A study on the optimization of athletes’ fitness assessment and training program based on sports biomechanics

Sport is a culturally rich physical activity, which aims to promote physical and mental health, and shows a special social phenomenon in the form of human physical activity. Athletes aim to achieve excellent results in sports competition by fully stimulating the maximum potential as the basis of the movement, and by continuously improving the physical ability and improving the movement technology [1-2]. In the context of the new era, based on the theories of sports biomechanics and other theories, high-precision instrumentation is utilized to detect and evaluate the data of athletes during and after exercise [3-4]. By exploring the behavioral characteristics and trajectories of human movement, sports biomechanics deeply explores the intrinsic laws of various sports and uses them to develop scientific and reasonable training plans for athletes [5-7]. It helps physical training and technical movement optimization in a biomechanical way to improve the stability and effectiveness of movements, thus significantly improving competitive sports performance [8-9].

In the field of sports science, especially in the study of sports biomechanics, the rapid development of intelligent technology promotes the study of biomechanical principles in the laws of human movement, which provides a reference for guiding athletes in sports training and improving their sports technology [10-12]. Combined with a large amount of sports biomechanics data, the use of intelligent algorithms can be used to analyze the strength output, energy consumption and other individual performance indicators, and take into account the athlete’s training needs, physical condition, past illness and injury risk and other factors, to customize a personalized training program for the athlete, to guide the athlete to adjust the posture, training intensity and other adjustments in order to improve the athlete’s strength and physical fitness [13-17]. Intelligent algorithms can also be used to construct biomechanical models of human movement, simulate different sports states, and analyze the data and performance of athletes during competitions and training, so as to realize the analysis and prediction of athletes’ performance and improve the competitive level of individuals and groups [18-21].

Exercise biomechanics plays an important role in the study of optimizing sport training programs. Fletcher, J. R. et al. optimized training patterns and equipment configuration for disabled athletes in specific sport events based on exercise biomechanics by considering athlete-specific barriers and performing mechanical assessments to achieve optimal physical preparation and performance [22]. Brazil, A. et al. revealed the overload and idiosyncratic characteristics of training choices of athletes from the perspective of sports biomechanics, in terms of perceptual cognition of physical training by considering the loading state of musculoskeletal versus extensor joints in order to facilitate the science as well as the coordination of training tasks [23]. Gou, X. et al. applied sports biomechanics to the process of optimizing the effect of sports training by assessing the key biomechanical factors of athletes’ movement techniques to improve the training program in order to maximize the effect of training and to assist coaches in developing a scientific training program [24]. Wang, B. conducted fitness tests on athletes using sports biomechanics methods and assessed the differences in the influence of the corresponding skeletal muscles on athletes’ performance based on the results [25]. Ma, Q. and Huo, P. established a human motion simulation model with knee flexion angle and muscle force changes, which was combined with experimental data on sports biomechanics to guide athletes’ training programs [26]. Al Ardha, M. A. et al. conducted an in-depth analysis of sports biomechanical factors affecting the effectiveness of hurdles training, and accordingly developed a training program to maximize the effects and minimize the injuries, which effectively enhanced the athletic performance of hurdlers [27]. ZhaoriGetu, H. and Li, C. showed that maintaining correct body postures during exercise can effectively enhance the effectiveness of physical training and reduce the probability of injury, and proposed a biomechanical feedback system based on an improved convolutional neural network to guide the training by capturing biomechanical responses in different movement states in order to improve the athlete’s training experience [28]. It can be found that sports biomechanics provides strong theoretical guidance and technical guarantee for athletes’ physical assessment and training guidance, and makes important contributions to the continuous development of sports science and the continuous progress of competitive sports.

In this paper, from the perspective of sports biomechanics, the video data of athletes in the process of physical training are collected and then extracted the athletes’ various limb stages and bone movement parameters according to the motion coordinate transformation method. Through the motion transformation of the local coordinate system of the joints, a simulation model of the human body posture is constructed in the computer system, and the calculation of the joint force and joint moment of the athlete’s body segments is carried out based on the human body’s multi-rigid body dynamics. The surface EMG information at the muscle belly of the athlete was monitored by combining wireless sensors and surface EMG wireless testing system. Finally, a fitness assessment and exercise monitoring system is constructed by combining the exercise simulation and EMG data acquisition module, and the analysis data presented by the system is used to assist in the formulation and decision-making of the exercise optimization training program.

