviation and movement rhythm error indicators. Joint angle deviation refers to the difference between the actual joint rotation angle and the standard angle. The principle is to compare the joint vectors of the two models using vector calculation methods, such as calculating the Euler angle difference of the hip joint. The movement rhythm error indicator involves comparing time series, such as the deviation between the speed of movement completion and the standard speed, quantified by analyzing the inter-frame interval. For example, in a martial arts training scenario, assuming the matching degree is 0.7, which is below the threshold, the deviation detection mechanism extracts data from the comparison results. First, it identifies the elbow joint angle deviation. If the elbow joint angle in the standard template is 150 degrees during a martial arts movement, while the practitioner's is 130 degrees, then this 20-degree deviation value is extracted. Simultaneously, for movement rhythm errors, the deviation detection mechanism checks the duration of the entire martial arts movement sequence. If the standard is 2 seconds while the practitioner's is 2.5 seconds, then a rhythm error of 0.5 seconds is calculated. The extraction process for these indicators relies on a data parsing tool, which iterates through the list of joint nodes in the comparison results, calculating the deviation one by one.

[0107] Step S320: Calculate the quantitative deviation values ​​of the joint angle deviation and movement rhythm error index based on the joint angle deviation and movement rhythm error index to obtain the deviation analysis results.

[0108] The quantitative deviation value of the joint angle deviation is obtained by the following formula:

[0109] (11)

[0110] In formula (11), The quantized deviation value represents the joint angle deviation. Indicates the total number of joints. Indicates the first The weighting coefficient of each joint, Indicates the first The actual angle of each joint Indicates the first Reference standard angles for each joint, Indicates the first The maximum allowable angle range of each joint. The control logic of formula (11) is to perform "normalized weighted summation" on the angle deviations of multiple joints to obtain the overall quantitative deviation value of the joint angle. The core is to combine "joint weight" and "maximum allowable range" to more reasonably evaluate the overall deviation of the joint angle.

[0111] The quantitative deviation value of the movement rhythm error index is obtained by the following formula:

[0112] (12)

[0113] In formula (12), This represents the quantitative deviation value of the movement rhythm error index. Indicates the time deviation weighting factor. Indicates the actual execution time of the action. Indicates the standard motion time. This represents the frequency deviation weighting factor. Indicates the number of beat segments. Indicates the first The actual frequency of the segment, Indicates the first The reference frequency of the segment. The control logic of formula (12) is to obtain the comprehensive quantitative deviation value of the action rhythm by weighted fusion of the two dimensions of "total time deviation" and "segment frequency deviation". The core is to take into account both "overall duration deviation" and "segment rhythm deviation" to more comprehensively evaluate the standardization of the action rhythm.

[0114] Based on the extracted joint angle deviation and movement rhythm error indices, the system uses machine learning algorithms to calculate the quantified deviation values ​​of these indices, thereby obtaining deviation analysis results. The machine learning algorithm here can be a support vector machine (SVM) model, used for classification and quantification of the degree of deviation. The principle is that the input deviation data is used as a feature vector, and after training, the SVM model outputs a quantified score, for example, mapping the angle deviation to a range of 0 to 1. In martial arts training, if a knee joint angle deviation of 15 degrees and a rhythm error of 0.3 seconds are extracted, the machine learning algorithm will input these into a pre-trained model, calculating a comprehensive quantified deviation value, such as 0.65, representing a moderate level of deviation, through feature weighting. This process involves multiple iterations to optimize accuracy. For example, in martial arts practice, the machine learning algorithm first standardizes the indices, such as converting the angle deviation into a percentage, then uses a neural network layer to process the rhythm error, combining both to generate analysis results, such as reporting overall movement incoordination caused by knee joint deviation. This quantification process ensures that the results can be used for subsequent guidance.

[0115] Furthermore, the method for martial arts movement instruction and digital dissemination based on digital twins proposed in this embodiment includes step S400:

[0116] Step S410: Extract information on high-deviation joints whose quantitative deviation values ​​exceed a preset deviation threshold from the deviation analysis results. If the number of joints in the extracted high-deviation joint information is lower than a preset number threshold, a secondary filter is performed using a data filtering tool to obtain a list of joints that meet the labeling requirements.

