Machine learning based hypermobility monitoring management method and system

Abstract

The application provides a kind of hyper-motion monitoring management method and system based on machine learning, it is related to motion monitoring management technical field, first based on hyper-motion monitoring scene demand, establish the associated mapping relationship of hyper-motion monitoring demand (include hyper-motion type and state evaluation demand) and machine learning model module (include feature analysis and state prediction module), then obtain the monitoring data set containing motion posture and trajectory data, carry out hierarchical feature analysis and obtain hierarchical feature set in combination with mapping result, then input hierarchical feature set into corresponding model module to generate monitoring analysis result containing posture evaluation and trajectory prediction, generate management scheme containing adjustment suggestion and safety prompt according to analysis result in combination with basic rule set, finally input management scheme into adaptation module for format and interface matching, and send to target monitoring terminal, realize the intelligent monitoring and management of hyper-motion.

Description

Technical Field

[0001] This invention relates to the field of motion monitoring and management technology, and more specifically, to a hyper-motion monitoring and management method and system based on machine learning. Background Technology

[0002] In high-performance sports such as athletic competitions, extreme sports, and professional training, accurate monitoring and scientific management of athletes' athletic status are crucial. Traditional high-performance monitoring and management methods mainly rely on manual observation and simple instrument measurements. Manual observation suffers from strong subjectivity and is prone to missing key information, making it difficult to comprehensively and accurately capture the complex details of the high-performance process. Simple instrument measurements often only obtain limited data and lack the ability to conduct in-depth analysis and comprehensive processing of the data.

[0003] In terms of data processing and analysis, most existing methods fail to fully consider the unique and complex nature of hyper-motion scenarios. There is a lack of accurate models to support the determination of hyper-motion types and states, making it difficult to provide personalized monitoring and management solutions based on different hyper-motion needs. Furthermore, the lack of effective rules and adaptation mechanisms in translating monitoring and analysis results into practical management measures leads to poor compatibility between management plans and actual monitoring terminals, hindering timely and accurate communication with athletes or relevant management personnel, thus affecting the effectiveness and quality of hyper-motion monitoring and management. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a hyper-motion monitoring and management method based on machine learning, the method comprising: Based on the application requirements of hyper-motion monitoring scenarios, an association mapping relationship between hyper-motion monitoring requirements and machine learning model modules is established, and the association mapping results between hyper-motion monitoring requirements and machine learning model modules are obtained. The hyper-motion monitoring requirements include hyper-motion type monitoring requirements and hyper-motion state evaluation requirements, and the machine learning model modules include hyper-motion feature analysis modules and hyper-motion state prediction modules. A hypermotion monitoring dataset is obtained, and a hierarchical feature parsing process is performed on the hypermotion monitoring dataset. Combined with the correlation mapping results, a hypermotion hierarchical feature set is obtained. The hypermotion monitoring dataset contains motion posture data and motion trajectory data during the hypermotion process, and the hypermotion hierarchical feature set contains posture hierarchical features and trajectory hierarchical features. The hypermotion hierarchical feature set is input into the machine learning model module corresponding to the association mapping result, and the interaction processing between features and model is performed to generate hypermotion monitoring and analysis results. The hypermotion monitoring and analysis results include hypermotion posture evaluation results and hypermotion trajectory prediction results. Based on the results of the hypermotion monitoring analysis and combined with the basic rule set for hypermotion monitoring management, a hypermotion monitoring management plan is generated. The hypermotion monitoring management plan includes hypermotion adjustment suggestions and hypermotion safety tips. The aforementioned hyper-motion monitoring management scheme is input into the hyper-motion monitoring terminal adaptation module, and the scheme format is matched with the terminal interface to obtain the terminal-adapted hyper-motion monitoring management scheme, which is then sent to the target hyper-motion monitoring terminal.

[0005] In another aspect, embodiments of the present invention also provide a hyper-motion monitoring and management system based on machine learning, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0006] Based on the above, this embodiment of the invention establishes a mapping relationship between hyper-motion monitoring requirements and machine learning model modules, accurately locating the core points of hyper-motion monitoring and the points of convergence between machine learning technology. It acquires and merges hyper-motion monitoring datasets for hierarchical feature analysis, and combines the mapping results to obtain a hyper-motion hierarchical feature set. This allows for comprehensive and detailed mining of key information in the hyper-motion data, fully reflecting the complex characteristics of posture and trajectory during hyper-motion. The hierarchical feature set is input into the corresponding machine learning model module for interactive processing, generating hyper-motion monitoring analysis results that include posture assessment and trajectory prediction. The powerful analytical capabilities of machine learning enable accurate understanding of the hyper-motion state. Based on the analysis results and a set of basic rules, a hyper-motion monitoring management scheme is generated. Finally, the hyper-motion monitoring management scheme is input into an adaptation module for format and interface matching and sent to the target terminal, ensuring timely and accurate communication and execution of the management scheme, significantly improving the intelligence level and practical effect of hyper-motion monitoring management. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the execution flow of the hyper-motion monitoring and management method based on machine learning provided in an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of exemplary hardware and software components of the machine learning-based hyper-motion monitoring and management system provided in an embodiment of the present invention. Detailed Implementation

[0009] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a machine learning-based hyper-motion monitoring and management method according to an embodiment of the present invention. The following is a detailed description of the machine learning-based hyper-motion monitoring and management method.

[0010] Step S110: Based on the application requirements of the hyper-motion monitoring scenario, establish the association mapping relationship between hyper-motion monitoring requirements and machine learning model modules, and obtain the association mapping result between hyper-motion monitoring requirements and machine learning model modules. The hyper-motion monitoring requirements include hyper-motion type monitoring requirements and hyper-motion state evaluation requirements. The machine learning model modules include hyper-motion feature analysis modules and hyper-motion state prediction modules.

[0011] In this embodiment, marathon running is used as the hyper-motion monitoring scenario. The application requirements of this hyper-motion monitoring scenario include real-time monitoring of athletes' movement types and assessing their movement status to ensure sports safety. Based on this, the association mapping relationship between hyper-motion monitoring requirements and machine learning model modules is established. The hyper-motion type monitoring requirement aims to accurately identify the athlete's current movement type, such as running on flat ground or running uphill; the hyper-motion status assessment requirement evaluates the athlete's physical condition and whether their movement posture is reasonable. The hyper-motion feature analysis module in the machine learning model module is used to extract and analyze features from the collected movement data, and the hyper-motion status prediction module predicts the athlete's subsequent movement status based on the analyzed features.

[0012] Step S111: Extract the key requirement elements from the application requirements of the hyper-motion monitoring scenario, classify the hyper-motion monitoring requirements based on the key requirement elements, and obtain the hyper-motion monitoring requirement types. The key requirement elements include the real-time requirements of hyper-motion monitoring and the accuracy requirements of hyper-motion monitoring. The hyper-motion monitoring requirement types correspond to hyper-motion type monitoring requirements and hyper-motion state assessment requirements.

[0013] In marathon monitoring scenarios, key requirements are extracted from application needs. Real-time requirements refer to the need for rapid analysis and feedback of athlete data to promptly assess athlete condition; accuracy requirements ensure accurate and reliable identification of movement types and state assessments. Based on these two key requirements, the super-movement monitoring needs are divided into super-movement type monitoring needs and super-movement state assessment needs. For example, in scenarios with high real-time requirements, super-movement type monitoring needs to quickly respond to changes in athlete movement types; while in scenarios with high accuracy requirements, super-movement state assessment needs to more accurately determine the athlete's muscle fatigue level and other states.

[0014] Step S112: Obtain a preset set of machine learning model module function descriptions. The set of machine learning model module function descriptions includes the function descriptions of the hypermotion feature analysis module and the hypermotion state prediction module. The function description of the hypermotion feature analysis module includes the function of feature extraction of hypermotion posture data and the function of feature extraction of hypermotion trajectory data. The function description of the hypermotion state prediction module includes the function of predicting the subsequent hypermotion state based on hypermotion features.

[0015] The pre-defined set of machine learning model module function descriptions is stored in the system. The description of the hyper-motion feature analysis module specifies that it can extract effective posture features from marathon runners' motion posture data, such as joint angle change data, and trajectory features from motion trajectory data, such as center of gravity position change data. The description of the hyper-motion state prediction module indicates that it can predict the athlete's subsequent motion state based on the feature data obtained from the hyper-motion feature analysis module, such as the amplitude of joint angle changes and the trend of center of gravity position changes, including whether there will be movement deformation or a tendency to deviate from the predetermined trajectory.

[0016] Step S113: Perform feature encoding processing on the hyper-motion monitoring requirement type to generate a hyper-motion monitoring requirement feature vector.