2

Method

2.1

Motion Capture and Recognition Analysis Methods

2.1.1

Motion Coordinate Transformation and Motion Simulation

Motion capture of the athlete’s human skeletal posture based on sports biomechanics [29] is to reproduce the real-world motion in a computer through digital modeling. In this paper, the necessary assumptions and simplifications are made to the skeletal muscle model based on anatomical and physiological foundations, and only the motion parameters of each limb segment of the athlete or the bones therein need to be captured, and then the relative coordinates under the joint localized coordinate system are obtained based on the geometrical model of the skeletal-muscular system, and then the motion transformations of the joint localized coordinate system are utilized to drive the transformation of the coordinates of the corresponding points in the computer model.

Take a muscle for example, if its starting point A is attached to the humerus and its stopping point B is attached to the radius, then according to the local coordinate system of the upper limb, its starting point A can be transformed by the shoulder joint which is attached to the humerus, and similarly, its stopping point B can be transformed by the brachioradialis joint which is attached to the radius. As shown in equation (1), at first, the coordinate (x_AG, y_AG, z_AG) of the starting point A under the coordinate system G of the geometric model of the human skeletal muscular system is converted into the local coordinate value (x_As, y_As, z_As) under the shoulder joint coordinate system, and T_ς and R_ς in the formula are respectively the translation and rotation transformation matrices of the shoulder joint coordinate system, wherein x_s, y_s, z_s is the coordinate value of the origin of the shoulder joint coordinate system in the geometric model coordinate system, v_xr1, v_xr2, v_xr3 is the three-coordinate value of the direction vector of the xth axis of the shoulder joint coordinate system, and so on. The formula is as follows: (1) $\begin{array}{l} (x_{A s}, y_{A s}, z_{A s}, 1) = (x_{A G}, y_{A G}, z_{A G}, 1) T_{s} R_{s} \\ = (\begin{matrix} x_{A G}, y_{A G}, z_{A G}, 1 \end{matrix}) [\begin{matrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ - x_{s} & - y_{s} & - z_{s} & 1 \end{matrix}] [\begin{matrix} v_{s x 1} & v_{s y 1} & v_{s z 1} & 0 \\ v_{s x 2} & v_{s y 2} & v_{s z 2} & 0 \\ v_{s x 3} & v_{s y 3} & v_{s z 3} & 0 \\ 0 & 0 & 0 & 1 \end{matrix}] \end{array}$

According to the same principle of coordinate conversion [30], the coordinate (x_BG, y_BG, z_BG) of the dead center B under the coordinate system G of the geometric model of the human skeletal muscular system can also be converted into the local coordinate values (x_Br, y_Br, z_Br) under the coordinate system of the humeraldial joint, where T_r and R are respectively the translation and rotation transformation matrices of the humeraldial joint coordinate system, where x_r, y_r, z_i is the coordinate value of the origin of the brachioradial joint coordinate system in the geometric model coordinate system, v_ns1, v_nx2, v_nx3 is the three-coordinate value of the direction vector of the xth axis of the humeraldial joint coordinate system, and so on. The formula is: (2) $\begin{array}{l} (x_{B r}, y_{B r}, z_{B r}, 1) = (x_{B G}, y_{B G}, z_{B G}, 1) I_{r} R_{r} \\ = (x_{B G}, y_{B G}, z_{B G}, 1) [\begin{matrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ - x_{r} & - y_{r} & - z_{r} & 1 \end{matrix}] [\begin{matrix} v_{r x 1} & v_{r y 1} & v_{r z 1} & 0 \\ v_{r x 2} & v_{r y 2} & v_{r z 2} & 0 \\ v_{r x 3} & v_{r y 3} & v_{r z 3} & 0 \\ 0 & 0 & 0 & 1 \end{matrix}] \end{array}$

The above method allows for a quick local coordinate transformation, which was utilized in this study for the upper limb muscle start and stop points.

