An edge computing device for periodic bilateral athletic movement effect assessment

Abstract

This invention provides an edge computing device for evaluating the effectiveness of periodic bilateral athletic movements, belonging to the fields of edge computing and sports technology. The device includes a main control MCU, an onboard IMU, and distributed IMU sensors. The main control MCU constructs an intelligent predictive evaluation model based on data collected by the IMUs. By solving a system of multiple regression equations in the conduction chain, it calculates the contribution weights of key positions to the center of mass movement and the force transmission efficiency index. Combining comprehensive force characteristics and unilateral and bilateral technical stability feature vectors, it constructs a bilateral overall motion coordination index. This invention utilizes edge computing for real-time processing, accurately quantifying force transmission efficiency and bilateral coordination, and comprehensively evaluating the overall effect of periodic athletic movements.

Description

Technical Field

[0001] This invention belongs to the field of edge computing and sports technology, and specifically refers to an edge computing device for evaluating the effects of periodic bilateral sports movements. Background Technology

[0002] In cyclical bilateral sports such as swimming, rowing, cross-country skiing, running, and cycling, athletes' performance highly depends on the coordination of limb exertion and the effectiveness of force transmission. Existing motion analysis technologies, which employ high-speed cameras and motion capture equipment, are greatly limited by venue and technical conditions and cannot provide real-time feedback; or they only focus on the kinematic trajectory of a single limb, lacking a quantitative assessment of bilateral synergy and the efficiency of force transmission from limbs to the trunk. Furthermore, traditional solutions using IMU data for analysis and evaluation rely on cloud processing, which suffers from latency in transmitting large amounts of raw data, making it difficult to meet the real-time guidance needs of high-frequency movements in complex sports application scenarios (underwater, outdoor environments, large spaces, and long durations). Therefore, there is an urgent need for a portable, wearable edge computing device that can perform real-time calculations and comprehensively evaluate bilateral motion coordination and force transmission effectiveness. Summary of the Invention

[0003] To address the technical problems existing in the prior art, the present invention provides an edge computing device for evaluating the effects of periodic bilateral physical movements, the edge computing device comprising: The main control MCU and onboard IMU sensor are fixed to the athlete's lumbosacral region, as well as distributed IMU sensors fixed to key positions of the relevant limbs for force exertion; The athlete's body is simplified as a body mass point with its center of mass in the lumbosacral region. The onboard IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data to obtain the velocity vector and acceleration vector of the body's center of mass at each moment. The velocity vector and acceleration vector of the body's center of mass are used to form the target output vector matrix. The accelerometer, gyroscope and magnetometer of the distributed IMU sensor collect time series data to obtain the velocity vector and acceleration vector at key positions at each moment, and then concatenate them into a global regression input vector; The main control MCU constructs an intelligent prediction and evaluation model, and based on the residuals and weights obtained from the intelligent prediction and evaluation model, as well as a pre-built edge computing algorithm, performs edge computing on the measured data of the athlete to be evaluated, and performs intelligent evaluation, including: Based on the global regression input vector, multiple strength feature values ​​are defined and calculated, and aggregated into a comprehensive strength feature vector for each key position. Multiple unilateral technical feature values ​​are defined and calculated, and aggregated into a unilateral technical stability feature vector for each key position. Multiple bilateral technical feature values ​​are defined and calculated, and aggregated into bilateral technical stability feature vectors for multiple key position pairs. The overall feature vector is composed of the comprehensive strength feature vector, the unilateral technical stability feature vector, and the bilateral technical stability feature vector. Based on the global regression input vector, construct and solve an intelligent prediction and evaluation model based on the multivariate regression equation system of the transmission chain; Using the weights of the solved multiple regression equations of the conduction chain, the contribution weights of the parameters at each key position to the velocity and acceleration of the body's center of mass are obtained. Using the residuals of the solved multiple regression equations of the conduction chain and the target output vector matrix, the force transmission efficiency index is defined and calculated. Based on the contribution weights, force transmission efficiency index, and overall characteristic vector of each key position, the impact of sports movements at each key position on the forward velocity and acceleration of the body's center of mass is comprehensively evaluated. By utilizing the contribution weights of each key position, the comprehensive force feature vector of each key position, the technical stability feature vector of one side of the body at each key position, and the technical stability feature vectors of both sides of the body for multiple key position pairs, a bilateral overall motion coordination index is constructed to comprehensively evaluate the overall effect of all key position movements.

[0004] The beneficial effects of the technical solution provided by this invention include at least the following: This invention deploys intelligent prediction and evaluation models and edge computing algorithms directly at the edge (main control MCU), completing the entire process from data acquisition and feature extraction to model solving without relying on cloud servers. This edge computing architecture significantly reduces data transmission latency, enabling millisecond-level real-time feedback for evaluating athletes' periodic sports movements. It effectively solves the problems of traditional cloud processing solutions, such as high latency and inability to meet the real-time guidance needs of high-frequency movements. More specifically: 1. This invention defines and calculates the following indices: explosive power index reflecting changes in motion vector; average impulse rate reflecting the effective force range; force decay coefficient reflecting the peak of the motion vector that maintains the force level; power output index and rotational power output index reflecting the instantaneous power accumulation of a specific limb within the assessment period; force smoothness index reflecting the smoothness of force application; and contribution ratio index reflecting the change in kinetic energy at key positions to the change in kinetic energy at the center of mass. These indices are then aggregated into a comprehensive force characteristic vector for key limb positions. This vector can be used to identify and compare patterns of force characteristics at different key limb positions of athletes, to identify and compare the force characteristics of the same key limb position in different athletes, and for comparing, classifying, and clustering the force characteristics of the same limb position in a group of athletes.