[0017] For the two types of hyper-motion monitoring needs—hyper-motion type monitoring needs and hyper-motion state assessment needs—feature encoding processing is performed separately. Key attributes for each type of need, such as the level of real-time requirements and accuracy requirements, are transformed into quantifiable feature values. For example, real-time requirements are divided into different levels and assigned corresponding values, and various accuracy requirements are also transformed into corresponding values. These values ​​are then arranged in a specific order to form a hyper-motion monitoring need feature vector. This hyper-motion monitoring need feature vector can comprehensively reflect the characteristics of hyper-motion monitoring needs.

[0018] Step S114: Perform feature encoding on the functional description of each machine learning model module in the set of machine learning model module functional descriptions to generate a machine learning model module functional feature vector.

[0019] The functional descriptions of the hyper-motion feature analysis module and the hyper-motion state prediction module also undergo feature encoding processing. Functional attributes such as feature extraction speed and the types of features extracted in the hyper-motion feature analysis module are converted into feature values; similarly, functional attributes such as prediction speed and prediction accuracy in the hyper-motion state prediction module are also converted into feature values. Subsequently, these feature values ​​are combined in a specific order to generate functional feature vectors corresponding to each machine learning model module, reflecting the functional characteristics of each module.

[0020] Step S115: Calculate the similarity between the hyper-motion monitoring requirement feature vector and the functional feature vector of each machine learning model module, generate a similarity matrix, and determine the matching relationship between the hyper-motion monitoring requirement type and the machine learning model module based on the similarity matrix.

[0021] The similarity between the feature vectors of hyper-motion monitoring needs and the feature vectors of the hyper-motion feature analysis module and the hyper-motion state prediction module is calculated using a specific algorithm. For example, the cosine similarity between the vectors is calculated, with a higher value indicating a higher similarity. The calculated similarity values ​​are arranged in rows and columns to form a similarity matrix, where the rows represent hyper-motion monitoring need types and the columns represent machine learning model modules. Based on the magnitude of the values ​​in the similarity matrix, it is determined which hyper-motion monitoring need type is most matched with which machine learning model module. For example, if the feature vector of hyper-motion monitoring needs has a high similarity to the feature vector of the hyper-motion feature analysis module, it indicates a high degree of matching between the two.

[0022] Step S116: Set a similarity threshold, and mark the hyper-motion monitoring demand type corresponding to the similarity value in the similarity matrix that is greater than or equal to the similarity threshold and the machine learning model module as a preliminary matching pair.

[0023] Based on the actual needs and experience of marathon monitoring, a suitable similarity threshold is set. The similarity matrix is ​​traversed, and elements with similarity values ​​greater than or equal to the threshold are marked as preliminary matching pairs for their corresponding hyper-motion monitoring requirement types and machine learning model modules. For example, with a threshold of 0.7, if the similarity value between the hyper-motion state assessment requirement and the hyper-motion state prediction module is 0.8, then they are marked as a preliminary matching pair.

[0024] Step S117: Verify the preliminary matching pairs. Check whether the functions of the machine learning model modules in the preliminary matching pairs fully cover the requirements of the corresponding hyper-motion monitoring requirement type. If they fully cover the requirements, the preliminary matching pairs are determined to be the final matching relationships. If they do not fully cover the requirements, the similarity threshold is readjusted or the functional description of the machine learning model modules is supplemented until a matching pair that fully covers the requirements is found and determined to be the final matching relationship.

[0025] For the initially identified matching pairs, a detailed examination is conducted to determine whether the functionality of the machine learning model module can meet all the requirements of the corresponding hyper-motion monitoring need type. For example, for the initial matching pair between the hyper-motion state assessment requirement and the hyper-motion state prediction module, it is checked whether the hyper-motion state prediction module can predict all state assessment requirements of athletes, such as muscle fatigue and joint injury risk. If it can fully cover all requirements, the matching pair is determined as the final matching relationship; if some requirements are not covered, such as the inability to predict joint injury risk, the similarity threshold is readjusted, or the functional description of the hyper-motion state prediction module is supplemented to cover the requirement, and then the matching verification is performed again until a suitable matching pair is found.

[0026] Step S118: Based on the matching relationship, construct an association mapping table between hyper-motion monitoring requirements and machine learning model modules. The association mapping table records at least one machine learning model module corresponding to each hyper-motion monitoring requirement type. The association mapping table is used as the association mapping result between hyper-motion monitoring requirements and machine learning model modules.

[0027] Based on the finalized matching relationships, an association mapping table is constructed. In this table, each hyper-motion monitoring requirement type corresponds to a row, and its corresponding machine learning model module is listed in that row. For example, hyper-motion type monitoring requirements correspond to the hyper-motion feature analysis module, and hyper-motion state assessment requirements correspond to both the hyper-motion feature analysis module and the hyper-motion state prediction module. This association mapping table clearly demonstrates the correspondence between hyper-motion monitoring requirements and machine learning model modules, serving as an important basis for subsequent processing.

[0028] Step S120: Obtain the hypermotion monitoring data set, perform hierarchical feature parsing processing on the hypermotion monitoring data set, and combine the association mapping results to obtain the hypermotion hierarchical feature set. The hypermotion monitoring data set includes motion posture data and motion trajectory data during the hypermotion process, and the hypermotion hierarchical feature set includes posture hierarchical features and trajectory hierarchical features.

[0029] During a marathon, a set of hypermotion monitoring data is acquired through sensor devices worn by athletes. This data includes posture data, such as changes in the angles of leg joints and arm swing angles; and trajectory data, such as changes in the athlete's position on the track and shifts in their center of gravity. Combining the association mapping results obtained in step S110, the types of features to be extracted are identified. Then, these data undergo hierarchical feature parsing processing, ultimately yielding a hypermotion hierarchical feature set containing both posture and trajectory hierarchical features.

[0030] Step S121: Collect raw data during the hypermotion process using a hypermotion data acquisition device, filter and process the raw data to remove invalid data, and obtain a hypermotion monitoring data set. The raw data includes the position data and angle data of the hypermotion object at different times. The hypermotion monitoring data set includes the motion posture data and motion trajectory data during the hypermotion process. The motion posture data includes the angle change data of each joint of the hypermotion object, and the motion trajectory data includes the position change data of the center of gravity of the hypermotion object.

[0031] Using specialized sports data acquisition equipment, such as inertial measurement units and GPS positioning devices, raw data of marathon athletes during their run is collected. This raw data includes the athlete's position at different times, such as latitude and longitude coordinates; and angle data, such as the angles of the knee and hip joints. The raw data is then filtered to remove invalid data caused by equipment malfunctions, signal interference, or other reasons, such as angle values ​​that deviate significantly from the normal range or data with excessively large positional jumps. After filtering, a valid set of hyperkinetic monitoring data is obtained, in which the motion posture data records in detail the angle changes of each joint, and the motion trajectory data reflects the changes in the position of the center of gravity.

[0032] Step S122: Based on the feature requirements of the machine learning model module associated with the hyper-motion type monitoring requirements in the association mapping results, determine the hierarchical dimensions of the hyper-motion monitoring dataset. The hierarchical dimensions include temporal hierarchical dimensions and spatial hierarchical dimensions.

[0033] Referring to the machine learning model modules associated with the hyper-motion type monitoring needs in the correlation mapping results, such as the feature requirements of the hyper-motion feature analysis module, we determine the dimensions for stratifying the hyper-motion monitoring dataset. The temporal stratification dimension divides the data into different time periods according to time intervals to analyze motion characteristics at different times; the spatial stratification dimension divides the data based on spatial location or body part, such as dividing motion posture data according to different joint parts, or dividing motion trajectory data according to different segments of a track.

[0034] Step S123: Based on the time layer dimension, the hypermotion monitoring data set is divided into multiple time layer data units according to the time interval. Each time layer data unit contains motion posture data and motion trajectory data within the corresponding time period.

[0035] Based on the defined time stratification dimension, appropriate time intervals are set, such as 10 seconds per interval. The hyper-motion monitoring dataset is divided according to this time interval, resulting in multiple time-layer data units. Each time-layer data unit contains the athlete's motion posture data and motion trajectory data within that time period, such as knee joint angle changes within a 10-second interval and center of gravity position changes within that time period.

[0036] Step S124: Based on the spatial layering dimension, the motion posture data in each time layer data unit is divided into multiple posture sub-layer data according to the joint parts, and the motion trajectory data is divided into multiple trajectory sub-layer data according to the spatial region.

[0037] Based on each time-level data unit, further subdivisions are made according to spatial layering. For motion posture data, it is divided into multiple posture sub-layers according to joint locations, such as the knee, hip, and ankle joints. Each posture sub-layer specifically records the angle changes of the corresponding joint within that time level. For motion trajectory data, it is divided into multiple trajectory sub-layers according to the spatial regions of the track, such as the starting area, flat areas in the middle, and uphill areas. Each trajectory sub-layer records the changes in the athlete's center of gravity position within the corresponding spatial region.