As shown in equation (3): (3) $\begin{array}{l} (x_{A M}, y_{A M}, z_{A M}, 1) = (x_{A s}, y_{A s}, z_{A s}, 1) {(R_{s})}^{- 1} {(T_{s})}^{- 1} \\ = (x_{A z}, y_{A z}, z_{A z}, 1) {[\begin{matrix} v_{x x 1} & v_{x y 1} & v_{z z 1} & 0 \\ v_{x x 2} & v_{x y 2} & v_{x z 2} & 0 \\ v_{x x 3} & v_{x y 3} & v_{x z 3} & 0 \\ 0 & 0 & 0 & 1 \end{matrix}]}^{- 1} [\begin{matrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ - x_{s} & - y_{s} & - z_{s} & 1 \end{matrix}] \end{array}$

The local coordinates (x_As, y_As, z_As) of the starting point A in the shoulder joint coordinate system are first converted to coordinates (x_AM, y_AM, z_AM) in the computer model coordinate system M by means of translation and rotation inverse transformations.

The local coordinates (x_Br, y_Br, z_Br) of the end point B of the athlete’s radial-humeral joint coordinate system can also be converted to coordinates (x_BM, y_BM, z_BM) of the computer model coordinate system M by the same principle of coordinate transformation: (4) $\begin{array}{l} (x_{B M}, y_{B M}, z_{B M}, 1) = (x_{B r}, y_{B r}, z_{B r}) {(R_{r})}^{- 1} {(T_{r})}^{- 1} \\ = (x_{B r}, y_{B r}, z_{B r}, 1) {[\begin{matrix} v_{r z 1} & v_{r y 1} & v_{r z 1} & 0 \\ v_{r z 2} & v_{r y 2} & v_{r z 2} & 0 \\ v_{r z 3} & v_{r y 3} & v_{r z 3} & 0 \\ 0 & 0 & 0 & 1 \end{matrix}]}^{- 1} [\begin{matrix} 1 & 0 & 0 & 0 \\ 0 & 1 & 0 & 0 \\ 0 & 0 & 1 & 0 \\ - x_{r} & - y_{r} & - z_{r} & 1 \end{matrix}] \end{array}$

This realizes the transformation of the movement coordinates of any muscle starting and stopping points of athletes in the process of sports training, and captures the movement of human body segments in reality only through a small number of specific marking points, which in turn drives the changes of the skeletal system and the muscle force lines in the computer, and also realizes the athletes’ movement posture capture and simulation.

2.1.2

Human body multi-rigid body dynamics calculations

To perform human multi-rigid body dynamics [31] calculations, it is necessary to analyze the forces on the body segments and to establish the half-balance equations containing the forces or moments to be solved. The following equilibrium equations are available for each rigid body in the non-dynamic basis problem: (5) ${\begin{matrix} m_{i} {\ddot{s}}_{i} = \sum F_{i} \\ {\dot{L}}_{i} = I_{i} {\dot{ω}}_{i} = \sum M_{i} \end{matrix}$

Where L_r is the momentum moment of rigid body i with respect to the center of mass, m_r and I_r are the mass of rigid body i and the rotational inertia about the center of simplification, ${\ddot{s}}_{i}$ and ${\dot{ω}}_{i}$ are the linear acceleration and angular acceleration of rigid body i, and $\sum F_{i}$ and $\sum M_{i}$ are the main vector of the force system and the main moment about the center of simplification, respectively, that act on rigid body i (the force system includes both externally-acting forces and moments as well as constraint counterforces and moments between and within the rigid bodies). Let the rigid body, whose center of mass is point P, be subjected to a constraint counterforce F^p and a constraint countermoment M^p at point o, and to an external force F^e at point P^e and to an external moment M^e. If the coordinate system is established at point o and used as the inertial base, we have, according to the Newton-Euler formula: (6) ${\begin{matrix} m \ddot{s} = F^{e} + F^{p} \\ I_{x} \ddot{α} + I_{y} \ddot{β} + I_{z} \ddot{γ} = M^{e} + M^{p} + M^{F} \end{matrix}$

where $\ddot{s}$ is the linear acceleration of the rigid body, $\ddot{α}, \ddot{β}, \ddot{γ}$ is the angular acceleration of the rigid body rotating about x, y, and z axes, respectively, I_x, I_y, I_z is the moment of inertia of the rigid body about the center of moments about x, y, and z axes, respectively, and M^F is the force F^e and F^p on the rigid body and the additional moment of inertia about the center of moments.