[0005] 2. This invention defines and calculates a continuity index that reflects the phase fluctuation of unilateral limb position movements, a phase smoothness index that reflects the instantaneous frequency variation of movements, a consistency index that reflects the phase consistency of cyclical movements, and a phase-amplitude coupling index that reflects the degree of coordination between the timing of force application and the maximum force burst. These defined and calculated indices are aggregated into a technical stability feature vector for key limb positions. This vector can be used to identify and compare patterns of technical stability characteristics of different unilateral key limb positions in athletes, to identify and compare the technical stability characteristics of the same key limb position in different athletes, and to compare, classify, and cluster the technical stability characteristics of the same limb position in a group of athletes.

[0006] 3. This invention defines and calculates the phase lock value, which reflects the stability of the phase difference between bilateral limb positions and movements; the phase synchronization accuracy, which quantifies the phase difference between the actual bilateral movements and the ideal standard movements; the phase difference fluctuation index, which reflects bilateral coordination; the synchronization duration ratio index, which reflects the duration of bilateral coordination within a specific time period; and the cross synchronization value, which quantifies the strength of bilateral synchronization and the time difference at optimal synchronization. The defined and calculated data are aggregated into a bilateral technical stability feature vector. This vector can be used to identify and compare the technical stability feature patterns of athletes' bilaterally symmetrical key limb positions, to identify and compare the technical stability features of the same symmetrical bilateral limbs of different athletes, and to compare, classify, and cluster the technical stability features of the same symmetrical bilateral limb positions among a group of athletes.

[0007] 4. The device of this invention can collect a large number of sample data of athlete motion vectors, which can be used to train an athlete strength and technical characteristic assessment and prediction model based on the conduction chain multiple regression equation system.

[0008] 5. This invention utilizes the weights of the solved multivariate regression equations to obtain the contribution weights of the motion vectors of different key limb positions of the athlete to the forward velocity and acceleration of the body's center of mass. Using these weights, the athlete's strength and technical characteristics can be evaluated by combining the comprehensive strength feature vector, the unilateral technical stability vector, the bilateral technical stability vector, and the various indices in the vectors.

[0009] 6. This invention utilizes the residuals of the solved multiple regression equations to define and calculate the force transmission efficiency index, which can be used to evaluate the predictive performance of the multiple linear regression model.

[0010] 7. This invention utilizes the weights of the solved multiple regression equations, as well as the aggregated comprehensive force feature vector, unilateral technical stability vector, and bilateral technical stability vector, to further define and calculate the force gradient index for evaluating the efficiency of force transmission in the joint chain, the phase coordination index for evaluating the coordination of phase motion between multiple degrees of freedom on one side of the body, and the bilateral precession index for examining whether the phase difference field has rotation. The defined and calculated indices are then aggregated into a bilateral overall motion coordination index. This index can be used to evaluate the overall force transmission, unilateral coordination, and bilateral synchronicity of all key positions of an athlete. The value of this index can be used to retrospectively analyze the force and technical characteristics of a specific key position.

[0011] 8. This invention can be used to evaluate the effect of a single training session for the same athlete, and to compare and evaluate historical training effects. It can also utilize built-in group sports data and artificial intelligence models to perform comparative, hierarchical classification, cluster analysis, and artificial intelligence (AI) evaluation of groups of athletes.

[0012] 9. The main control MCU of the device of this invention is responsible for edge computing and intelligent evaluation, the edge AI module is responsible for AI-related calculations (similar to the function of a GPU), and the Zigbee network layer module is responsible for the synchronization control of data collected by multiple IMUs and reporting the collected data to the MCU. The algorithms, programs and frameworks of edge computing and intelligent evaluation are put into the MCU, and the real-time preprocessing and calculation of the collected data are completed locally. This avoids the high latency and high bandwidth dependency mode of transmitting a large amount of raw data to the cloud (external) for calculation, and transforms it into local real-time closed-loop processing. This effectively avoids the risk of congestion of wireless transmission links, significantly reduces the data transmission latency and power consumption from the end to the cloud (external), and improves the real-time response capability of the system.

[0013] 10. The device of this invention can still independently complete core intelligent sensing and decision analysis tasks even when the wireless network signal is poor or disconnected, without relying on external computing power, thus enhancing the system's independent working ability and environmental robustness in complex sports application scenarios.

[0014] 11. The device of this invention avoids the risk of uploading sensitive raw data to the cloud. All data is anonymized on the device or directly converted into evaluation descriptions and decision instructions. It can also combine built-in physical and physiological data such as athlete gender, age, height, weight, and heart rate to conduct comprehensive artificial intelligence evaluation, thus eliminating the risk of leakage of user privacy and confidential data in the transmission link or cloud storage from the source. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a structural diagram of an edge device provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the locations of each IMU provided in an embodiment of the present invention; Figure 3 This is an axial schematic diagram of the onboard IMU and the geodetic coordinate system at the body's center of mass provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of an IMU fixed to the wrist according to an embodiment of the present invention; Figure 5 This is an axial schematic diagram of an IMU sensor fixed to the back of the palm and its local coordinate system provided in an embodiment of the present invention. Detailed Implementation