[0038] Step S125: Perform feature extraction processing on each posture sub-layer data to obtain posture layer features, which include joint angle change amplitude features and joint angle change frequency features.

[0039] For each posture sub-layer data, such as the knee joint posture sub-layer data, feature extraction is performed. The variation range of the joint angle within the corresponding time period of that sub-layer data is calculated, i.e., the difference between the maximum and minimum angle values; simultaneously, the frequency of joint angle changes is calculated, i.e., the number of angle changes per unit time. These extracted features are used as the posture layer features corresponding to that posture sub-layer data.

[0040] Step S1251: Preprocess the joint angle change data in each attitude sub-layer data, extract the joint angle value at each time step from the preprocessed joint angle change data, calculate the difference between the joint angle values ​​at adjacent time steps, and obtain the joint angle change sequence.

[0041] The joint angle variation data in the attitude sub-layer data is preprocessed, including noise removal and data smoothing, to improve data quality. After preprocessing, the joint angle value at each time step is extracted, and the difference between the joint angle values ​​at two adjacent time steps is calculated. These differences are arranged in chronological order to form a joint angle variation sequence, which reflects the change of joint angles over time.

[0042] Step S1252: In the sequence of joint angle changes, find the maximum and minimum values, calculate the difference between the maximum and minimum values, and use this difference as the feature of the joint angle change amplitude.

[0043] Traverse the sequence of joint angle changes, find the maximum and minimum values, and then calculate the difference between them. This difference can intuitively reflect the maximum range of joint angle changes within that time period, and can be used as a feature of the joint angle change amplitude.

[0044] Step S1253: Count the number of consecutive data segments with the same direction of change in the joint angle change sequence to determine the number of cycles of joint angle change.

[0045] Observe the direction of change in the joint angle variation sequence, i.e., whether the angle is increasing or decreasing. Count the number of consecutive data segments with the same direction of change. Each consecutive data segment represents one cycle of joint angle change, thus determining the number of cycles of joint angle change.

[0046] Step S1254: Based on the time length and number of cycles corresponding to the attitude sublayer data, calculate the average time length of each cycle, and use the reciprocal of the average time length as the joint angle change frequency feature.

[0047] Given the time length corresponding to the attitude sublayer data, divide this time length by the number of cycles of joint angle change to obtain the average time length of each cycle. Then, use the reciprocal of the average time length as the joint angle change frequency feature, which reflects the speed of joint angle change.

[0048] Step S1255: Associate and store the joint angle change amplitude feature and the joint angle change frequency feature as the attitude layering feature corresponding to the attitude sub-layer data.

[0049] The extracted joint angle change amplitude features and joint angle change frequency features are associated and stored according to the corresponding posture sub-layer data to ensure that each posture sub-layer data has its corresponding posture layer features, which facilitates subsequent integration and analysis.

[0050] Step S126: Perform feature extraction processing on each trajectory sub-layer data to obtain trajectory layer features, which include the distance feature of the change in center of gravity position and the direction feature of the change in center of gravity position.

[0051] For each trajectory sub-layer data, such as the trajectory sub-layer data for an uphill area, feature extraction is performed. The distance of the change in center of gravity position within the corresponding time period and spatial region of that sub-layer data is calculated, i.e., the straight-line distance between the starting and ending positions; simultaneously, the direction of the change in center of gravity position is determined, such as forward, left, or right. These features are then used as the trajectory layer features corresponding to that trajectory sub-layer data.

[0052] Step S127: Summarize all pose layer features and trajectory layer features to obtain a hypermotion layer feature set.

[0053] The posture-level features extracted from each posture sub-layer data and the trajectory-level features extracted from each trajectory sub-layer data are all aggregated together to form a complete hyper-motion-level feature set. This hyper-motion-level feature set comprehensively includes the posture and trajectory features of athletes during the marathon.

[0054] Step S130: Input the hypermotion hierarchical feature set into the machine learning model module corresponding to the association mapping result, perform feature-model interaction processing, and generate hypermotion monitoring and analysis results. The hypermotion monitoring and analysis results include hypermotion posture evaluation results and hypermotion trajectory prediction results.

[0055] The hypermotion hierarchical feature set obtained in step S120 is input into the corresponding machine learning model modules, namely the hypermotion feature analysis module and the hypermotion state prediction module, according to the guidance of the association mapping results. In the model modules, the feature data interacts with the model for processing. The hypermotion feature analysis module performs in-depth analysis of the features, and the hypermotion state prediction module makes predictions based on the analysis results, ultimately generating a hypermotion monitoring and analysis result that includes hypermotion posture evaluation results and hypermotion trajectory prediction results.

[0056] Step S131: Extract the machine learning model module corresponding to the current hyper-motion monitoring requirement from the association mapping result, and determine the input interface parameter requirements of the machine learning model module. The input interface parameter requirements include the dimension requirements of the feature data and the format requirements of the feature data.

[0057] By reviewing the correlation mapping results, identify the machine learning model modules corresponding to the current marathon monitoring needs, such as the hyper-exercise state assessment needs, including hyper-exercise feature analysis and hyper-exercise state prediction modules. Clarify the input interface parameter requirements for these model modules, including the dimension of the feature data (i.e., the length of the feature vector) and the format of the feature data (e.g., floating-point data type, array storage format, etc.).

[0058] Step S132: Perform format adjustment processing on the hyper-motion hierarchical feature set to make the hyper-motion hierarchical feature set meet the input interface parameter requirements of the machine learning model module, and obtain the adjusted hyper-motion hierarchical feature set.

[0059] Based on the input interface parameter requirements of the machine learning model module, the format of the hyper-motion hierarchical feature set is adjusted. For example, if the model requires the feature data dimension to be of a specific length, the features are augmented or reduced in dimensionality; if the model requires a specific data format, the feature data is converted to the corresponding format, such as converting list-format features to array-format features. After adjustment, a modified hyper-motion hierarchical feature set that meets the requirements is obtained.

[0060] Step S133: Input the adjusted hyper-motion layered feature set into the hyper-motion feature analysis module. The hyper-motion feature analysis module performs correlation analysis on the posture layered features in the adjusted hyper-motion layered feature set to determine the correlation between angle changes between different joint parts, performs continuity analysis on the trajectory layered features to determine the continuous trend of the center of gravity position change, and performs standardization processing on the feature values ​​during the analysis process to ensure scale consistency, thereby obtaining the hyper-motion feature analysis results.

[0061] The adjusted hyper-motion hierarchical feature set is input into the hyper-motion feature analysis module. This module first performs correlation analysis on the posture hierarchical features, such as analyzing the relationship between knee and hip joint angle changes to determine whether their movements are coordinated, thereby identifying the correlation of angle changes between different joint parts. Simultaneously, it performs continuity analysis on the trajectory hierarchical features, observing the changes in the center of gravity position across different time periods and spatial regions to determine its continuous trend, such as whether it is stable forward or has a lateral shift trend. During the analysis, all feature values ​​are standardized to the same scale range to eliminate the dimensional influence between different features, ensuring the accuracy of the analysis results, and finally obtaining the hyper-motion feature analysis results.

[0062] Step S134: Input the hypermotion feature analysis results into the hypermotion state prediction module. The hypermotion state prediction module calls the preset prediction algorithm. Based on the standardized angle change correlation and the continuous trend of center of gravity position change in the hypermotion feature analysis results, it predicts the subsequent posture changes and trajectory changes of the hypermotion object and obtains the hypermotion state prediction results.

[0063] Step S1341: Load a preset prediction algorithm using the hypermotion state prediction module. The prediction algorithm is a prediction algorithm based on temporal features.

[0064] After the hypermotion state prediction module starts, it calls a pre-defined prediction algorithm based on temporal features through an internal interface. This prediction algorithm is pre-stored in the module's algorithm library, and its core architecture includes a feature input layer, a temporal feature extraction layer, a feature fusion layer, and a prediction output layer. During algorithm loading, the module can automatically verify whether the basic parameters on which the algorithm depends are complete, such as the temporal window length and feature dimension adaptation parameters, to ensure that the prediction algorithm can run normally.

[0065] Step S1342: Convert the angular change correlation in the hyper-motion feature analysis results into a correlation matrix, which records the correlation coefficients of angular changes at different joint locations.

[0066] Angle change correlation data were extracted from the hyper-motion feature analysis results, including correlation data for different joint combinations such as knee and hip joints, and ankle and knee joints. These correlation data were then categorized and arranged according to joint location to construct a correlation matrix. The rows and columns of the matrix correspond to different joint locations, and the matrix elements are the correlation coefficients for corresponding joint pairs. These correlation coefficients were obtained by analyzing the degree of coordination in joint angle changes within historical synchronous motion data, reflecting the coordinated relationship of joint movements.