Thus for any rigid body i, there is a kinetic equation in matrix form: (7) $[\begin{matrix} m_{i} & 0 \\ 0 & I_{i} \end{matrix}] [\begin{matrix} {\ddot{s}}_{i} \\ {\dot{ω}}_{i} \end{matrix}] = [\begin{matrix} F_{i}^{e} \\ M_{i}^{e} \end{matrix}] + [\begin{matrix} F_{i}^{p} \\ M_{i}^{p} \end{matrix}] + [\begin{matrix} 0 \\ M_{i}^{F} \end{matrix}]$

where $I = [\begin{matrix} I_{x} & I_{y} & I_{z} \end{matrix}], \dot{ω} = {[\begin{matrix} \ddot{α} & \ddot{β} & \ddot{γ} \end{matrix}]}^{T}$ .

Taking the right upper limb of human body as an example, the upper limb can be divided into three body segments, namely upper arm, forearm, and hand, which are defined as Body Segment 1, Body Segment 2, and Body Segment 3, when the fingers of hand are ignored. A reference coordinate system is established with the center of the joints O as the origin, x axes as the sagittal backward, y axes as the coronal rightward, and z axes as the vertical upward axes. Where P_i(i = 1, 2, 3) is the center of mass of body segment i, ${\ddot{α}}_{i}, {\ddot{β}}_{i}, {\ddot{γ}}_{i}$ is the angular acceleration of body segment i rotating around x, y, and z axes, respectively, and ${\ddot{s}}_{i}$ is the linear acceleration of body segment i. G_i = m_ig is the gravitational force on body segment i, where m_i is the mass of body segment i. $F_{i}^{p}$ and $M_{i}^{p}$ are the reaction force and joint moment, respectively, on body segment i from its proximal neighboring body segments. $F_{i}^{d}$ and $M_{i}^{d}$ are the distal joint force and joint moment, respectively, applied to body segment i, and $F_{3}^{e}$ and $M_{3}^{e}$ are the external force and external moment, respectively, applied to body segment 3 at point $P_{3}^{e}$ , i.e., there is an external load acting on only the end body segment. The end body segment 3 is analyzed first, and then Newton’s third law is gradually deduced to body segment 2 and body segment 1.

The end body segment 3 is analyzed and is obtained from the Newton-Euler equations: (8) ${\begin{matrix} m_{3} {\ddot{s}}_{3} = F_{3}^{e} + F_{3}^{p} + G_{3} \\ I_{3 x} {\ddot{α}}_{3} + I_{3 y} {\ddot{β}}_{3} + I_{3 z} {\ddot{γ}}_{3} = M_{3}^{e} + M_{3}^{p} + M_{3}^{F} \end{matrix}$

where I_3x, I_3y, I_3z is the moment of inertia of the body segment 3 about the center of moments around the x-axis, y-axis, and z-axis, respectively, and $M_{3}^{F}$ is the principal moment of the spatial force system (including inertial forces) about the center of moments to which the body segment 3 is subjected. The principal vector of the force system through the simplified center and the principal moment of the force system about the simplified center are obtained: (9) ${\begin{array}{l} M_{x}^{F} = y F_{z} - z F_{y} \\ M_{y}^{F} = z F_{x} - x F_{z} \\ M_{z}^{F} = x F_{y} - y F_{x} \end{array}$

where x, y, z is the triple coordinate value of the simplified force point with respect to the simplified origin o.