[0117] The following formula is used to define the criteria for judging high deviation joints:

[0118] (13)

[0119] In formula (13), Indicates the first This is an indicator of whether a joint is a high-deviation joint. Indicates the first The quantitative deviation values ​​of joint angle deviation and movement rhythm error indicators. This represents a preset deviation threshold. When the quantitative deviation values ​​of joint angle deviation and movement rhythm error indicators exceed the threshold, the joint is marked as a high deviation joint. The control logic of formula (13) is to perform a binary classification of each joint as "high deviation / non-high deviation" by "comparing the deviation value with the threshold". The core is to quickly screen out the joints that need to be corrected. In martial arts training, it is necessary to quickly locate the "joints that need to be corrected" (for example, when there are multiple joint deviations, high deviation joints are processed first). Formula (13) realizes the automatic marking of high deviation joints through simple threshold judgment, which is convenient for subsequent targeted generation of correction suggestions (for example, for...). The elbow joint, with a focus on guiding its angle adjustment.

[0120] The following formula is used to determine whether the number of extracted high-deviation joints is lower than a preset threshold, and whether secondary filtering is needed:

[0121] (14)

[0122] In formula (14), Indicates the condition for determining quantity. This indicates the number of joints with high deviation. The preset threshold number is indicated. The control logic of formula (14) is to determine whether high-deviation joints need to be filtered a second time by comparing the actual number of high-deviation joints with the preset threshold. The core is to avoid the situation where the number of high-deviation joints is too small (which may lead to missed judgments). In martial arts movements, joint deviations are often "multi-joint linkages" (for example, knee joint deviation may be accompanied by hip joint deviation). If the number of high-deviation joints is too small, it may be due to the threshold being set too strictly, resulting in missed judgments. Formula (14) ensures that no joints that need to be corrected are missed by judging the number of joints, thus improving the comprehensiveness of deviation recognition.

[0123] The following formula describes the process of secondary filtering of high-deviation joint areas using data filtering tools:

[0124] (15)

[0125] In formula (15), This represents the list of joint parts that meet the labeling requirements after secondary filtering. This represents the filtering function of the data filtering tool. This indicates the initial high deviation joint location. The filtering conditions for labeling requirements are indicated. The control logic of formula (15) is based on the preset labeling requirements, and performs secondary filtering on the initial high deviation joint set to obtain the final joint list that meets the requirements. The core is to refine the filtering rules and correct the deviation of the initial identification. In martial arts training, there is a clear labeling priority for "joints that need to be corrected" (such as giving priority to the core joints of force exertion). Formula (15) can supplement the joints that were missed by secondary filtering and remove irrelevant joints, so that the final high deviation joint list is more in line with the actual needs of training guidance.

[0126] The process of extracting information on high-deviation joints from deviation analysis results involves parsing the previously calculated quantified deviation values. Deviation analysis results are typically a structured dataset containing quantified deviation scores for each joint. For example, in martial arts training, if the results show a deviation value of 0.75 for the shoulder joint and 0.85 for the knee joint, these scores are comprehensive indicators output by machine learning algorithms. During extraction, the system traverses the dataset, filtering out joints with scores higher than a specific threshold. For instance, if the threshold is set to 0.7, the shoulder and knee joints are extracted as high-deviation information. This extraction process relies on a data query module that uses an SQL-like query language to filter data. For example, the query statement might specify "SELECT joint_name FROM deviation_results WHERE score>0.7," resulting in a list of joints such as the shoulder and knee, ensuring that subsequent processing focuses on the problem area.

[0127] In a martial arts training scenario, assuming the deviation analysis results show a hip joint deviation of 0.82 and an ankle joint deviation of 0.65, if the preset threshold is 0.7, the hip joint is extracted as a high-deviation area, while the ankle joint is not extracted because it is below the threshold. Next, if the extracted high-deviation joint information is below the preset threshold (which may refer to the number of extracted joints or the average deviation level, e.g., fewer than three), a secondary filtering is triggered, processed by a data filtering tool. This data filtering tool is a rule-based filter that applies multiple layers of conditions for filtering. For example, it first sorts by deviation value, then excludes noisy data (e.g., ignoring temporary deviations below 0.1), ultimately obtaining a list of joints that meet the annotation requirements. For instance, filtering from the initially extracted two joints might yield a list of one core joint.

[0128] Step S420: Based on the list of joint parts, use image processing tools to generate an initial visualization feedback report, perform preliminary annotation for high deviation parts, and determine the annotation content.