[0067] Step S1343: Convert the continuous trend of the center of gravity position change in the hypermotion feature analysis results into a trend vector, wherein the trend vector records the rate of change of the center of gravity position in different directions.

[0068] To analyze the continuous trend of changes in the center of gravity position, the data is decomposed into three spatial dimensions: front-back, left-right, and up-down. A sliding window process is applied to the data in each dimension, and the average rate of change within each window is calculated. These rate values ​​are then arranged chronologically to form a trend vector. Each element of the trend vector corresponds to the rate of change of the center of gravity in a specific direction within a particular time window, and the vector length is consistent with the preset number of time windows.

[0069] Step S1344: Perform eigenvalue standardization on the correlation matrix and trend vector respectively to obtain standardized correlation matrix and standardized trend vector. Then, input the standardized correlation matrix and standardized trend vector into the feature processing layer of the prediction algorithm for fusion processing to generate fused feature vector.

[0070] For each element of the association matrix, a min-max normalization method is applied to map the values ​​to a specific interval, eliminating scale differences in the association coefficients of different joints. For each dimension of the trend vector, a z-score normalization method is applied, transforming the data based on the mean and standard deviation to ensure the comparability of the rates of change in each direction. The normalized association matrix is ​​then flattened into a one-dimensional vector and concatenated with the normalized trend vector in the order of feature dimensions before being input into the feature processing layer of the prediction algorithm. The feature processing layer performs a non-linear transformation on the concatenated features through a fully connected network, generating a fused feature vector that incorporates comprehensive information on joint associations and center-of-gravity trends.

[0071] Step S1345: Input the fused feature vector into the prediction layer of the prediction algorithm. The prediction layer processes the fused feature vector based on the model parameters trained on the historical hypermotion state data, and predicts the joint angle and center of gravity position of the hypermotion object at multiple subsequent moments.

[0072] The fused feature vectors are fed into the prediction layer of the prediction algorithm. This layer employs a gated recurrent unit structure, using weight and bias parameters obtained from historical training to perform temporal modeling of the features. During computation, the feature input at each time step is dynamically updated by combining the hidden state of the previous time step, and the predicted joint angles and center of gravity positions are calculated sequentially for multiple subsequent time steps. The reasonableness of the output values ​​is checked in real time during prediction; if the predicted value exceeds the physiological limits of motion, an internal correction mechanism is triggered to adjust the output.

[0073] Step S1346: Arrange the predicted joint angle values ​​at multiple subsequent time points in chronological order to obtain the subsequent posture changes of the hyper-motion object.

[0074] The predicted angle values ​​for each joint at multiple subsequent time points are organized chronologically and by joint location. For each joint, a time-angle change sequence is generated, containing the specific angle value and direction of angle change at each predicted time point. The sequence data for all joints are then aggregated to form a record of the subsequent posture changes of the hyper-motion object, including temporal features and joint coordination features of the posture changes.

[0075] Step S1347: Arrange the predicted center of gravity positions at multiple subsequent moments in chronological order to obtain the subsequent trajectory changes of the hyper-moving object.

[0076] The predicted center of gravity positions are organized according to time sequence and spatial coordinate dimensions (x, y, z) to generate a predicted center of gravity trajectory sequence. Each element in the sequence contains the coordinate value at the predicted time, the change in position between that time and the previous time, and the direction vector of the change. Trajectory smoothing is used to eliminate prediction jitter, ensuring continuous and reasonable trajectory changes, ultimately forming a record of the subsequent trajectory changes of the hyper-moving object.

[0077] Step S1348: Integrate the subsequent attitude changes and trajectory changes of the hyper-motion object to obtain the hyper-motion state prediction result.

[0078] The attitude change and trajectory change data are aligned along the time axis to establish a temporal correlation between them. For each prediction moment, the joint angle data and center of gravity position data at that moment are associated to form a spatiotemporally fused state record. During the integration process, a state confidence indicator is added, and a confidence value is assigned to the prediction result at each moment based on the historical prediction error data. Finally, a hyper-motion state prediction result that includes attitude prediction, trajectory prediction, and spatiotemporal correlation is generated.

[0079] Step S135: Integrate the hypermotion feature analysis results and hypermotion state prediction results to generate hypermotion monitoring and analysis results. The hypermotion monitoring and analysis results include hypermotion attitude assessment results and hypermotion trajectory prediction results. The hypermotion attitude assessment results are determined based on the correlation of angle changes, and the hypermotion trajectory prediction results are determined based on the continuous trend of center of gravity position changes and the predicted trajectory changes.

[0080] The angular change correlation data are extracted from the hyperkinetic feature analysis results and combined with the physiological standard range of joint movement to generate hyperkinetic posture assessment results, including joint coordination scores, abnormal posture identification markers, and descriptions of the degree of abnormality. Trajectory change data are extracted from the hyperkinetic state prediction results and combined with the continuous trend of center of gravity position changes to generate hyperkinetic trajectory prediction results, including future trajectory path prediction, trajectory deviation risk assessment, and key node position prediction. The posture assessment results and trajectory prediction results are integrated according to time dimension and feature category, and a result generation timestamp and data source identifier are added to form a complete hyperkinetic monitoring and analysis result.

[0081] Step S140: Based on the hypermotion monitoring analysis results and the set of basic rules for hypermotion monitoring management, generate a hypermotion monitoring management plan, which includes hypermotion adjustment suggestions and hypermotion safety tips.

[0082] The system invokes a set of fundamental rules for hyper-motion monitoring and management, which includes attitude safety rules, trajectory safety rules, and risk assessment rules. Hyper-motion monitoring analysis results are matched against these fundamental rules one by one. For indicators in the attitude assessment results that exceed the safety range, corresponding hyper-motion adjustment suggestions are generated according to the adjustment rules. For risk points in the trajectory prediction results, hyper-motion safety warnings are generated according to the risk warning rules. The adjustment suggestions and safety warnings are prioritized and integrated into a hyper-motion monitoring and management scheme. This scheme includes the applicable scenarios, implementation methods, and expected effects for each suggestion and warning.

[0083] Step S141: Obtain the basic rule set for hyper-motion monitoring and management. The basic rule set includes hyper-motion posture safety rules and hyper-motion trajectory safety rules. The hyper-motion posture safety rules record the safe range of the angles of each joint of the hyper-motion object, and the hyper-motion trajectory safety rules record the safe area of ​​the center of gravity of the hyper-motion object.

[0084] The basic rule set is retrieved from the database through the rule management interface. This basic rule set is stored in a structured format and includes fields such as rule identifier, rule type, applicable conditions, threshold parameters, and processing methods. In the hyper-motion posture safety rules, safe angle ranges are defined for each joint at different stages of motion, such as the safe angle range of the knee joint during running and the maximum extension angle of the hip joint. In the hyper-motion trajectory safety rules, safe zone boundaries for the center of gravity position in different terrains (flat ground, uphill, downhill) are defined, such as the safe distance for left and right offset of the track centerline and the safe rate of change of altitude.

[0085] Step S142: Compare and analyze the hypermotion posture assessment results in the hypermotion monitoring and analysis results with the hypermotion posture safety rules to determine whether the current joint angle of the hypermotion object exceeds the safety range. If it exceeds the safety range, determine the joint parts that need to be adjusted and the adjustment direction, and obtain posture adjustment suggestions.

[0086] Extract the joint angle data from the hyper-motion posture assessment results and compare them with the corresponding safe ranges in the hyper-motion posture safety rules. For each joint, calculate the difference between the actual angle and the upper and lower limits of the safe range. If the difference is positive and exceeds the allowable deviation, it is determined to be outside the safe range. For joints outside the range, determine the adjustment direction (e.g., decreasing the angle, increasing the angle) and reference values ​​for the adjustment range based on the adjustment guidelines in the rules and the correlation of joint movement. At the same time, associate the corresponding muscle exertion adjustment suggestions to form posture adjustment suggestions.

[0087] Step S1421: Extract the real-time angle values ​​and angle change rate data of each joint from the hypermotion posture assessment results, and organize them into a posture assessment dataset according to the joint location.

[0088] The real-time angle values ​​of each joint, including the current angles of major motion joints such as the knee, hip, and ankle, are extracted from the hyperkinetic posture assessment results. Simultaneously, the rate of change of each joint angle is extracted, calculated by the angle difference at consecutive time points. This data is then categorized by joint name to form a posture assessment dataset containing joint identifiers, real-time angle values, and angle change rates, with each data item accompanied by a collection timestamp.

[0089] Step S1422: Extract the safety angle range and angle change rate threshold of the corresponding joint from the hypermotion posture safety rules, and generate a rule comparison parameter set.