Let the coordinates of point P₃ be (x_p3, y_p3, z_p3), the coordinates of point $P_{3}^{e}$ be $(x_{p 3}^{e}, y_{p 3}^{e}, z_{p 3}^{e})$ , and the coordinates of point P₂₃ be $(x_{p_{23}}, y_{p_{23}}, z_{p_{23}})$ , where the x, y, z of the subscript indicates the coordinate axes corresponding to the components of the axes of the vector in question. I.e: (10) ${\begin{matrix} F_{3 x}^{*} + F_{3 y}^{*} - m_{3} {\ddot{s}}_{3 x} = 0 \\ F_{3 y}^{*} + F_{3 z}^{*} - n_{3} {\ddot{s}}_{3 y} = 0 \\ F_{3 z}^{*} + F_{3 x}^{*} + G_{3} - m_{3} {\ddot{s}}_{3 z} = 0 \\ M_{x x}^{*} + M_{y y}^{*} - I_{x x} \ddot{α} - y_{3} p_{2} s_{x z} + z p_{3} m_{3} {\ddot{s}}_{3 z} + y_{3} p_{2} G_{3} = 0 \\ M_{y z}^{*} + M_{z x}^{*} - I_{y z} \ddot{β} - z p_{3} m_{3} {\ddot{s}}_{3 z} + x p_{3} m_{3} {\ddot{s}}_{3 x} - x p_{3} G_{3} = 0 \\ M_{z z}^{*} + M_{x y}^{*} - I_{z z} \ddot{γ} - y p_{3} m_{3} {\ddot{s}}_{3 y} + y p_{3} G_{3} = 0 \end{matrix}$

Solving yields the joint force and joint moment applied to body segment 3 as: (11) ${\begin{matrix} F_{3 x}^{p} = - F_{3 x}^{s} + m_{3} {\ddot{s}}_{3 x} \\ F_{3 y}^{p} = - F_{3 y}^{s} + m_{3} {\ddot{s}}_{3 y} \\ F_{3 z}^{p} = - F_{3 z}^{s} - G_{3} + m_{3} {\ddot{s}}_{3 z} \\ M_{3 x}^{p} = - M_{3 x}^{s} + I_{3} \ddot{α} + γ_{P} m_{3} {\ddot{s}}_{3 z} - z_{P} m_{3} {\ddot{s}}_{3 y} - γ_{P} G_{3} - ξ_{P}^{z} F_{3 z}^{s} + z_{P}^{z} F_{3 y}^{s} - y_{P}^{z} F_{3 z}^{p} + z_{P}^{2} F_{3 z}^{p} + Z_{P}^{2} F_{3 y}^{p} \\ M_{3 y}^{p} = - M_{3 y}^{s} + I_{3} \ddot{β} + Z_{P} m_{3} {\ddot{s}}_{3 z} - x_{P} m_{3} {\ddot{s}}_{3 y} + x_{P}^{2} F_{3 z}^{p} - ξ_{P}^{z} F_{3 x}^{s} + x_{P}^{x} F_{3 z}^{s} - z_{P}^{z} F_{3 x}^{p} + x_{P}^{2} F_{3 z}^{p} \\ M_{3 z}^{p} = - M_{3 z}^{s} + I_{3} \ddot{γ} + x_{P} m_{3} {\ddot{s}}_{3 y} - γ_{P} m_{3} {\ddot{s}}_{3 x} - ξ_{P}^{x} F_{3 y}^{s} + y_{P}^{x} F_{3 z}^{s} - x_{P}^{x} F_{3 y}^{p} + y_{P}^{2} F_{3 x}^{p} \end{matrix}$ (12) ${\begin{matrix} F_{1 x}^{p} = - F_{1 x}^{d} + m_{1} {\ddot{s}}_{1 x} \\ F_{1 y}^{p} = - F_{1 y}^{d} + m_{1} {\ddot{s}}_{1 y} \\ F_{1 z}^{p} = - F_{1 z}^{d} - G_{1} + m_{1} {\ddot{s}}_{1 z} \\ M_{1 x}^{p} = - M_{1 x}^{d} + I_{1} {\ddot{α}}_{1} - y_{p} m_{1} {\ddot{s}}_{2 x} + z_{p} m_{1} {\ddot{s}}_{2 y} - y_{p} G_{1} - y_{p 1} F_{1 z}^{d} + z_{p 1} F_{1 z}^{d} \\ M_{1 y}^{p} = - M_{1 y}^{d} + I_{1} {\ddot{β}}_{1} - z_{p} m_{1} {\ddot{s}}_{2 x} + x_{p} m_{1} {\ddot{s}}_{2 z} + x_{p} G_{1} - Z_{p 1} F_{1 x}^{d} + x_{p 1} F_{1 x}^{d} \\ M_{1 z}^{p} = - M_{1 z}^{d} + I_{1} {\ddot{γ}}_{1} - x_{p} m_{1} {\ddot{s}}_{2 y} + y_{p} m_{1} {\ddot{s}}_{2 z} + y_{p} G_{1} - Y_{p 1} F_{1 y}^{d} + y_{p 1} F_{1 y}^{d} \end{matrix}$

So far, the solution of joint force and joint moment in the training process of athletes has been realized.