[0129] The following formula is used to define the criteria for initial labeling of high-deviation areas:

[0130] (16)

[0131] In formula (16), Indicates the first The labeling status of each joint is set; a value of 1 indicates that labeling is required, and a value of 0 indicates that labeling is not required. Indicates the first Evaluation values ​​of the degree of deviation of joint angle deviation and movement rhythm error indicators. This represents the deviation threshold parameter that triggers annotation. The control logic of formula (16) determines whether a single joint needs to be annotated by comparing the deviation assessment value with the annotation threshold. The core is to generate a "annotated / unannotated" status label for joints with high deviation. Annotation in martial arts training needs to focus on "joints with sufficiently large deviations". Formula (16) achieves automated screening of annotation objects through simple threshold judgment, avoiding redundant annotation of joints with slight deviations, and allowing subsequent correction guidance to focus more on key issues.

[0132] The process of generating an initial visual feedback report based on a list of joints involves the synthesis of digital images. Image processing tools, such as variants of the OpenCV library, first load an image of the practitioner's digital model. Then, they perform preliminary annotations on high-deviation areas in the list, such as the elbow joint. These annotations include text labels and color highlighting; for example, circling the elbow joint in red and labeling it "Deviation angle: 18 degrees." The annotations are determined by referencing the quantified deviation values ​​from deviation analysis results to ensure accurate reflection of the problem. In martial arts training, if the list includes the wrist joint, the image processing tool generates a report image with arrows overlaid at the joint location to indicate the direction of deviation. This preliminary annotation helps users intuitively understand the problem.

[0133] Step S430: Use an image rendering tool to overlay the pre-established standard motion trajectory data onto the initial visualization feedback report, obtain the overlaid trajectory comparison view, and determine whether it meets the display requirements.

[0134] The following formula is used to define the criteria for meeting the display requirements:

[0135] (17)

[0136] In formula (17), This indicates that the display requires the judgment result to be met. This represents the total number of key points on the trajectory. Indicates the first The position coordinates of key points on a standard trajectory. Indicates the first The location coordinates of key points on the visualized trajectory. The maximum allowable deviation threshold is indicated. When the average deviation is less than or equal to the threshold, it is judged to meet the display requirements. The control logic of formula (17) determines whether the visualized trajectory meets the display accuracy requirements by using the "average position deviation of the trajectory key points". The core is to verify the "degree of fit between the visualized trajectory and the standard trajectory". Visual guidance for martial arts training requires the trajectory display to be accurate enough (otherwise it will mislead the practitioners). Formula (17) verifies the "average deviation of the trajectory key points" to ensure that the movement trajectory displayed on the visualization interface is consistent with the standard trajectory, and avoids the training effect being affected by display errors.

[0137] The image rendering tool overlays pre-established standard motion trajectory data onto the initial visualization feedback report to obtain a comparison view of the overlaid trajectory. The image rendering tool works by using a graphics API (Application Programming Interface) such as WebGL (Web Graphics Library) for layered rendering. For example, the standard trajectory is overlaid as a semi-transparent layer on the practitioner's image. The trajectory data is a pre-stored time-series point set, such as a curve from the start point to the end point. After overlaying, it is judged whether it meets display requirements, such as a clarity higher than 80% and no overlapping blur. If it meets these requirements, the process continues.

[0138] Step S440: If the superimposed trajectory comparison view meets the display requirements, the report integration tool is used to adjust the format of the superimposed trajectory comparison view and the annotation content to obtain the annotation feedback structure.

[0139] The annotated feedback structure is derived using the following formula:

[0140] (18)

[0141] In formula (18), This indicates the final feedback structure after processing by the report integration tool. Indicates time The display requires a weight function. This indicates the overlay trajectory comparison view in time. The state value, Indicates the time period of the marked content. Formatting adjustment parameters, and These represent the start and end times of the processing, respectively. The control logic of formula (18) is to integrate and fuse "display weight, trajectory view status, and annotation format" in the time dimension to generate a complete annotation feedback structure. The core is to integrate the feedback elements (display priority, trajectory comparison, and annotation format) at different time points into continuous final feedback.