[0090] For each joint in the attitude assessment dataset, the corresponding safety parameters are retrieved from the hypermotion attitude safety rules, including the upper limit of the safety angle, the lower limit of the safety angle, and the maximum allowable rate of angle change. These parameters are then categorized and organized by joint to generate a rule comparison parameter set. Each entry in the parameter set includes a joint identifier, a safety angle range, a rate threshold, and out-of-range judgment conditions.

[0091] Step S1423: Compare the real-time angle values ​​of each joint in the attitude evaluation dataset with the safe angle range in the rule comparison parameter set one by one, and calculate the angle deviation value. The angle deviation value is calculated as follows: real-time angle value - upper limit of safe angle (if real-time angle value > upper limit of safe angle) or angle deviation value = lower limit of safe angle - real-time angle value (if real-time angle value < lower limit of safe angle).

[0092] For each joint, compare its real-time angle value with the safe angle range. If the real-time angle value is greater than the upper limit of the safe angle, calculate the angle deviation as the real-time angle value minus the upper limit of the safe angle; if the real-time angle value is less than the lower limit of the safe angle, calculate the angle deviation as the lower limit of the safe angle minus the real-time angle value; if the real-time angle value is within the safe range, the angle deviation value is zero. Record the angle deviation value and deviation direction (positive exceeding the upper limit, negative exceeding the lower limit) for each joint.

[0093] Step S1424: Compare the rate of angle change of each joint in the posture evaluation dataset with the threshold of the rate of angle change in the rule comparison parameter set to determine whether there is a rate exceeding the limit.

[0094] The rate of change of angle of each joint is compared with the corresponding rate threshold. If the rate of change of angle is greater than the rate threshold, it is determined that the rate is out of limit, and the magnitude and duration of the out-of-limit are recorded. If it is not exceeded, the rate is determined to be normal.

[0095] Step S1425: Based on the combined angle deviation value and rate exceeding the limit, evaluate the out-of-range level of each joint. The out-of-range level is divided into slight out-of-range, moderate out-of-range, and severe out-of-range.

[0096] The grading criteria are set based on the magnitude of the angle deviation value, and a comprehensive evaluation is conducted in conjunction with the rate exceeding the limit. If the angle deviation value is small and there is no rate exceeding the limit, it is rated as a slight exceeding the limit; if the angle deviation value is moderate or there is a slight rate exceeding the limit, it is rated as a moderate exceeding the limit; if the angle deviation value is large or there is a significant rate exceeding the limit, it is rated as a severe exceeding the limit.

[0097] Step S1426: For joints that exceed the range of moderate or severe limits, determine the joint locations and directions to be adjusted based on the adjustment guidelines in the rules for safe movement postures and in conjunction with joint motion correlation analysis.

[0098] For joints assessed as moderately or severely exceeding limits, obtain corresponding adjustment guidelines from the rules for safe movement postures, and clarify the adjustment goals (such as restoring the angle to a safe range). Through joint motion correlation analysis, determine the motion influence relationship between this joint and other joints to avoid adjusting a single joint causing new postural imbalances. Based on the direction of angle deviation and the principles of joint kinetics, determine the specific adjustment direction, such as reducing the bending angle when the knee joint angle exceeds the upper limit.

[0099] Step S1427: Based on the adjustment direction and angle deviation values, calculate the suggested adjustment range, and combine the joint coordination relationship in the hypermotion feature analysis results to generate a posture adjustment suggestion that includes the adjustment location, adjustment direction, and suggested range.

[0100] Based on the magnitude of the angular deviation, calculate the recommended adjustment range proportionally to ensure that the adjusted angle falls within a safe range. Referring to the joint coordination data in the hyper-motion feature analysis results, the adjustment range must be adapted to the motion states of other related joints to avoid creating new incoordination. Organize the information such as the adjustment location, direction, recommended range, and timing into a structured posture adjustment recommendation.

[0101] Step S143: Compare and analyze the predicted trajectory of hypermotion in the hypermotion monitoring and analysis results with the hypermotion trajectory safety rules to determine whether the subsequent center of gravity of the hypermotion object will exceed the safe area. If it will exceed the safe area, determine the area to be avoided and the direction of trajectory adjustment to obtain trajectory adjustment suggestions.

[0102] Extract the future center of gravity position sequence from the hyper-motion trajectory prediction results and perform a spatiotemporal comparison with the safe zone boundary in the hyper-motion trajectory safety rules. For the center of gravity position at each prediction time, determine whether it is within the safe zone; if it exceeds the safe zone, mark it as a risk point. Analyze the distribution of risk points to determine the area to be avoided (e.g., the left side of a specific road segment). Based on the safe zone boundary and the current trajectory trend, calculate the trajectory adjustment direction vector (e.g., rightward shift, deceleration and turning), and generate trajectory adjustment suggestions based on terrain features, including the adjustment start position, adjustment direction, and adjustment magnitude reference.

[0103] Step S144: Based on the attitude adjustment suggestions and trajectory adjustment suggestions, generate hypermotion adjustment suggestions, which include specific joint adjustment parameters and trajectory adjustment paths.

[0104] The posture adjustment suggestions, including the adjustment areas, directions, and magnitudes, are converted into specific joint adjustment parameters, such as the specific angle to be reduced in knee flexion and the adjustment time window. The avoidance areas and adjustment directions in the trajectory adjustment suggestions are converted into trajectory adjustment paths, including the coordinates of key points, path turning angles, and suggested speeds for each segment. The posture and trajectory adjustment suggestions are then validated for synergy to ensure no conflicts arise during implementation. These are then integrated into a complete hyper-kinematic adjustment suggestion, accompanied by an explanation of the adjustment implementation priorities and expected effects.

[0105] Step S145: Based on the risk levels of the hypermotion posture assessment results and hypermotion trajectory prediction results in the hypermotion monitoring and analysis results, and combined with the risk warning rules in the basic rule set, generate a hypermotion safety warning. The hypermotion safety warning includes a risk level description and risk avoidance measures.

[0106] The out-of-range levels of each joint are extracted from the hyper-motion posture assessment results, and the risk levels of risk points are extracted from the hyper-motion trajectory prediction results. A comprehensive risk level is calculated according to the risk assessment rules in the basic rule set. Based on the comprehensive risk level and risk warning rules, a risk level description is generated, such as "moderate posture risk accompanied by trajectory deviation risk." According to the risk type (posture risk, trajectory risk), corresponding risk avoidance measures are matched from the rules, such as "immediately adjust the knee joint angle to a safe range" and "turn in advance to avoid the left-side risk area," forming a hyper-motion safety warning that includes the predicted time of risk occurrence, the scope of risk impact, and specific avoidance steps.

[0107] Step S146: Integrate hyper-motion adjustment suggestions and hyper-motion safety tips to generate a hyper-motion monitoring and management plan.

[0108] Hyper-exercise adjustment suggestions and safety alerts are sorted chronologically and by risk level, prioritizing high-risk alerts and urgent adjustment suggestions. Implementation time windows, conditions, and effectiveness verification standards are added, clearly defining the implementing entity for each suggestion and alert (e.g., athlete self-adjustment, coach remote guidance). A consistency check is performed on the plan content to ensure no contradictions between adjustment suggestions and safety alerts. The final hyper-exercise monitoring and management plan includes adjustment suggestions, safety alerts, implementation instructions, and verification standards. The plan is stored in Extensible Markup Language (XML) format for easy subsequent terminal adaptation.

[0109] Step S150: Input the super motion monitoring management scheme into the super motion monitoring terminal adaptation module, perform matching processing between the scheme format and the terminal interface, obtain the terminal-adapted super motion monitoring management scheme, and send it to the target super motion monitoring terminal.

[0110] After receiving the hyper-motion monitoring management scheme, the hyper-motion monitoring terminal adaptation module parses the scheme's structure and content to obtain the interface parameter information of the target hyper-motion monitoring terminal. Based on the interface's data format requirements, transmission protocol, and rate limits, it performs format conversion, protocol encapsulation, and rate adaptation processing on the scheme. After processing, it generates a terminal-adapted hyper-motion monitoring management scheme and sends it to the target terminal via the communication interface, simultaneously recording the transmission status and timestamp to ensure accurate delivery of the scheme.

[0111] Step S151: Obtain the interface parameter information of the target hypermotion monitoring terminal. The interface parameter information includes the data transmission protocol, data format requirements, and data receiving rate requirements.

[0112] The device files of the target hypersonic monitoring terminal are queried through the terminal management system to extract interface parameter information. Data transmission protocols include transport layer protocols and application layer protocols, such as Transmission Control Protocol (TCP) and Message Queuing Telemetry Transmission Protocol (MQTP). Data format requirements include data encoding methods (such as Extensible Markup Language (EXPLAIN) and JavaScript object notation), field definitions, and data type specifications. Data reception rate requirements include parameters such as maximum transmission rate and data packet size limits. These parameters are organized into a structured set of interface parameters for subsequent solution adaptation processing.