[0017] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0018] like Figure 1 As shown, this embodiment of the invention provides an edge computing device for evaluating the effects of periodic bilateral physical movements: The edge computing device includes: a main control MCU and onboard IMU sensors fixed to the athlete's lumbosacral region, and distributed IMU sensors fixed to key limb force application points (e.g., wrist, knee, ankle, etc.), (and other components such as an edge AI acceleration module and a Zigbee coordinator). Figure 2 As shown; The athlete's body is simplified as a body mass point with its center of mass in the lumbosacral region. The onboard IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data to obtain the velocity vector and acceleration vector of the body's center of mass at each moment. The velocity vector and acceleration vector of the body's center of mass are used to form the target output vector matrix. The accelerometer, gyroscope and magnetometer of the distributed IMU sensor collect time series data to obtain the velocity vector and acceleration vector at key positions at each moment, and then concatenate them into a global regression input vector; The main control MCU constructs an intelligent prediction and evaluation model, and based on the residuals and weights obtained from the intelligent prediction and evaluation model, as well as a pre-built edge computing algorithm, performs edge computing on the measured data of the athlete to be evaluated, and performs intelligent evaluation, including: Based on the global regression input vector, multiple strength feature values ​​are defined and calculated, and aggregated into a comprehensive strength feature vector for each key position. Multiple unilateral technical feature values ​​are defined and calculated, and aggregated into a unilateral technical stability feature vector for each key position. Multiple bilateral technical feature values ​​are defined and calculated, and aggregated into bilateral technical stability feature vectors for multiple key position pairs. The overall feature vector is composed of the comprehensive strength feature vector, the unilateral technical stability feature vector, and the bilateral technical stability feature vector. Based on the global regression input vector, construct and solve an intelligent prediction and evaluation model based on the multivariate regression equation system of the transmission chain; Using the weights of the solved multiple regression equations of the conduction chain, the contribution weights of the parameters at each key position to the velocity and acceleration of the body's center of mass are obtained. Using the residuals of the solved multiple regression equations of the conduction chain and the target output vector matrix, the force transmission efficiency index is defined and calculated. Based on the contribution weights, force transmission efficiency index, and overall characteristic vector of each key position, the impact of sports movements at each key position on the forward velocity and acceleration of the body's center of mass is comprehensively evaluated. By utilizing the contribution weights of each key position, the comprehensive force feature vector of each key position, the technical stability feature vector of one side of the body at each key position, and the technical stability feature vectors of both sides of the body for multiple key position pairs, a bilateral overall motion coordination index is constructed to comprehensively evaluate the overall effect of all key position movements.

[0019] Optionally, such as Figure 3 As shown, the onboard IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data, which is obtained in the geodetic coordinate system X, Y, Z, for each moment. The velocity vector of the body's center of mass forward and acceleration vector The target output vector matrix is ​​composed of the velocity and acceleration vectors of the body's center of mass moving forward. ; The distributed IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data for each moment in the geodetic coordinate system, or other coordinate systems defined as needed. Key positions for limb exertion (like Figure 4 The velocity vector at the wrist (as shown) and acceleration vector , Key positions for limb exertion of and acceleration vector The motion vector that constitutes this key position ,Will Concatenate them into the global regression input vector: .

[0020] The main control MCU defines and calculates multiple force feature values ​​based on the global regression input vector, and aggregates them into a comprehensive force feature vector for each key location, specifically including: S1-1, Using the global regression input vector In ,in, It is a critical position number. These are the directional component numbers. Define and calculate the following eigenvalues: (1) Extreme value characteristics: ; ; .

[0021] in, This is the start time of the period that needs to be evaluated. It is the end time; (2) Mean and root mean square characteristics: ; .

[0022] in, The total number of sampling points, i.e. arrive The number of data points within a time period; For discrete sampling time points, The sampling sequence number; (3) Time integral characteristics: ; .

[0023] S1-2. Using the above characteristic values, define and calculate the following force characteristic values: (1) The explosive force index describing the change of motion vector per unit time:

[0024] in, For time intervals; For motion vectors such as velocity and acceleration, the significance of defining this exponent lies in its ability to quantify the time cost for these motion vectors to reach their maximum value. (2) Average impulse rate within the effective force range:

[0025] in, The Hilbert envelope represents the set of effective force application points. It is the contour of the signal extracted by low-pass filtering and amplitude detection, which can more stably reflect the continuous changing trend of force. The absolute value sign emphasizes that the filtering criteria are based on the magnitude of the amplitude. Indicates the entire evaluation period [ , Within, envelope signal The maximum value that can be achieved This is the threshold scaling factor. This indicates the total duration of the effective power range; The execution of sports movements includes preparatory movements, buffer movements, and transitional movements. Simply averaging the impulse over the entire time span dilutes the impulse of the effective force exertion. The significance of defining this index lies in its ability to adaptively extract the total duration of the "effective force exertion zone" from the collected raw signal, and also in its ability to utilize... The threshold for the "effective force application range" is preset; then the absolute value of the motion vector is integrated within this range and then averaged to eliminate the dilution effect of preparatory actions, buffer actions and transitional actions. (3) Force attenuation coefficient after peak value:

[0026] in, From peak Descending to Time, It is a natural constant; In sports, relying solely on the peak value of the motion vector and the explosive power index of the previously defined motion vector change is insufficient to evaluate the level of strength maintenance; the significance of defining this coefficient lies in: quantitatively describing the level of decline in peak value. (4) Power output index of mechanical power output:

[0027] Existing specialized equipment and methods can measure the power and energy consumption of the entire human body during sports activities; however, there are currently no dedicated equipment and methods for measuring the power accumulation and energy consumption of specific limbs, requiring wearable devices and specific algorithms for measurement and calculation. The significance of defining this index is that it calculates the accumulation of instantaneous power of a specific limb within the evaluation period, which is physically close to energy consumption and can be used to reflect the energy conversion effect of limb movements. (5) Rotational power output index of rotational power output:

[0028] in, Angular velocity of rotation Both rotational angular acceleration and rotational angular acceleration are obtained from the IMU. For limb movements involving rotation (such as wrist rotation, trunk and limb twisting), the significance of defining this index is that it calculates the cumulative instantaneous rotational power of a specific limb within the evaluation time period, which can be used to reflect the rotational energy conversion effect of limb movements. (6) Force smoothness index for smooth force application:

[0029] The fluidity of athletic movements is influenced by a combination of factors, including technical skill, coordination of the musculoskeletal system, and strength. Currently, there is no scientific method to quantify this. The significance of defining this index lies in using the reciprocal of the variance of the differential motion signal to quantify the smoothness of force transmission. The value indicates that there is a pause or jerking during force application, and that there is energy leakage.

[0030] (7) Key positions The contribution ratio of the change in kinetic energy to the change in the kinetic energy of the center of mass is:

[0031] The significance of defining this index is to compare the kinetic energy of limb movement with the kinetic energy of the center of mass moving forward, in order to observe the contribution rate of this part of the limb movement to the forward movement of the center of mass. S1-3. Aggregate the calculated force characteristic values ​​into key locations. Comprehensive strength feature vector: .