[0142] If the overlaid trajectory comparison view meets the display requirements, a report integration tool is used to adjust the format of the overlaid trajectory comparison view and the annotation content to obtain the annotated feedback structure. The report integration tool is a template-based processor that adjusts the layout, such as placing the view in the center and the annotations on the side. Format adjustments include resolution unification and font standardization. For example, in martial arts, the final structure is a PDF report containing a comparison view showing spinal trajectory deviation and annotations such as "suggested posture adjustment." This process ensures that the feedback structure is easy for users to interpret and apply to actual training improvements.

[0143] Furthermore, the method for martial arts movement instruction and digital dissemination based on digital twins proposed in this embodiment includes step S500:

[0144] Step S510: Generate a correction suggestion sequence based on the labeled feedback structure, and use a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence.

[0145] The following formula is used to evaluate the effect of the proposed sequence order adjustment:

[0146] (19)

[0147] In formula (19), Indicates the first The score of a dynamically presented sequence. Indicates the first The intensity of each feedback, Indicates the variability of the feedback. This represents the weighting coefficient for the feedback. This represents the total number of dynamically presented sequences. The control logic of formula (19) is to perform a weighted summation of the "feedback intensity and variability" of the revised suggestion sequences to quantify the effect score after the sequence order adjustment. The core is to evaluate the rationality of the sequence by "strengthening high-value feedback and weakening redundant feedback". Martial arts revision suggestions need to be "sorted by importance and redundancy reduced". Formula (19) quantifies the feedback value by "intensity-variability" and then combines it with weighted summation to evaluate whether the sequence "prioritizes the display of high-value feedback and weakens redundant content", which helps to select the optimal revision suggestion presentation order.

[0148] The process of generating a sequence of corrective suggestions based on the annotated feedback structure unfolds by analyzing video data of users practicing martial arts movements. For example, assuming a practitioner is learning the "Cloud Hands" movement in Tai Chi, the system first annotates the video, identifying key points such as the arm's angle, hand position, and body balance. If the annotated feedback indicates an arm angle deviation of 15 degrees, the system generates a series of corrective suggestions, such as "first adjust to relax the shoulders, then slowly raise the arms to chest level." These corrective suggestions form a sequence, ensuring that the system guides the practitioner step-by-step from basic adjustments to advanced optimizations.

[0149] When adjusting the dynamic presentation order of correction suggestion sequences using a sequence optimization algorithm, a priority-based algorithm is employed for dynamic adjustment. Specifically, this sequence optimization algorithm is similar to a variant of a genetic algorithm; it calculates the weight of each suggestion based on the learner's historical data and real-time feedback. For example, if the learner repeatedly makes mistakes in balance, the sequence optimization algorithm will prioritize placing balance-related suggestions at the beginning of the sequence. Through iterative optimization, the presentation order of the correction suggestion sequences changes from a static list to a dynamic path. For instance, the initial sequence might be "shoulder relaxation - arm raise - hand turn," but if user fatigue is detected after optimization, it might be adjusted to "rest first - shoulder relaxation - arm raise," thereby improving learning efficiency. This adjustment not only considers the severity of the deviation but also incorporates user preferences, ensuring that the correction suggestion sequences are more personalized.

[0150] Step S520: Simultaneously package the standardized martial arts movement template, practitioner movement data, deviation analysis results, and adjusted correction suggestions into a standardized digital dissemination package, supporting output and dissemination through multiple formats such as short videos, 3D interactive demonstrations, and online courses.

[0151] The following formula is used to generate the revised proposal sequence, supporting propagation of multiple output forms:

[0152] (20)

[0153] In formula (20), Indicates the first Adjusted motion suggestion vectors for frames. Indicates the template action in the first position. The feature vector of a frame, Indicates the practitioner is in the first The action feature vector of the frame, The adjustment coefficient is used to balance the integration of standardized martial arts movement templates and practitioners' movements. The control logic of formula (20) is to generate personalized correction suggestion movement vectors that are adapted to practitioners by "blending the differences between template movements and practitioners' movements". The core is to make a balance adjustment between "standard templates" and "practitioners' current movements". Martial arts correction suggestions need to "both conform to the standard and adapt to the individual differences of practitioners" (for example, practitioners with limited limbs cannot fully align with the template). Formula (20) adjusts the coefficients to achieve this. It generates "personalized standard movement suggestions"—correcting deviations while avoiding requiring trainees to perform movements beyond their capabilities, thus improving the feasibility of corrective suggestions.