2.2

Kinematic data acquisition

Based on the principles of sports biomechanics, this study simultaneously combined athletes’ kinematic data for fitness assessment and training program optimization. Surface EMG wireless test system produced by Delsye Company in the United States was used for monitoring, which mainly includes base station, wireless transmission electrode sensors, EMGworks acquisition software [32], EMGworks analysis software, 4 silver bars on the sensors are tightly affixed to the muscle belly of the monitoring muscle to collect surface EMG information, 16 sensors can be used simultaneously to monitor the surface EMG of 16 muscles The EMG signals are used for analysis, the bandwidth of EMG signals is 20-450hz, the sampling rate of EMG signals is 2Khz, the acquisition and analysis software is installed on a laptop (WiFi transmission) or a tablet (Bluetooth transmission) for use. Advantages of the system the electrodes for collecting surface EMG signals realize wireless transmission, which does not affect the athletes’ technical movements during the monitoring process. The analysis software has a powerful function of processing data, and at the same time has the function of realizing timely feedback for athletes.

2.3

Physical fitness assessment and training decision support system construction

2.3.1

Basic model for training decision support systems

According to the basic features and functions of sports training decision support system, the basic model of sports training decision support system can be constructed. The basic model of a complete decision support system mainly includes the decision support system itself, the real system, the external environment and the decision maker, and there is an intrinsic close relationship among the four. The basic model of sports training optimization decision support system is shown in Figure 1. Among them, the decision makers (coaches) are in the main position, they use their own knowledge and experience, the response output of the decision support system, and combined with their management of the “real system” to analyze, so as to make a decision on the optimization of athletes’ training. For a “real system”, the questions asked and the data manipulated are the output streams, while the decision makers’ decisions are the input streams. The lower part of the figure shows the basic data related to the decision support system, which mainly includes processed information from the “real system”, environmental information, information related to the training behavior of athletes, MIS information, and statistical information, etc. The lower part of the figure shows the basic data related to the decision support system. On the right side of the figure is the decision support system, which consists of a database system, a model repository system and a human-computer interaction system. The decision maker (i.e., the coach) uses his knowledge and experience, combined with the corresponding output of the decision support system, to make decisions on the optimization of the sports training program for the “real system” he manages.

2.3.2

Decision support systems for sports fitness and training

On the basis of the analysis of the decision support system for the assessment and monitoring of athletes’ training fitness, the next task to be carried out is the design of the training optimization decision-making system, whose overall structural scheme is shown in Figure 2. The system design in the development phase of the athlete training optimization decision support system first determines the specialized physical components, structure and development platform of the design, and then the whole system design process is divided into preliminary design and detailed design. The preliminary design of the decision support system is mainly to complete the overall design of the system, problem decomposition and problem synthesis. The detailed design of the decision support system is mainly to carry out the detailed design of the data and model, the design of the data includes the data file design and database design, and the model design includes the model algorithm design and model library design. The decision support system design in this study is mainly the design of system data, model and knowledge, in the process of data design in the form of a database, the design of the model includes the design of the model data description file and the model description file organization and storage of the model library, the mathematical model is expressed in the form of mathematical equations and algorithmic design is proposed for the model equations, and the model is designed in the form of effective algorithms and then the computer is utilized to language to prepare the program. The design of the structural scheme of the athletes’ training fitness assessment monitoring and decision support system adopts the multi-library synergistic technology of database, model library and knowledge library, and realizes the diagnosis and evaluation of athletes’ sports training functions, training control and other system functions through the management of the unified interface.