[0032] The significance of defining this aggregation vector is that it forms a set of vectors that can be used to quantitatively assess and identify key limb locations. The strength feature vector is similar to a "strength feature fingerprint"; it can be used for pattern recognition and comparison of the strength features of different limbs of athletes, as well as for the classification and clustering analysis of the strength features of the same limbs of group athletes, and for artificial intelligence-assisted training.

[0033] Optionally, the main control MCU defines and calculates multiple unilateral technical feature values ​​based on the global regression input vector, and aggregates them into a unilateral technical stability feature vector for each key location, specifically including: S2-1, For each key location scalar components Using Hilbert transform to obtain Construct analytical signals , Extracting the instantaneous phase:

[0034] Extracting instantaneous frequency:

[0035] S2-2. Define and calculate the following unilateral technical characteristic values ​​for the body: (1) Continuity index:

[0036] in, This represents the phase difference between adjacent moments. This indicates mapping the phase difference to... , This indicates that for all points in time... Define and calculate variance: This is used to measure the dispersion of phase difference fluctuations. Used to normalize variance to Within the range; A larger value indicates a more continuous phase change and stable operation; conversely, a smaller value indicates a more obvious discontinuity. Even if the macroscopic trajectory of the movement is continuous, if the phase changes drastically (such as wrist shaking during swimming), it can still cause problems. Low values ​​may also appear; the significance of defining this index is to detect microscopic lag. (2) Phase smoothness index:

[0037] in, It is the instantaneous frequency of the action extracted through Hilbert transform. for The mean, for standard deviation The coefficient of variation; The larger the value, the more stable the instantaneous frequency and the smoother the rhythm of the movement; the smaller the value, the greater the frequency fluctuation. Average speed and instantaneous frequency cannot reflect subtle fluctuations in movement frequency. The significance of defining this index lies in detecting changes in rhythm through the coefficient of variation of instantaneous frequency; for example, the minute fluctuations in stride frequency during running and the frequency of limb strokes during swimming. Capable of precise capture; (3) Phase coherence index:

[0038] The larger the value, the better the phase consistency. The significance of defining this index is to use vector averaging to quantify whether each action falls on the same phase point for cyclical periodic actions. (4) Phase-amplitude coupling index of mutual information between force amplitude and phase: ; This indicates the fineness of neuromuscular control, among which For mutual information; If the amplitude of limb force occurs at an angle unfavorable to optimal force exertion, it will affect the forward movement of the center of gravity; the significance of defining this index is to detect the degree of coordination between the timing of force exertion (phase) and the maximum force burst (amplitude), and to describe the precision control performance of neuromuscular control. S2-3. Establish the above indices as key positions. One-sided technical stability feature vector: .

[0039] The significance of defining this aggregation vector is that it forms a technical feature vector that can be used to quantitatively evaluate and identify the position of a unilateral limb, similar to a "technical feature fingerprint"; it can be used for pattern recognition and comparison of different limb technical features of athletes, as well as for classification and cluster analysis of technical features of the same limb position in a group of athletes, and for artificial intelligence-assisted training.

[0040] Optionally, the main control MCU defines and calculates multiple technical feature values ​​for both sides of the body based on the global regression input vector, and aggregates them into a technical stability feature vector for both sides of the body at each key location, specifically including: S3-1, For each key location scalar components Using Hilbert transform to obtain Construct analytical signals , Extracting the instantaneous phase: ; For the p-th pair of key positions that are symmetrical on the left and right sides and Calculate the instantaneous phase difference between the two sides. ; For example, using the X-axis acceleration of the right wrist Calculate its phase Then analyze and calculate the left wrist. Obtain the phase difference During a complete paddling cycle, Monotonically increasing from 0 to .

[0041] S3-2. Define and calculate the following technical characteristic values ​​for both sides of the body: (1) Phase lock value:

[0042] If the average time difference between the movements of both limbs is small, but if there are fluctuations in the phase difference... Low values ​​may occur. The significance of defining this index is to quantify the stability of the phase difference between the two sides throughout the entire action cycle. (2) Phase synchronization accuracy:

[0043] in, The ideal phase difference under different motion modes; The "gold standard" of phase difference (e.g., 180-degree phase reversal of the freestyle stroke) can be set according to actual training requirements, i.e., the ideal phase difference. The significance of defining this index lies in quantifying the deviation between actual movements and standards, which can be used to precisely adjust training requirements and goals; (3) Phase difference fluctuation index:

[0044] The significance of defining this index lies in calculating the dispersion of the phase difference; large fluctuations indicate poor coordination between the left and right sides. (4) Synchronization duration ratio index:

[0045] in, The allowable deviation threshold; For longer-duration sports activities (e.g., running, swimming), in addition to observing and evaluating synchronicity within instantaneous or short time intervals, it is also necessary to evaluate the persistence of bilateral coordination synchronicity within intervals. The significance of defining this index is that it allows setting an allowable deviation threshold. Then, the time percentage of the ideal synchronous phase difference within the window that is less than the threshold is counted. (5) Cross-correlation synchronization: ; .

[0046] Among them, time lag The normalized cross-correlation function is the time interval during which one signal is shifted forward or backward relative to another signal. , This indicates that two signals have a given lag. The covariance trend under; Used to eliminate the influence of signal amplitude; operators Quantization indicates the degree of synchronization, with values ​​ranging from... Between these values, a value close to 1 indicates strong synchronicity of bilateral movements, a value close to -1 indicates anti-synchronicity, and a value of 0 indicates no correlation. Operators in To extract The time difference at which optimal synchronization is achieved; The significance of defining the above indices lies in specifically quantifying the degree of synchronization and calculating the time difference when optimal coordination is achieved; S3-3. Form the technical stability feature vector on both sides of the p-th key position by the above calculated exponents: .