[0154] The process of simultaneously encapsulating standardized martial arts movement templates, practitioner movement data, deviation analysis results, and adjusted correction suggestion sequences into a standardized digital dissemination package involves data packaging technology. The standardized martial arts movement template refers to a predefined ideal movement model, such as the standard 3D skeletal framework of Tai Chi, including joint coordinates and movement trajectories. Practitioner movement data is collected from sensors, such as real-time joint positions captured using a mobile phone camera. Deviation analysis results are quantitative indicators derived by comparing the template with actual data; for example, calculating the Euclidean distance to determine an arm deviation of 0.2 meters. These elements, along with the optimized suggestion sequences, are then encapsulated into a standardized digital dissemination package in JSON format, supporting multiple output formats. This standardized digital dissemination package can generate short videos showcasing the superimposed animation of the template movement and the user's deviation; it can also be transformed into a 3D interactive demonstration, allowing users to rotate and view the deviation in a VR (Virtual Reality) environment; and it can be integrated into online courses as an interactive module for students to download. Through this encapsulation, the standardized digital dissemination package achieves cross-platform compatibility, ensuring efficient digital dissemination.

[0155] Step S530: Simultaneously extract user interaction data from the visual feedback report, dynamically update the practitioner's digital twin model and the dissemination content library, and achieve synergistic optimization of precise movement guidance and efficient digital dissemination.

[0156] The following formula is used to balance the contributions of feedback data and current model parameters, thereby achieving dynamic optimization of the model:

[0157] (twenty one)

[0158] In formula (21), This represents the updated parameters of the digital twin model. This represents the user interaction data impact factor extracted from the visual feedback report. Indicates the parameters of the current digital twin model. This indicates the updating of weight coefficients. The control logic of formula (21) is to achieve dynamic optimization of the digital twin model through "weighted fusion of feedback data and current parameters". The core is to balance "new information from user feedback" and "stability of the existing model". In martial arts training, the digital twin needs to "update with the changes in the practitioner's movements without frequent fluctuations". Formula (21) uses weights to achieve this. Dynamically balancing the "real-time feedback" and the "stability of the model"—for example, reducing the timeliness of feedback when the learner's movement deviation is small. Retain the current model; increase the value if the bias is large. Quickly adapt to new actions.

[0159] The following formula can be used to optimize the selection and updating of the content library and improve the effectiveness of digital communication:

[0160] (twenty two)

[0161] In formula (22), Indicators representing the efficiency of digitally disseminated content. Indicates the first Response rate of each piece of content disseminated. Indicates the first The weight of each piece of content, The total number of contents in the content library is represented by a weighted average, and the dissemination efficiency is calculated. The control logic of formula (22) is to quantify the efficiency of digital dissemination by "weighted average of the response rate of the disseminated content", thereby guiding the selection and updating of the content library. The core is to highlight the contribution of high-value content. Digital dissemination of martial arts needs to prioritize the retention / updating of "high response, high weight" content (such as "Warrior Second Form Correction Tutorial" which has a high response rate and high weight). Formula (22) quantifies efficiency by weighted average, which can guide the optimization of the content library: if the weighted contribution value of a certain content is low, it can be replaced with higher quality content, thereby improving the overall dissemination effect.

[0162] The process of extracting user interaction data from visual feedback reports and dynamically updating the learner's digital twin model and the content library is driven by machine learning models. For example, the visual feedback report is a system-generated chart showing user interaction data such as click frequency and pause duration during practice. After extracting this data, the learner's digital twin model—a virtual 3D human replica simulating the user's body shape and movement habits—is dynamically updated. For instance, if the report shows that the user frequently neglects leg posture, the learner's digital twin model will enhance the accuracy of leg simulation. Simultaneously, the content library—a database storing movement templates and suggestions—is updated by adding new data, such as videos of users' improved movements, enabling content iteration. In this way, precise movement guidance and efficient digital dissemination are synergistically optimized. For example, the system can provide more targeted guidance videos during the user's next practice session, reducing repetitive errors and ultimately improving overall learning effectiveness and expanding dissemination influence.