3

Results and discussion

3.1

Analysis of the accuracy of athletic training data

3.1.1

Data collection and processing

The acquisition software for this study was under the Vicon Nexus computer software for kinematic, kinetic, and surface EMG data acquisition. In this study, kinematic data were captured by 12 Vicon Vantage V5 cameras for 3D motion capture at 120 HZ. Two AMTI force tables (model HPS400600) were used for kinetic data acquisition at 2000 Hz. The surface EMG signals of the athletes were collected and analyzed by wireless transmission of electrode sensors and EMGworks acquisition software at a frequency of 2000 Hz.

The subjects of this experiment were 24 athletes from a middle school sports team in Beijing. After opening the windows and ventilating the room, the indoor temperature was maintained at about 25 degrees Celsius to ensure that the environment did not interfere with the test acquisition. The test environment was darkened to keep it free of reflective materials except for the Marker points. The testers familiarized themselves with the test procedure, prepared the experimental consumables, informed the subjects of the test and signed an informed consent form, and then recorded the height, weight, lower limb length, knee width, ankle width, elbow width, and wrist width of the subjects, the purpose of which was to construct a skeleton model for the Vicon Nexus. Subjects were asked to perform 10-15 minutes of preparatory activities, including jogging, stretching, and dynamic bouncing and stretching, to increase nervous system excitability, reduce muscle stickiness, and prevent muscle strains.

3.1.2

Comparative validation of kinematic data

1)

Verification of movement posture angle data

Taking the angle change data of an athlete’s arm, leg, knee and ankle as samples, the results of the motion data collected by the Vicon device were compared and analyzed for errors with the system analysis results, and the results of the comparative analysis of the motion angles of the body parts are shown in Fig. 3, with (a)-(d) representing the results of the analysis of the angle changes of the arms, legs, knees and ankles, respectively. The results show that the two are highly similar, proving the reliability of the system. Calculating the data error between the two, it is found that the errors between the athlete’s arm, leg, knee and ankle angles recognized by the system and the real values are 2.36°, 0.10°, 1.12° and 0.38°, respectively, which are small, indicating that the accuracy of the motion capture and recognition method proposed in this paper is high.

2)

Calculation and Verification of Motion Residual Force

The reasonable range of motion residual force is 0-25N, and the residual moment is 0-50Nm. Because the squatting and jumping movements, which require large movements of the hip, knee and ankle joints, are different from the gait and running movements in the training process of athletes, the values can be slightly out of the reasonable range. The results of the residual analysis of the subjects are shown in Table 1, which shows that the residual force (0.068N~2.892N) and residual moment (0.231Nm~18.539Nm) of the 24 subjects in this study are within a reasonable range.

3)

Surface EMG data validation

The EMG data collected by the sensors were preprocessed using EMGworks software, which sequentially performed a 4th order 50 Hz high pass filtering process, full wave rectification, and a 4th order 20 Hz low pass filtering process to produce a linear envelope. The EMG signals were then carefully visually inspected, and low-pass filtering at lower frequencies was used to treat any high-frequency noise interference present in the signals. The linear envelope of EMG was then normalized using the amplitude maxima of the block of muscle during the action cycle, and finally the linear envelope was normalized to 101 data points from the beginning to the end of the action and the data was saved as a file in CSV format. The specific movement phase was selected, and Scale scaling, IK inverse kinematics, ID inverse dynamics and SO optimization analysis were run sequentially to obtain the muscle activation changes during the jumping process, and the data of the right rectus femoris muscle was selected, which was subjected to the intra-group similarity analysis with the rectus femoris muscle signals after the envelope. The results of EMG linear envelope and EMGworks software analysis are shown in Fig. 4, and the two showed some differences at the peak of EMG data changes, such as the true values at 26.45% and 38.44% of the exercise cycle and the analyzed values of the EMG data were 0.600, 0.464 and 0.856, 0.731, respectively. However, the average error was maintained at around 0.072, and the comparison results showed that the trends of the two EMG data were highly similar, proving the reliability of the model.

Table 1.