[0047] The significance of defining this vector lies in constructing a high-dimensional "two-sided technical stability evaluation map." This vector allows for the rapid diagnosis of the type of two-sided asymmetry, such as: high... ,Low This indicates a steady rhythm, but the angle is slightly off; low. ,high The angle is correct, but the rhythm is slightly off; high. ,big It indicates a strong correlation, but with a fixed delay; it can be used for pattern recognition and comparison of technical features of athletes' bilateral limbs, as well as classification and cluster analysis of technical features of the same bilateral limbs in a group of athletes, and artificial intelligence-assisted training.

[0048] Optionally, the main control MCU constructs and solves an intelligent prediction and evaluation model based on the multivariate regression equation system of the transmission chain according to the global regression input vector, specifically including: The intelligent prediction and evaluation model based on the multiple regression equation system of the transmission chain is constructed as follows: ; .

[0049] in, and The regression coefficient matrix and As a bias term, the constraint equations conform to physical reality (if there is no action at all critical locations, the input...). When zero, the inherent base velocity and acceleration of the body's center of mass; when solving using the least squares method, this can be achieved by extending the design matrix and combining it with the regression coefficient matrix. and (Solve together) This is the upper limit of the optimal time lag (to avoid exceeding the reasonable lag time during actual calculations (e.g., the action cycle)). )), and The residual, in a physical sense, represents the difference between the observed value and the linear regression prediction. Its sources include key locations not included in the model and changes in the centroid dynamics caused by external forces (e.g., changes in the resistance of the fluid (air or water) around the limb during movement, sensor noise, and offset). During the model training phase, the sample data obtained from the test is used first. , , Find the optimal regression coefficient matrix and and bias terms and The prediction equations are obtained as follows: ; .

[0050] During the model testing / application phase, new and further tested results will be obtained. The data is input into the prediction equation system to obtain the predicted value, and then the residual is calculated using the difference between the predicted value and the actual measured value. ; .

[0051] Thus, the regression coefficient matrix and Bias terms and residual and This ultimately forms a complete intelligent prediction and evaluation model based on the multiple regression equations of the transmission chain: ; .

[0052] Optionally, the main control MCU uses the weights of the solved multiple regression equations of the conduction chain to obtain the contribution weights of the parameters at each key position to the velocity and acceleration of the body's center of mass. Using the residuals of the solved multiple regression equations of the conduction chain and the target output vector matrix, it defines and calculates the force transmission efficiency index, specifically including: By using the weights of the solved multiple regression equations of the transmission chain, the key positions can be obtained. The weights of the parameters contributing to the velocity of the body's center of mass:

[0053] and key positions The weights of the parameters contributing to the acceleration of the body's center of mass:

[0054] If residual and The smaller the value, the closer the model is to reality. To represent the center-of-mass motion that cannot be explained in a multiple regression model, use The ratio of the two represents the total intensity of the forward movement of the body's center of mass. The smaller the value, the smaller the proportion of the model that is not interpretable; Define and calculate the force transmission efficiency index: .

[0055] Optionally, the main control MCU comprehensively evaluates the impact of sports movements at each key position on the forward velocity and acceleration of the body's center of mass based on the contribution weight, force transmission efficiency index, and overall feature vector at each key position, specifically including: The closer to 1, the better the prediction of the multiple linear regression model, and the higher the explanatory power of each limb movement on the center of mass movement. if A lower value (e.g., 0.3) indicates poor model prediction. Specific reasons could be as follows: (1) The model is missing key force points: Data on important key force points (such as core muscle groups and joints not covered by sensors) were not included in the model; (2) Incoordination of technical movements / compensation: The direction of force exerted by each part is chaotic and cancels each other out, or there are a lot of redundant movements that are not related to the target movement, resulting in low efficiency of force transmission; (3) There are external disturbances (such as air, water flow, etc.): These external forces act directly on the center of mass, but the data is not included in the model input; Within a specific time period, or a sliding time window, the contribution weight of each key position i is used, combined with the eigenvalues ​​of the overall feature vector and each feature component, to comprehensively evaluate and judge the impact of this position on forward velocity and acceleration. If regression analysis reveals the contribution weight of a certain location (such as the right wrist) It is very large, and further examination is needed on its corresponding... and ,if Power output index and force smoothness index If they are all very high, it indicates that the power output from this position is strong and the movement is smooth; if Phase-amplitude coupling index in A higher level indicates precise timing of force application and a high level of neuromuscular control. If the contribution weights of both positions are low, further examination is needed. Phase difference fluctuation index If it is low, it indicates that the coordination and synchronicity of the movements in this symmetrical position are poor.

[0056] Optionally, the main control MCU utilizes the contribution weights of each key position, the comprehensive force feature vector of each key position, the technical stability feature vector of one side of the body at each key position, and the technical stability feature vectors of both sides of the body for multiple key position pairs to construct a bilateral overall motion coordination index, comprehensively evaluating the overall effect of motion at all key positions, specifically including: (1) Define and calculate the force potential function:

[0057] in, It is a location The weight of the contribution to the velocity of the center of mass at time t. It is the comprehensive force feature vector at position i. The L2 norm; Define and compute the force gradient vector:

[0058] in, Indicates from position To adjacent positions The direction of the link, It is a location and location spatial coordinates, It refers to the physical distance between locations; Define and calculate the force gradient exponent:

[0059] The significance of defining this index lies in assessing the efficiency of force transmission along the joint chain; High values ​​indicate force along the joint chain. Positive and efficient transmission, This indicates that the force is blocked or leaked in the direction of transmission; (2) Define the phase vector field:

[0060] in, , , These represent the instantaneous phases of position i at degrees of freedom 1, 2, and 3 (generally 3 degrees of freedom, such as flexion / extension, adduction / abduction, and rotation); Calculate phase divergence:

[0061] in, It is the generalized coordinate of the d-th degree of freedom; Define the phase coordination index:

[0062] The significance of defining this index lies in assessing the coordination of phase motion among multiple degrees of freedom on one side of the body, and quantifying the degree of "convergence" or "divergence" of phase motion.