[0163] Please see Figure 2This embodiment provides a martial arts movement guidance and digital dissemination system based on digital twins, used to execute the aforementioned martial arts movement guidance and digital dissemination method based on digital twins. It includes a standardized limb joint data acquisition module 10, a joint position matching degree determination module 20, a deviation analysis result acquisition module 30, a feedback structure determination module 40, and a movement guidance and digital dissemination collaborative optimization module 50. The standardized limb joint data acquisition module 10 acquires three-dimensional joint coordinate data and motion parameter data from the practitioner's limbs using motion sensor devices. It processes the joint three-dimensional coordinate data and motion parameter data using a preset coordinate transformation algorithm to form a unified input sequence, obtaining standardized limb joint data. The joint position matching degree determination module 20 constructs a digital twin model of the practitioner based on the standardized limb joint data, simultaneously acquires a standard template model of the corresponding martial arts movement from a preset database, and uses a three-dimensional mapping algorithm to superimpose and compare the practitioner's digital twin model and the standard template model to determine the preliminary joint position matching degree. The deviation analysis result acquisition module 30 is used to determine the preliminary joint position matching degree... If the value is below a preset threshold, a deviation detection mechanism is triggered. This mechanism extracts joint angle deviation and movement rhythm error indices from the comparison results between the practitioner's digital twin model and the standard template model. It calculates the quantified deviation values ​​of these indices to obtain the deviation analysis results. The feedback structure determination module 40 extracts information on high-deviation joints from the deviation analysis results, where the quantified deviation value exceeds a preset deviation threshold. It generates a visual feedback report to mark these high-deviation joints and overlays the standard movement trajectory onto the report using image rendering technology to determine the marked feedback structure. The movement guidance and digital dissemination collaborative optimization module 50 generates a correction suggestion sequence based on the marked feedback structure. It uses a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence and simultaneously encapsulates the standardized martial arts movement template, practitioner movement data, deviation analysis results, and the adjusted correction suggestion sequence into a standardized digital dissemination package. This package supports dissemination through short videos, 3D interactive demonstrations, and online courses. It also extracts user interaction data from the visual feedback report to dynamically update the practitioner's digital twin model and the dissemination content library.

[0164] The digital twin-based martial arts movement instruction and digital dissemination method and system provided in this embodiment, compared with existing technologies, acquires three-dimensional joint coordinates and motion parameter data from the practitioner's limbs using motion sensors. A preset coordinate transformation algorithm processes the data to form a standardized input sequence, resulting in standardized limb joint data. This constructs a digital twin model of the practitioner. Simultaneously, a standard template model of the corresponding martial arts movement is retrieved from a preset database. A three-dimensional mapping algorithm is used to overlay and compare the two models to determine the initial joint position matching degree. If the matching degree is lower than a preset threshold, a deviation detection mechanism is triggered. Joint angle deviation and movement rhythm error indicators are extracted from the comparison results, and quantified values ​​are calculated using machine learning algorithms to obtain the deviation analysis results. Further, information on high-deviation joints is extracted from the results, and a visual feedback report is generated to mark these high-deviation areas. Standard movement trajectories are overlaid using image rendering technology to determine the marked feedback structure. Based on this feedback structure, a correction suggestion sequence is generated, and a sequence optimization algorithm is used to adjust the dynamic presentation order. Simultaneously, standardized martial arts movement templates, practitioner movement data, deviation analysis results, and the adjusted correction suggestion sequence are packaged into a standardized digital dissemination package. This package supports dissemination through short videos, 3D interactive demonstrations, and online courses. User interaction data is extracted from the visual feedback report to dynamically update the practitioner's digital twin model and the dissemination content library, achieving synergistic optimization of precise movement guidance and efficient digital dissemination. This embodiment achieves real-time deviation detection, precise feedback, and dynamic optimization of martial arts movements, improving training accuracy and safety, and significantly enhancing the overall guidance effect. The construction of a digital twin of a martial arts practitioner allows for 3D reconstruction and virtual display of martial arts movements, supporting recording, playback, and immersive interactive experiences, forming a visual digital martial arts performance. This not only improves the standardization level of martial arts teaching and training but also provides an innovative technological path for the digital presentation, immersive dissemination, and cultural inheritance of traditional martial arts, possessing high educational and cultural promotion value.