[0113] Step S152: Input the hyper-motion monitoring management scheme into the hyper-motion monitoring terminal adaptation module. The hyper-motion monitoring terminal adaptation module parses the structure of the hyper-motion monitoring management scheme and extracts the data streams corresponding to the hyper-motion adjustment suggestions and hyper-motion safety prompts in the hyper-motion monitoring management scheme.

[0114] After the hyper-motion monitoring and management solution is input into the adaptation module, the module starts the solution parser, which parses the structure according to the solution's format specifications and identifies the content blocks corresponding to hyper-motion adjustment suggestions and hyper-motion safety prompts. The data stream of each block is extracted, separating text description data, parameter data, and instruction data. The extracted data stream undergoes integrity verification to ensure no data loss or corruption; if any data is missing, a retransmission mechanism is triggered to obtain the complete data.

[0115] Step S1521: The super motion monitoring terminal adaptation module calls the scheme parsing engine and loads the parsing rules corresponding to the super motion monitoring management scheme format. The parsing rules include scheme structure definition, field identification rules and data type mapping relationship.

[0116] After the adaptation module starts, it calls the corresponding solution parsing engine based on the format type of the hyper-motion monitoring management solution (such as Extensible Markup Language). The parsing engine loads predefined parsing rules, where the solution structure definition clearly defines the hierarchical structure of the solution (such as the nesting relationship between root nodes and child nodes); the field identification rules specify the naming conventions for fields in adjustment suggestions and safety prompts; and the data type mapping relationship defines the correspondence between various types of data (text, numeric, boolean values) in the solution and the data types of the terminal interface.

[0117] Step S1522: Based on the parsing rules, traverse the content of the hyper-motion monitoring and management scheme, identify the hyper-motion adjustment suggestion block and the hyper-motion safety prompt block in the scheme, and record the start and end positions of each block.

[0118] The parsing engine traverses the scheme content according to the structure definition in the parsing rules, identifying the block tags corresponding to hyper-motion adjustment suggestions and hyper-motion safety prompts through field identification rules. For each block, it records its start and end byte positions in the scheme, and extracts the block's metadata (such as block type and generation time). If an incomplete or incorrectly formatted block tag is found, the parsing engine will attempt to repair it based on the context. If the repair fails, it will be marked as an abnormal block and the error information will be recorded.

[0119] Step S1523: Based on the start and end positions of the blocks, extract the data content within the hypermotion adjustment suggestion block and the data content within the hypermotion safety prompt block to form a structured adjustment suggestion data stream and a safety prompt data stream.

[0120] Based on the recorded start and end positions, the complete content of the hyper-motion adjustment suggestion block and hyper-motion safety warning block is extracted from the hyper-motion monitoring and management scheme. For the adjustment suggestion block, specific data such as joint adjustment parameters (e.g., knee joint angle adjustment range, hip joint force correction value) and trajectory adjustment paths (e.g., steering node coordinate sequence, slope adaptation trajectory correction coefficient) are extracted and organized into a structured adjustment suggestion data stream according to parameter type and path node order. Each data item is labeled with the corresponding movement part identifier and adjustment priority. For the safety warning block, risk level descriptions (e.g., muscle fatigue risk level, joint load risk level) and risk avoidance measures (e.g., deceleration interval suggestions, action correction points) are extracted and formed into a safety warning data stream according to risk type and measure execution order. Each warning message is associated with the corresponding risk assessment basis and triggering conditions. During the extraction process, the unstructured text descriptions are semantically converted, transforming the natural language adjustment requirements and warning content into machine-recognizable parameter formats and instruction codes to ensure the uniformity of the data stream format. If duplicate data or conflicting information exists within a block, deduplication is performed based on the record with the newest timestamp, and the original location of the conflicting data is marked for subsequent tracing.

[0121] Step S1524: Based on the data format requirements in the interface parameter information of the target hyper-motion monitoring terminal, perform field mapping processing on the structured adjustment suggestion data stream and the safety prompt data stream respectively, and convert the field names and data types in the data stream into a format compatible with the terminal interface.

[0122] Obtain the data format specifications defined in the target hyper-motion monitoring terminal interface parameter information, including field naming rules, data type restrictions (such as integer, floating-point, and string), and field length constraints. For the adjustment suggestion data stream, map "Joint Adjustment Parameters" to the "MotionAdjustParam" field in the terminal interface, and map "Trajectory Adjustment Path" to the "TrajectoryModifyPath" field. Convert data types according to terminal requirements, such as retaining a specified number of decimal places for floating-point values ​​of angle adjustment ranges. For the safety prompt data stream, map "Risk Level Description" to the "RiskLevelDesc" field, and "Risk Avoidance Measures" to the "RiskAvoidanceMeasure" field, ensuring that the length of string-type prompt content does not exceed the maximum number of bytes limited by the terminal interface. During the field mapping process, establish a mapping relationship lookup table to record the correspondence between the original fields and the terminal interface fields. If special data with no corresponding field in the terminal interface is encountered, create an extended field and indicate the data meaning in the field description. At the same time, add the extended field information to the metadata area of ​​the data stream. After mapping is completed, verify whether the data type and length of each field meet the terminal requirements. Truncate or convert the format of any non-compliant fields until all fields meet the terminal interface format specifications.

[0123] Step S1525: Merge the format-converted adjustment suggestion data stream and security prompt data stream, arrange the fields according to the data packet structure required by the terminal interface, and generate the initial data stream.

[0124] Following the data packet structure template defined by the target hyper-motion monitoring terminal interface, the order of the adjustment suggestion data stream and the safety prompt data stream within the data packet is determined. Typically, key adjustment parameter fields are placed at the beginning of the data packet, followed by the safety prompt fields. The two data streams, after format conversion, are merged sequentially. During the merging process, duplicate metadata fields are removed, and a complete metadata record (including data generation time, data source identifier, and data version number) is retained. The merged data stream undergoes field order verification to ensure consistency with the field sequence required by the terminal interface. If there is a discrepancy, the field positions are rearranged according to the template. After generating the initial data stream, the total length of the data stream is calculated to ensure it does not exceed the maximum data packet length supported by the terminal interface. If it does, it is marked as a data stream requiring fragmentation, and the fragmentation start position and the number of fragments are recorded.

[0125] Step S153: Based on the data transmission protocol in the interface parameter information of the target hyper-motion monitoring terminal, perform protocol encapsulation processing on the format-converted data stream to generate data packets that conform to the transmission protocol.

[0126] The protocol encapsulation module is invoked to load the data transmission protocol specifications supported by the target hyper-motion monitoring terminal, including the protocol header format, data verification method, and command identifier definition. A protocol header is added before the initial data stream, containing information such as the protocol version number, data packet type identifier (adjustment suggestion packet or security prompt packet), data length field, and terminal device address. The data length field is filled with the actual number of bytes in the initial data stream. A checksum is calculated on the encapsulated data packet using a protocol-specified verification algorithm (such as CRC checksum) on the data packet content (including the protocol header and initial data stream), and the resulting checksum is added to the checksum field at the end of the data packet. If the transmission protocol requires data encryption, the initial data stream portion is encrypted using the terminal's preset encryption key. The encrypted data replaces the original initial data stream position, and an encryption identifier is added to the protocol header. After encapsulation, the overall structure of the data packet is checked to ensure it conforms to the protocol specifications, with a focus on verifying the integrity of the header fields, the validity of the checksum, and the correctness of the encryption identifier, ensuring that the generated data packet can be correctly recognized and parsed by the terminal.

[0127] Step S154: Test whether the transmission rate of the data packet meets the data reception rate requirements in the interface parameter information of the target hyper-motion monitoring terminal. If not, adjust the fragment size of the data packet until the transmission rate meets the requirements, and obtain the data packet corresponding to the hyper-motion monitoring management scheme after terminal adaptation.

[0128] Establish a test connection with the target hyper-motion monitoring terminal, send encapsulated data packets in a simulated transmission environment, and monitor the data packet transmission rate (number of bytes successfully transmitted per unit time) in real time. Compare the actual transmission rate with the data reception rate requirement in the terminal interface parameter information. If the actual rate is greater than or equal to the reception rate requirement, the transmission rate is considered to meet the requirements and no adjustment is needed. If the actual rate is lower than the reception rate requirement, initiate a fragmentation adjustment mechanism. Calculate a reasonable fragment size based on the rate difference, typically splitting the original data packet proportionally into multiple smaller data packets to reduce the transmission pressure on individual data packets. When adjusting the fragment size, ensure that the length of each fragment data packet does not exceed the maximum fragment length specified by the transmission protocol, and that the number of fragments does not exceed the maximum number of concurrent fragments supported by the terminal. Re-encapsulate the data stream after adjusting the fragment size, generating multiple fragment data packets, and test the transmission rate again until the transmission rate of all fragment data packets meets the terminal's reception requirements. Record the final fragment size parameters, fragment quantity, and transmission rate compliance value, and add this information to the additional information field of the data packet.