[0063] Define the two-sided phase difference vector field:

[0064] in, , , These are the left and right phase differences of the p-th joint in the 1st, 2nd, and 3rd degrees of freedom, respectively; Calculate the phase difference curl:

[0065] Define the two-sided precession index:

[0066] The significance of defining this index lies in observing whether the phase difference field exists; This indicates no rotation, and stable left-right coordination; This indicates that there is a rotational pattern in the left-right coordination, and it is necessary to calculate and determine the specific rotational pattern and corresponding motion characteristics or items based on IMU data; Each of the above positions of , , Defined as a tensor:

[0067] Define the bilateral overall motor coordination index as follows: For all locations The tensor is averaged and normalized to obtain ,get Location of for:

[0068] For example: four positions of for: ; .

[0069] in, It is the identity matrix under ideal and coordinated conditions. Represents the square root of the sum of the squares of all elements in the matrix; Define this index The significance lies in evaluating the overall power transmission, unilateral coordination, and bilateral synchronicity across all key positions; A score of 1 indicates good power transfer, unilateral coordination, and bilateral synchronization, resulting in a high degree of overall coordination. A 1 indicates a coordination defect, which can be addressed through analysis. The components, and tracing back to the aforementioned key locations. The problem is located by analyzing the comprehensive strength characteristics and unilateral technical characteristics of the combined force and the bilateral technical stability characteristics of the paired p.

[0070] Optionally, the main control MCU also calculates and evaluates, for swimming, the impact of hand rotation (roll, yaw, pitch) on swimming speed for different strokes. like Figure 5 As shown, an IMU is fixed on the palm (or inside the palm). In a local coordinate system with the IMU's center of mass as the origin, the angular velocity vectors of roll, yaw, and pitch around the x, y, and z axes of this local coordinate system are obtained. ) and angular acceleration vector ; For a single hand position i, the motion vector is obtained as follows: The motion vectors of the left and right palms are concatenated to obtain the global regression input vector. ; Still using time-series data from the accelerometer, gyroscope, and magnetometer of the IMU in the lumbosacral region, each time step is obtained in the X, Y, Z geodetic coordinate system. The velocity vector of the body's center of mass forward and acceleration vector ,by Output a vector matrix for the target; Define and calculate the overall feature vector consisting of the comprehensive strength feature vector of the palm, the technical stability feature vector of one side of the body, and the technical stability feature vector of both sides of the body (the specific method is as described above, and will not be repeated here). Define and calculate the force gradient index, phase coordination index, and bilateral rotational precession index of the palm, and define and calculate the overall motion coordination index of both palms (the specific method is as described above, and will not be repeated here). An intelligent prediction and evaluation model is established to assess the effect of the axial rotation of the palm around the local coordinate system on the forward velocity and acceleration of the body's center of mass, and the evaluation is carried out (the specific method is as described above, and will not be repeated here).

[0071] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An edge computing device for evaluating the effects of periodic bilateral physical movements, characterized in that, The edge computing device includes: a main control MCU and an onboard IMU sensor fixed to the athlete's lumbosacral region, and distributed IMU sensors fixed to key positions of the relevant limbs for force exertion; The athlete's body is simplified as a body mass point with its center of mass in the lumbosacral region. The onboard IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data to obtain the velocity vector and acceleration vector of the body's center of mass at each moment. The velocity vector and acceleration vector of the body's center of mass are used to form the target output vector matrix. The accelerometer, gyroscope and magnetometer of the distributed IMU sensor collect time series data to obtain the velocity vector and acceleration vector at key positions at each moment, and then concatenate them into a global regression input vector; The main control MCU constructs an intelligent prediction and evaluation model, and based on the residuals and weights obtained from the intelligent prediction and evaluation model, as well as a pre-built edge computing algorithm, performs edge computing on the measured data of the athlete to be evaluated, and performs intelligent evaluation, including: Based on the global regression input vector, multiple strength feature values ​​are defined and calculated, and aggregated into a comprehensive strength feature vector for each key position. Multiple unilateral technical feature values ​​are defined and calculated, and aggregated into a unilateral technical stability feature vector for each key position. Multiple bilateral technical feature values ​​are defined and calculated, and aggregated into bilateral technical stability feature vectors for multiple key position pairs. The overall feature vector is composed of the comprehensive strength feature vector, the unilateral technical stability feature vector, and the bilateral technical stability feature vector. Based on the global regression input vector, construct and solve an intelligent prediction and evaluation model based on the multivariate regression equation system of the transmission chain; Using the weights of the solved multiple regression equations of the conduction chain, the contribution weights of the parameters at each key position to the velocity and acceleration of the body's center of mass are obtained. Using the residuals of the solved multiple regression equations of the conduction chain and the target output vector matrix, the force transmission efficiency index is defined and calculated. Based on the contribution weights, force transmission efficiency index, and overall characteristic vector of each key position, the impact of sports movements at each key position on the forward velocity and acceleration of the body's center of mass is comprehensively evaluated. By utilizing the contribution weights of each key position, the comprehensive force feature vector of each key position, the technical stability feature vector of one side of the body at each key position, and the technical stability feature vectors of both sides of the body for multiple key position pairs, a bilateral overall motion coordination index is constructed to comprehensively evaluate the overall effect of all key position movements.

2. The edge computing device according to claim 1, characterized in that, The onboard IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data, which is then used to obtain the data for each moment in the X, Y, Z geodetic coordinate system. The velocity vector of the body's center of mass forward and acceleration vector The target output vector matrix is ​​composed of the velocity and acceleration vectors of the body's center of mass moving forward. ; The distributed IMU sensor's accelerometer, gyroscope, and magnetometer collect time-series data for each moment in the geodetic coordinate system, or other coordinate systems defined as needed. Key positions for limb exertion velocity vector and acceleration vector , Key positions for limb exertion of and acceleration vector The motion vector that constitutes this key position ,Will Concatenate them into the global regression input vector: .