[0165] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A method for martial arts movement instruction and digital dissemination based on digital twins, characterized in that, Includes the following steps: S100. Obtain joint three-dimensional coordinate data and motion parameter data from the exerciser's limbs through a motion sensor device, and process the joint three-dimensional coordinate data and motion parameter data using a preset coordinate transformation algorithm to form a uniform input sequence to obtain standardized limb joint data. S200. Construct a digital twin model of the practitioner based on the standardized limb joint data, simultaneously obtain the standard template model of the corresponding martial arts movement from the preset database, and use a three-dimensional mapping algorithm to superimpose and compare the digital twin model of the practitioner and the standard template model to determine the preliminary joint position matching degree. S300. If the initial joint position matching degree is lower than the preset threshold, the deviation detection mechanism is triggered. The joint angle deviation and movement rhythm error index are extracted from the comparison results between the exerciser's digital twin model and the standard template model. The quantitative deviation values ​​of the joint angle deviation and movement rhythm error index are calculated to obtain the deviation analysis results. S400. Obtain information on high-deviation joints whose quantitative deviation values ​​exceed a preset deviation threshold from the deviation analysis results, generate a visual feedback report to mark the high-deviation joints, and use image rendering technology to overlay the standard motion trajectory onto the visual feedback report to determine the marked feedback structure. S500. Generate a correction suggestion sequence based on the labeled feedback structure, and use a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence. Simultaneously, encapsulate the standardized martial arts movement template, practitioner movement data, deviation analysis results, and the adjusted correction suggestion sequence into a standardized digital dissemination package, which supports output and dissemination through multiple forms such as short videos, 3D interactive demonstrations, and online courses. At the same time, extract user interaction data from the visualized feedback report and dynamically update the practitioner digital twin model and dissemination content library.

2. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 1, characterized in that, Step S100 includes: S110. Obtain joint three-dimensional coordinate data and motion parameter data from the exerciser's limbs through motion sensor devices; S120. A preset coordinate transformation algorithm is used to process the joint three-dimensional coordinate data and motion parameter data, converting the three-dimensional coordinate data and motion parameter data from different sources into a unified format input sequence to obtain standardized limb joint data.

3. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 1, characterized in that, Step S200 includes: S210. Based on standardized limb joint data, use graphics rendering tools to generate a digital twin model of the practitioner; S220. Obtain the standard template model of the corresponding martial arts movement from the preset database, and use a data synchronization tool to align the practitioner's digital twin model with the standard template model on the timeline. S230. A three-dimensional mapping algorithm is used to spatially overlay the time-axis-aligned digital twin model of the trainee with the standard template model to determine the preliminary joint position matching degree.

4. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 1, characterized in that, Step S300 includes: S310. If the joint position matching degree is lower than the preset threshold, the deviation detection mechanism is triggered to extract the joint angle deviation and movement rhythm error index from the comparison results between the practitioner's digital twin model and the standard template model. S320. Based on the joint angle deviation and the movement rhythm error index, calculate the quantitative deviation values ​​of the joint angle deviation and the movement rhythm error index to obtain the deviation analysis results.

5. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 1, characterized in that, Step S400 includes: S410. Extract information on high-deviation joints whose quantitative deviation values ​​exceed a preset deviation threshold from the deviation analysis results. If the number of joints in the extracted high-deviation joint information is lower than a preset number threshold, perform secondary filtering using a data filtering tool to obtain a list of joints that meet the labeling requirements. The following formula is used to define the criteria for judging high deviation joints: ； in, Indicates the first This is an indicator of whether a joint is a high-deviation joint. Indicates the first The quantitative deviation values ​​of joint angle deviation and movement rhythm error indicators. This indicates a preset deviation threshold. When the quantified deviation values ​​of the joint angle deviation and movement rhythm error indicators exceed the threshold, the joint is marked as a high deviation joint. The following formula is used to define the criteria for secondary filtering: ； in, Indicates the condition for determining quantity. This indicates the number of joints with high deviation. This indicates a preset quantity threshold; The following formula describes the process of secondary filtering of high-deviation joint areas using data filtering tools: ； in, This represents the list of joint parts that meet the labeling requirements after secondary filtering. This represents the filtering function of the data filtering tool. This indicates the initial high deviation joint location. Indicates the filtering criteria for annotation requirements; S420. Based on the list of joint parts, use image processing tools to generate an initial visualization feedback report, perform preliminary annotation for high deviation parts, and determine the annotation content; S430. Use an image rendering tool to overlay the pre-established standard motion trajectory data onto the initial visualization feedback report, obtain the overlaid trajectory comparison view, and determine whether it meets the display requirements. S440. If the superimposed trajectory comparison view meets the display requirements, the report integration tool is used to adjust the format of the superimposed trajectory comparison view and the annotation content to obtain the annotation feedback structure.

6. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 5, characterized in that, In step S420, the following formula is used to define the judgment criteria for preliminary annotation of high deviation areas: ； in, Indicates the first The labeling status of each joint is set; a value of 1 indicates that labeling is required, and a value of 0 indicates that labeling is not required. Indicates the first Evaluation values ​​of the degree of deviation of joint angle deviation and movement rhythm error indicators. This represents the deviation threshold parameter that triggers the annotation.

7. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 6, characterized in that, In step S430, the following formula is used to define the judgment condition that meets the display requirements: ； in, This indicates that the display requires the judgment result to be met. This represents the total number of key points on the trajectory. Indicates the first The position coordinates of key points on a standard trajectory. Indicates the first The location coordinates of key points on the visualized trajectory. This indicates the maximum allowable deviation threshold. When the average deviation is less than or equal to the threshold, the display requirements are considered met.

8. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 7, characterized in that, In step S440, the annotated feedback structure is obtained using the following formula: ； in, This indicates the final feedback structure after processing by the report integration tool. Indicates time The display requires a weight function. This indicates the overlay trajectory comparison view in time. The state value, Indicates the time period of the marked content. Formatting adjustment parameters, and These represent the start and end times of the processing, respectively.

9. The method for martial arts movement instruction and digital dissemination based on digital twins according to claim 1, characterized in that, Step S500 includes: S510. Generate a correction suggestion sequence based on the labeled feedback structure, and use a sequence optimization algorithm to adjust the dynamic presentation order of the correction suggestion sequence; S520 simultaneously encapsulates standardized martial arts movement templates, practitioner movement data, deviation analysis results, and adjusted correction suggestions into a standardized digital dissemination package, supporting output and dissemination through multiple formats such as short videos, 3D interactive demonstrations, and online courses. S530. Simultaneously, user interaction data is extracted from the visualized feedback report, and the practitioner's digital twin model and dissemination content library are dynamically updated to achieve synergistic optimization of precise movement guidance and efficient digital dissemination.

10. A martial arts movement instruction and digital dissemination system based on digital twins, used to execute the martial arts movement instruction and digital dissemination method based on digital twins as described in any one of claims 1 to 9, characterized in that, include: The standardized limb joint data acquisition module (10) is used to acquire three-dimensional coordinate data and motion parameter data of the joints from the exerciser's limbs through a motion sensor device, and to process the three-dimensional coordinate data and motion parameter data of the joints using a preset coordinate transformation algorithm to form an input sequence in a unified format to obtain standardized limb joint data. The joint position matching degree determination module (20) is used to construct a digital twin model of the practitioner based on the standardized limb joint data, synchronously obtain the standard template model of the corresponding martial arts movement from the preset database, and use a three-dimensional mapping algorithm to superimpose and compare the digital twin model of the practitioner and the standard template model to determine the preliminary joint position matching degree. The deviation analysis result acquisition module (30) is used to trigger the deviation detection mechanism if the initial joint position matching degree is lower than the preset threshold, extract the joint angle deviation and movement rhythm error index from the comparison results of the exerciser digital twin model and the standard template model, calculate the quantitative deviation value of the joint angle deviation and movement rhythm error index, and obtain the deviation analysis result. The feedback structure determination module (40) is used to obtain information on high deviation joints whose quantitative deviation values ​​exceed a preset deviation threshold from the deviation analysis results, generate a visual feedback report to mark the high deviation joints, and use image rendering technology to overlay the standard motion trajectory on the visual feedback report to determine the marked feedback structure. The motion guidance and digital dissemination collaborative optimization module (50) is used to generate a correction suggestion sequence based on the labeled feedback structure, and to adjust the dynamic presentation order of the correction suggestion sequence using a sequence optimization algorithm. Simultaneously, it encapsulates the standardized martial arts motion template, practitioner motion data, deviation analysis results, and the adjusted correction suggestion sequence into a standardized digital dissemination package, which supports output and dissemination through short videos, 3D interactive demonstrations, and online courses. At the same time, it extracts user interaction data from the visualized feedback report and dynamically updates the practitioner digital twin model and dissemination content library.

Images