[0129] Step S155: Send the data packet corresponding to the terminal-adapted hyper-motion monitoring management scheme to the target hyper-motion monitoring terminal.

[0130] A formal connection is established with the target hyper-motion monitoring terminal through the data transmission interface. During the connection process, the terminal device identity is verified (by comparing the device's unique identifier with a preset whitelist) to ensure that data is sent to the correct terminal device. Data packets adapted to the terminal are sent sequentially according to the fragment order. The transmission status is monitored in real time during transmission, including the success flag, retransmission count, and transmission latency for each data packet. If a fragment data packet fails to be sent (without acknowledgment after exceeding the maximum retransmission count), the fragment data packet is retransmitted, and the reason for the failure is recorded (e.g., network fluctuations, temporary terminal offline). After all fragment data packets have been sent, the system waits for the terminal's reception confirmation information. The confirmation information should include the reception status of all fragments and the data integrity verification result. If the terminal reports complete data reception, the transmission is considered successful, and the transmission completion time and terminal confirmation information are recorded. If the terminal reports missing data or verification failure, the corresponding data packet is retransmitted according to the missing fragment number until the terminal confirms complete data reception. After transmission is complete, the data transmission connection is closed, transmission resources are released, and a detailed log of this transmission (including transmission time, number of data packets, transmission rate, and error handling records) is stored in the local log system.

[0131] Figure 2 The illustration shows exemplary hardware and software components of a machine learning-based hyper-motion monitoring and management system 100 that can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used in the machine learning-based hyper-motion monitoring and management system 100 and to perform the functions in this application.

[0132] The machine learning-based hyper-motion monitoring and management system 100 can be a general-purpose server or a special-purpose server, both of which can be used to implement the machine learning-based hyper-motion monitoring and management method of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0133] For example, the machine learning-based hyper-motion monitoring and management system 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the machine learning-based hyper-motion monitoring and management system 100 may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The machine learning-based hyper-motion monitoring and management system 100 also includes an I / O interface 150 between the computer and other input / output devices.

[0134] For ease of explanation, only one processor is described in the machine learning-based hyper-motion monitoring and management system 100. However, it should be noted that the machine learning-based hyper-motion monitoring and management system 100 of this application may also include multiple processors, and therefore the steps executed by one processor described in this application may also be executed jointly by multiple processors or individually. For example, if the processor of the machine learning-based hyper-motion monitoring and management system 100 executes steps A and B, it should be understood that steps A and B may also be executed jointly by two different processors or individually by one processor. For example, the first processor executes step A, the second processor executes step B, or the first processor and the second processor jointly execute steps A and B.

[0135] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned machine learning-based hyper-motion monitoring and management method is implemented.

[0136] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A hyper-motion monitoring and management method based on machine learning, characterized in that, The method includes: Based on the application requirements of hyper-motion monitoring scenarios, an association mapping relationship between hyper-motion monitoring requirements and machine learning model modules is established, and the association mapping results between hyper-motion monitoring requirements and machine learning model modules are obtained. The hyper-motion monitoring requirements include hyper-motion type monitoring requirements and hyper-motion state evaluation requirements, and the machine learning model modules include hyper-motion feature analysis modules and hyper-motion state prediction modules. A hypermotion monitoring dataset is obtained, and a hierarchical feature parsing process is performed on the hypermotion monitoring dataset. Combined with the correlation mapping results, a hypermotion hierarchical feature set is obtained. The hypermotion monitoring dataset contains motion posture data and motion trajectory data during the hypermotion process, and the hypermotion hierarchical feature set contains posture hierarchical features and trajectory hierarchical features. The hypermotion hierarchical feature set is input into the machine learning model module corresponding to the association mapping result, and the interaction processing between features and model is performed to generate hypermotion monitoring and analysis results. The hypermotion monitoring and analysis results include hypermotion posture evaluation results and hypermotion trajectory prediction results. Based on the results of the hypermotion monitoring analysis and combined with the basic rule set for hypermotion monitoring management, a hypermotion monitoring management plan is generated. The hypermotion monitoring management plan includes hypermotion adjustment suggestions and hypermotion safety tips. The aforementioned hyper-motion monitoring management scheme is input into the hyper-motion monitoring terminal adaptation module, and the scheme format is matched with the terminal interface to obtain the terminal-adapted hyper-motion monitoring management scheme, which is then sent to the target hyper-motion monitoring terminal.

2. The hyper-motion monitoring and management method based on machine learning according to claim 1, characterized in that, Based on the application requirements of the hyper-motion monitoring scenario, an association mapping relationship is established between hyper-motion monitoring requirements and machine learning model modules, and the association mapping results between hyper-motion monitoring requirements and machine learning model modules are obtained, including: Key requirement elements are extracted from the application requirements of hyper-motion monitoring scenarios. Based on these key requirement elements, hyper-motion monitoring requirements are classified to obtain hyper-motion monitoring requirement types. The key requirement elements include the real-time requirements and accuracy requirements of hyper-motion monitoring. The hyper-motion monitoring requirement types correspond to hyper-motion type monitoring requirements and hyper-motion state assessment requirements. Obtain a preset set of machine learning model module function descriptions. The set of machine learning model module function descriptions includes the function descriptions of the hyper-motion feature analysis module and the hyper-motion state prediction module. The function description of the hyper-motion feature analysis module includes the function of feature extraction of hyper-motion posture data and the function of feature extraction of hyper-motion trajectory data. The function description of the hyper-motion state prediction module includes the function of predicting the subsequent hyper-motion state based on hyper-motion features. The hyper-motion monitoring requirement type is subjected to feature encoding processing to generate a hyper-motion monitoring requirement feature vector; Perform feature encoding on the functional description of each machine learning model module in the set of machine learning model module functional descriptions to generate a machine learning model module functional feature vector. Calculate the similarity between the feature vector of hyper-motion monitoring demand and the feature vector of each machine learning model module, generate a similarity matrix, and determine the matching relationship between hyper-motion monitoring demand type and machine learning model module based on the similarity matrix; Based on the matching relationship, an association mapping table is constructed between hyper-motion monitoring requirements and machine learning model modules. The association mapping table records at least one machine learning model module corresponding to each hyper-motion monitoring requirement type. The association mapping table is used as the association mapping result between hyper-motion monitoring requirements and machine learning model modules.

3. The hyper-motion monitoring and management method based on machine learning according to claim 2, characterized in that, The step of determining the matching relationship between hyper-motion monitoring demand types and machine learning model modules based on the similarity matrix includes: Set a similarity threshold, and mark the hyper-motion monitoring demand type corresponding to the similarity value in the similarity matrix that is greater than or equal to the similarity threshold as a preliminary matching pair with the machine learning model module; The preliminary matching pairs are verified to check whether the functions of the machine learning model modules in the preliminary matching pairs fully cover the requirements of the corresponding hyper-motion monitoring requirement type. If they are fully covered, the preliminary matching pairs are determined to be the final matching relationships. If the coverage is not complete, the similarity threshold is readjusted or the functional description of the machine learning model module is supplemented until a matching pair that fully covers the functional requirements is found, and this is determined as the final matching relationship.

4. The hyper-motion monitoring and management method based on machine learning according to claim 1, characterized in that, The process involves acquiring a hyper-motion monitoring data set, performing hierarchical feature parsing on the hyper-motion monitoring data set, and combining the association mapping results to obtain a hyper-motion hierarchical feature set, including: Raw data during the hypermotion process is collected by a hypermotion data acquisition device. The raw data is then filtered and processed to remove invalid data, resulting in a hypermotion monitoring data set. The raw data includes the position and angle data of the hypermotion object at different times. The hypermotion monitoring data set includes the motion posture data and motion trajectory data during the hypermotion process. The motion posture data includes the angle change data of each joint of the hypermotion object, and the motion trajectory data includes the position change data of the center of gravity of the hypermotion object. Based on the feature requirements of the machine learning model modules associated with the hyper-motion type monitoring needs in the association mapping results, the hierarchical dimensions of the hyper-motion monitoring dataset are determined, including the temporal hierarchical dimension and the spatial hierarchical dimension. Based on the time-layer dimension, the hypermotion monitoring dataset is divided into multiple time-layer data units according to time intervals. Each time-layer data unit contains motion posture data and motion trajectory data within the corresponding time period. Based on the spatial layer dimension, the motion posture data in each time layer data unit is divided into multiple posture sub-layer data according to the joint location, and the motion trajectory data is divided into multiple trajectory sub-layer data according to the spatial region. Feature extraction processing is performed on each attitude sub-layer data to obtain attitude layer features, which include joint angle change amplitude features and joint angle change frequency features. Feature extraction processing is performed on each trajectory sub-layer data to obtain trajectory layer features, which include the distance feature of the change in center of gravity position and the direction feature of the change in center of gravity position. All pose layer features and trajectory layer features are summarized to obtain a hypermotion layer feature set.