3. The edge computing device according to claim 2, characterized in that, The main control MCU defines and calculates multiple force feature values ​​based on the global regression input vector, and aggregates them into a comprehensive force feature vector for each key location, specifically including: S1-1, Using the global regression input vector In ,in, It is a critical position number. These are the directional component numbers. Define and calculate the following eigenvalues: (1) Extreme value characteristics: ； ； ； in, This is the start time of the period that needs to be evaluated. It is the end time; (2) Mean and root mean square characteristics: ； ； in, The total number of sampling points, i.e. arrive The number of data points within a time period; For discrete sampling time points, The sampling sequence number; (3) Time integral characteristics: ； ； S1-2. Using the above characteristic values, define and calculate the following force characteristic values: (1) The explosive force index describing the change of motion vector per unit time: ； in, For time intervals; (2) Average impulse rate within the effective force range: ； in, The Hilbert envelope represents the set of effective force application points. It is the contour of the signal extracted by low-pass filtering and amplitude detection, which can more stably reflect the continuous changing trend of force. The absolute value sign emphasizes that the filtering criteria are based on the magnitude of the amplitude. Indicates the entire evaluation period [ , Within, envelope signal The maximum value that can be achieved This is the threshold scaling factor. This indicates the total duration of the effective power range; (3) Force attenuation coefficient after peak value: ； in, From peak Descending to Time, It is a natural constant; (4) Power output index of mechanical power output: ； (5) Rotational power output index of rotational power output: ； in, Angular velocity of rotation Both rotational angular acceleration and rotational angular acceleration are obtained from the IMU. (6) Force smoothness index for smooth force application: ； (7) Key positions The contribution ratio of the change in kinetic energy to the change in the kinetic energy of the center of mass is: ； S1-3. Aggregate the force characteristic values ​​defined and calculated above into key locations. Comprehensive strength feature vector: 。 4. The edge computing device according to claim 2, characterized in that, The main control MCU defines and calculates multiple unilateral technical feature values ​​based on the global regression input vector, and aggregates them into a technical stability feature vector for each key location on one side of the body, specifically including: S2-1, For each key location scalar components Using Hilbert transform to obtain Construct analytical signals , Extracting the instantaneous phase: ； Extracting instantaneous frequency: ； S2-2. Define and calculate the following unilateral technical characteristic values ​​for the body: (1) Continuity index: ； in, This represents the phase difference between adjacent moments. This indicates mapping the phase difference to... , This indicates that for all points in time... Define and calculate variance: This is used to measure the dispersion of phase difference fluctuations. Used to normalize variance to Within the range; A larger value indicates a more continuous phase change and stable operation; conversely, a smaller value indicates a more obvious discontinuity. (2) Phase smoothness index: ； in, It is the instantaneous frequency of the action extracted through Hilbert transform. for The mean, for standard deviation The coefficient of variation; The larger the value, the more stable the instantaneous frequency and the smoother the rhythm of the movement; the smaller the value, the greater the frequency fluctuation. (3) Phase coherence index: ； A larger value indicates better phase consistency; (4) Phase-amplitude coupling index of mutual information between force amplitude and phase: ； This indicates the fineness of neuromuscular control, among which For mutual information; S2-3. Establish the above indices as key positions. One-sided technical stability feature vector: 。 5. The edge computing device according to claim 2, characterized in that, The main control MCU defines and calculates multiple technical feature values ​​for both sides of the body based on the global regression input vector, and aggregates them into technical stability feature vectors for both sides of the body at each key location, specifically including: S3-1, For each key location scalar components Using Hilbert transform to obtain Construct analytical signals , Extracting the instantaneous phase: ； For the p-th pair of key positions that are symmetrical on the left and right sides and Calculate the instantaneous phase difference between the two sides. ; S3-2. Define and calculate the following technical characteristic values ​​for both sides of the body: (1) Phase lock value: ； (2) Phase synchronization accuracy: ； in, The ideal phase difference under different motion modes; (3) Phase difference fluctuation index: ； (4) Synchronization duration ratio index: ； in, The allowable deviation threshold; (5) Cross-correlation synchronization: ； ； Among them, time lag The normalized cross-correlation function is the time interval during which one signal is shifted forward or backward relative to another signal. , This indicates that two signals have a given lag. The covariance trend under; Used to eliminate the influence of signal amplitude; Operators Quantization indicates the degree of synchronization, with values ​​ranging from... Between these values, a value close to 1 indicates strong synchronicity of bilateral movements, a value close to -1 indicates anti-synchronicity, and a value of 0 indicates no correlation. Operators in To extract The time difference at which optimal synchronization is achieved; S3-3. The indices defined and calculated above are used to form the technical stability feature vector on both sides of the p-th key position: 。 6. The edge computing device according to claim 2, characterized in that, The main control MCU constructs and solves an intelligent prediction and evaluation model based on the global regression input vector and the multivariate regression equation system of the transmission chain, specifically including: The intelligent prediction and evaluation model based on the multiple regression equation system of the transmission chain is constructed as follows: ； ； in, and The regression coefficient matrix and As bias terms, the constraint equations conform to physical reality. It is the upper limit of the set optimal time lag. and The residual, in a physical sense, represents the difference between the observed value and the linear regression prediction. Its sources include key locations not included in the model and changes in the centroid dynamics caused by external forces. During the model training phase, the sample data obtained from the test is used first. , , Find the optimal regression coefficient matrix and and bias terms and The prediction equations are obtained as follows: ； ； During the model testing / application phase, new and further tested results will be obtained. The data is input into the prediction equation system to obtain the predicted value, and then the residual is calculated using the difference between the predicted value and the actual measured value. ； ； Thus, the regression coefficient matrix and Bias terms and residual and This ultimately forms a complete intelligent prediction and evaluation model based on the multiple regression equations of the transmission chain: ； 。 7. The edge computing device according to claim 6, characterized in that, The main control MCU, using the weights of the solved multiple regression equations of the conduction chain, obtains the contribution weights of the parameters at each key position to the velocity and acceleration of the body's center of mass. Using the residuals of the solved multiple regression equations of the conduction chain and the target output vector matrix, it defines and calculates the force transmission efficiency index, specifically including: By using the weights of the solved multiple regression equations of the transmission chain, the key positions can be obtained. The weights of the parameters contributing to the velocity of the body's center of mass: ； and key positions The weights of the parameters contributing to the acceleration of the body's center of mass: ； If residual and The smaller the value, the closer the model is to reality. To represent the center-of-mass motion that cannot be explained in a multiple regression model, use The ratio of the two represents the total intensity of the forward movement of the body's center of mass. The smaller the value, the smaller the proportion of the model that is not interpretable; Define and calculate the force transmission efficiency index: 。 8. The edge computing device according to claim 7, characterized in that, The main control MCU comprehensively evaluates the impact of sports movements at each key position on the forward velocity and acceleration of the body's center of mass, based on the contribution weight, force transmission efficiency index, and overall feature vector at each key position. Specifically, this includes: The closer to 1, the better the prediction of the multiple linear regression model, and the higher the explanatory power of each limb movement on the center of mass movement. if A lower value indicates poor model prediction. Let's analyze the specific reasons: (1) The model is missing key points: data on important key points of force application are not included in the model; (2) Incoordination of technical movements / compensation: The direction of force exerted by each part is chaotic and cancels each other out, or there are a lot of redundant movements that are not related to the target movement, resulting in low efficiency of force transmission; (3) External disturbances exist: These external forces act directly on the centroid, but the data is not included in the model input; Within a specific time period, or a sliding time window, the contribution weight of each key position i is used, combined with the eigenvalues ​​of the overall feature vector and each feature component, to comprehensively evaluate and judge the impact of this position on forward velocity and acceleration. If regression analysis reveals the contribution weight of a certain position It is very large, and further examination is needed on its corresponding... and ,if Power output index and force smoothness index If they are all very high, it indicates that the power output from this position is strong and the movement is smooth; if Phase-amplitude coupling index in A higher level indicates precise timing of force application and a high level of neuromuscular control. If the contribution weights of both positions are low, further examination is needed. Phase difference fluctuation index If it is low, it indicates that the coordination and synchronicity of the movements in this symmetrical position are poor.

9. The edge computing device according to claim 8, characterized in that, The main control MCU utilizes the contribution weights of each key position, the comprehensive force feature vector of each key position, the technical stability feature vector of one side of the body at each key position, and the technical stability feature vectors of both sides of the body for multiple key position pairs to construct a bilateral overall motion coordination index, comprehensively evaluating the overall effect of motion at all key positions, specifically including: (1) Define and calculate the force potential function: ； in, It is a location The weight of the contribution to the velocity of the center of mass at time t. It is the comprehensive force feature vector at position i. The L2 norm; Define and compute the force gradient vector: ； in, Indicates from position To adjacent positions The direction of the link, It is a location and location spatial coordinates, It refers to the physical distance between locations; Define and calculate the force gradient exponent: ； (2) Define and calculate the phase vector field: ； in, , , These represent the instantaneous phases of position i at degrees of freedom 1, 2, and 3, respectively. Define and calculate phase divergence: ； in, It is the generalized coordinate of the d-th degree of freedom; Define and calculate the phase coordination index: ； Define and compute the two-sided phase difference vector field: ； in, , , These are the left and right phase differences of the p-th joint in the 1st, 2nd, and 3rd degrees of freedom, respectively; Define and calculate the phase difference curl: ； Define and calculate the two-sided precession index: ； Each of the above positions of , , Define and compute it as a tensor: ； Define and calculate the bilateral overall motion coordination index. as follows: For all locations The tensor is averaged and normalized to obtain Location of for: ； ； in, It is the identity matrix under ideal and coordinated conditions. Represents the square root of the sum of squares of all elements in the matrix; based on the bilateral overall motion coordination index. Evaluate the overall force transmission, unilateral coordination, and bilateral synchronicity across all key positions: A score of 1 indicates good power transfer, unilateral coordination, and bilateral synchronization, resulting in a high degree of overall coordination. The number 1 indicates a coordination defect, which leads to further analysis. Each component, and tracing key locations. The problem is located by analyzing the comprehensive strength characteristics and unilateral technical characteristics of the combined force and the bilateral technical stability characteristics of the paired p.

10. The edge computing device according to any one of claims 1 to 9, characterized in that, The main control MCU also calculates and evaluates the impact of hand rotation on swimming speed for different strokes. With the IMU fixed on the palm of the hand, the angular velocity vectors of roll, yaw, and pitch around the x, y, and z axes of the local coordinate system, with the IMU's center of mass as the origin, are obtained. ) and angular acceleration vector ; For a single hand position i, the motion vector is obtained as follows: The motion vectors of the left and right palms are concatenated to obtain the global regression input vector. ; Still using time-series data from the accelerometer, gyroscope, and magnetometer of the IMU in the lumbosacral region, each time step is obtained in the X, Y, Z geodetic coordinate system. The velocity vector of the body's center of mass forward and acceleration vector ,by Output a vector matrix for the target; Define and calculate the overall feature vector consisting of the comprehensive force feature vector of the palm, the technical stability feature vector of one side of the body, and the technical stability feature vector of both sides of the body; Define and calculate the force gradient index, phase coordination index, and bilateral rotational precession index of the palm; define and calculate the overall motion coordination index of both palms. An intelligent predictive evaluation model was established to assess the impact of the axial rotation of the palm around a local coordinate system on the forward velocity and acceleration of the body's center of mass, and the model was evaluated.

Images