5. The hyper-motion monitoring and management method based on machine learning according to claim 4, characterized in that, The process of extracting features from each pose sub-layer data to obtain pose layer features includes: The joint angle change data in each attitude sub-layer is preprocessed, and the joint angle value at each time step is extracted from the preprocessed joint angle change data. The difference between the joint angle values ​​at adjacent time steps is calculated to obtain the joint angle change sequence. In the sequence of joint angle changes, find the maximum and minimum values, calculate the difference between the maximum and minimum values, and use this difference as the feature of the joint angle change amplitude. The number of consecutive data segments with the same direction of change in the joint angle change sequence is counted to determine the number of cycles of joint angle change. Based on the time length and number of cycles corresponding to the attitude sublayer data, calculate the average time length of each cycle, and use the reciprocal of the average time length as the joint angle change frequency feature. The joint angle change amplitude feature and the joint angle change frequency feature are associated and stored as the attitude layering feature corresponding to the attitude sub-layer data.

6. The hyper-motion monitoring and management method based on machine learning according to claim 1, characterized in that, The step of inputting the hyper-motion hierarchical feature set into the machine learning model module corresponding to the association mapping result, performing feature-model interaction processing, and generating hyper-motion monitoring and analysis results includes: Extract the machine learning model module corresponding to the current hyper-motion monitoring needs from the association mapping results, and determine the input interface parameter requirements of the machine learning model module. The input interface parameter requirements include the dimension requirements of the feature data and the format requirements of the feature data. The hypermotion hierarchical feature set is formatted to meet the input interface parameter requirements of the machine learning model module, resulting in the adjusted hypermotion hierarchical feature set. The adjusted hyper-motion layered feature set is input into the hyper-motion feature analysis module. The hyper-motion feature analysis module performs correlation analysis on the posture layered features in the adjusted hyper-motion layered feature set to determine the correlation of angle changes between different joint parts, performs continuity analysis on the trajectory layered features to determine the continuous trend of the center of gravity position change, and performs standardization processing on the feature values ​​during the analysis process to ensure scale consistency, thereby obtaining the hyper-motion feature analysis results. The results of the hyper-motion feature analysis are input into the hyper-motion state prediction module. The hyper-motion state prediction module calls the preset prediction algorithm and predicts the subsequent posture and trajectory changes of the hyper-motion object based on the standardized angle change correlation and the continuous trend of center of gravity position change in the hyper-motion feature analysis results, thus obtaining the hyper-motion state prediction results. The results of hypermotion feature analysis and hypermotion state prediction are integrated to generate hypermotion monitoring and analysis results. The hypermotion monitoring and analysis results include hypermotion attitude assessment results and hypermotion trajectory prediction results. The hypermotion attitude assessment results are determined based on the correlation of angle changes, and the hypermotion trajectory prediction results are determined based on the continuous trend of center of gravity position changes and the predicted trajectory changes.

7. The hyper-motion monitoring and management method based on machine learning according to claim 6, characterized in that, The hyper-motion feature analysis results are input into the hyper-motion state prediction module. The hyper-motion state prediction module calls a preset prediction algorithm and, based on the standardized angle change correlation and center of gravity position change continuous trend in the hyper-motion feature analysis results, predicts the subsequent posture and trajectory changes of the hyper-motion object, obtaining the hyper-motion state prediction result, including: The super-motion state prediction module loads a preset prediction algorithm, which is a prediction algorithm based on temporal features. The angular change correlation in the hyper-motion feature analysis results is transformed into a correlation matrix, which records the correlation coefficients of angular changes at different joint locations; The continuous trend of the change in the center of gravity position in the hyper-motion feature analysis results is transformed into a trend vector, which records the rate of change of the center of gravity position in different directions. The correlation matrix and trend vector are standardized by eigenvalue standardization to obtain standardized correlation matrix and standardized trend vector. Then, the standardized correlation matrix and standardized trend vector are input into the feature processing layer of the prediction algorithm for fusion processing to generate fused feature vector. The fused feature vector is input into the prediction layer of the prediction algorithm. The prediction layer processes the fused feature vector based on the model parameters trained on historical hypermotion state data, and predicts the joint angle and center of gravity position of the hypermotion object at multiple subsequent moments. The predicted joint angles at multiple subsequent moments are arranged in chronological order to obtain the subsequent posture changes of the hyper-motion object. Arrange the predicted center of gravity positions at multiple subsequent moments in chronological order to obtain the subsequent trajectory changes of the hyper-moving object; By integrating the subsequent attitude and trajectory changes of the hyper-motion object, the hyper-motion state prediction result is obtained.

8. The hyper-motion monitoring and management method based on machine learning according to claim 1, characterized in that, Based on the results of the hyper-motion monitoring analysis and combined with the set of basic rules for hyper-motion monitoring management, a hyper-motion monitoring management scheme is generated, including: Obtain a set of basic rules for hyper-motion monitoring and management. The set of basic rules includes hyper-motion posture safety rules and hyper-motion trajectory safety rules. The hyper-motion posture safety rules record the safe range of the angles of each joint of the hyper-motion object. The hyper-motion trajectory safety rules record the safe area of ​​the center of gravity of the hyper-motion object. The hypermotion posture assessment results in the hypermotion monitoring and analysis results are compared and analyzed with the hypermotion posture safety rules to determine whether the current joint angle of the hypermotion object exceeds the safety range. If it exceeds the safety range, the joint parts that need to be adjusted and the adjustment direction are determined, and posture adjustment suggestions are obtained. The prediction results of hypermotion trajectory in the hypermotion monitoring and analysis results are compared and analyzed with the hypermotion trajectory safety rules to determine whether the subsequent center of gravity position of the hypermotion object will exceed the safe area. If it will exceed the safe area, the area to be avoided and the direction of trajectory adjustment are determined, and trajectory adjustment suggestions are obtained. Based on the posture adjustment suggestions and trajectory adjustment suggestions, a hypermotion adjustment suggestion is generated, which includes specific joint adjustment parameters and trajectory adjustment paths. Based on the risk levels of the hypermotion posture assessment results and hypermotion trajectory prediction results in the hypermotion monitoring and analysis results, and combined with the risk warning rules in the basic rule set, a hypermotion safety warning is generated. The hypermotion safety warning includes a risk level description and risk avoidance measures. By integrating hyper-motor adjustment recommendations and hyper-motor safety tips, a hyper-motor monitoring and management plan is generated.

9. The hyper-motion monitoring and management method based on machine learning according to claim 1, characterized in that, The process of inputting the hyper-motion monitoring management scheme into the hyper-motion monitoring terminal adaptation module, performing scheme format matching with the terminal interface to obtain the terminal-adapted hyper-motion monitoring management scheme, and sending it to the target hyper-motion monitoring terminal includes: Obtain the interface parameter information of the target hyper-motion monitoring terminal, wherein the interface parameter information includes data transmission protocol, data format requirements, and data receiving rate requirements; The hyper-motion monitoring management solution is input into the hyper-motion monitoring terminal adaptation module. The hyper-motion monitoring terminal adaptation module parses the structure of the hyper-motion monitoring management solution and extracts the data streams corresponding to the hyper-motion adjustment suggestions and hyper-motion safety prompts in the hyper-motion monitoring management solution. Based on the data format requirements in the interface parameter information of the target hyper-motion monitoring terminal, the extracted data stream is format converted to make the data stream format consistent with the format required by the terminal interface, thus obtaining the format-converted data stream; Based on the data transmission protocol in the interface parameter information of the target hyper-motion monitoring terminal, the data stream after format conversion is encapsulated to generate data packets that conform to the transmission protocol. Test whether the transmission rate of the data packet meets the data reception rate requirements in the interface parameter information of the target hyper-motion monitoring terminal. If not, adjust the fragment size of the data packet until the transmission rate meets the requirements, and obtain the data packet corresponding to the hyper-motion monitoring management scheme after terminal adaptation. Send the data packet corresponding to the terminal-adapted hyper-motion monitoring management solution to the target hyper-motion monitoring terminal.

10. A hyper-motion monitoring and management system based on machine learning, characterized in that, The machine learning-based hyper-motion monitoring and management system includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the machine learning-based hyper-motion monitoring and management method according to any one of claims 1-9.

Images