The results of the subjects’ residual differences

Subjects	FX(N)	FY(N)	FZ(N)	MX(Nm)	MY(Nm)	MZ(Nm)
1	1.646	0.548	0.209	9.435	8.329	7.751
2	2.351	0.88	2.144	7.738	3.554	13.329
3	2.89	1.773	2.115	9.862	9.208	8.533
4	1.808	0.11	2.705	1.767	7.507	15.426
5	1.597	2.334	1.64	10.407	7.552	6.278
6	1.64	0.487	1.683	15.345	6.738	15.869
7	2.305	1.928	2.502	7.724	8.715	10.974
8	0.388	0.542	0.905	15.158	1.531	16.643
9	2.892	1.431	1.432	2.847	11.654	4.174
10	0.717	1.253	0.612	0.231	16.359	5.965
11	1.766	2.566	2.48	0.564	3.336	15.152
12	1.072	1.57	0.368	4.191	17.559	8.202
13	2.583	0.345	2.546	7.362	13.466	18.351
14	2.561	0.224	1.676	16.03	11.9	4.629
15	2.524	2.867	0.806	17.758	8.003	6.077
16	1.699	0.58	2.325	0.882	0.629	3.905
17	2.413	2.567	1.906	7.891	5.876	5.268
18	1.359	2.562	2.818	6.479	14.143	8.635
19	0.068	2.533	0.494	13.259	5.389	17.997
20	2.824	2.155	2.878	0.489	3.128	11.234
21	1.703	0.579	1.568	4.997	2.367	7.322
22	0.353	2.388	2.125	18.306	11.427	16.163
23	1.889	1.181	2.726	2.491	5.374	16.115
24	1.049	2.835	0.379	10.294	18.539	18.066

3.2

Analysis of the optimization effect of sports training

3.2.1

Experimental Objects

Twenty-four athletes were randomly divided into an optimization training group and a traditional training group by random grouping, with 12 athletes in the optimization training group applying the sports physical fitness monitoring, assessment and training program optimization decision-making system, and 12 athletes in the traditional training group still using the traditional way of training, and each of them underwent a 10-week sports training intervention. The basic information of the athletes in the two groups is shown in Table 2, which shows that there is no significant difference in the age (P=0.558>0.05), height (P=0.264>0.05), weight (P=0.352>0.05), and years of training (P=0.358>0.05) of the athletes.

Table 2.

Basic information of the experimental object

Basic information	Optimize training group (n=12)		Traditional training group (n=12)		T	P
Basic information	Mean	SD	Mean	SD	T	P
Age	16.59	1.52	16.84	0.94	0.529	0.558
Height (cm)	185.63	9.48	186.42	10.75	1.114	0.264
Weight (kg)	74.52	9.65	74.82	9.76	1.063	0.352
Training period	5.26	1.32	5.29	1.54	0.952	0.358

3.2.2

Experimental design

From the cycle of sports training and sports biomechanics theory, all sports cycle training program arrangements must follow, determine the main competition date → determine the main competition stage → determine the main competition period → determine the entire training cycle of the four steps, in turn, to develop a scientific training cycle plan. The practical application of physical fitness training program also needs to follow the above steps, based on the sports team’s competition schedule, the training program into the whole cycle of training. The athletic team in this high school is a traditional strong team in Beijing, and in terms of the annual competition schedule of the sports team in 2023, the 16th Games in mid-July is the key event for the team to prepare for. According to the team’s annual training program, after the Beijing Youth Championships in early February, the team enters a recovery period, and March-July is a complete training cycle to prepare for the Games. In conjunction with the team’s cycle schedule, this study conducted a 10-week physical and athletic training intervention from March 16-May 31, 2023. Through expert interviews, this study decided to divide the training ratio of the three phases of physical-motor function training as 1, 2, and 3, i.e., the basic motor function training was scheduled in the first 2 weeks (the recovery period scheduled in the previous cycle), the general motor function training phase was scheduled in weeks 3-6 (the preparation period), and the specialized motor function training phase was in weeks 7-10 (the later part of the preparation period - the earlier part of the competition period). Training was conducted twice a week for 150 min each time on Mondays and Wednesdays. The pre-intervention test was conducted on March 14-15, 2023, and the post-test was conducted on June 1-2, 2023.

Athletes in the optimized training group used the fitness assessment monitoring and training optimization decision-making system proposed in this paper to assist the intervention in the process of physical training, and the coaches combined the physical fitness data and sports data analyzed by the system to develop a reasonable training optimization plan for each athlete. In addition to the different contents of the optimization training group and the traditional training group, the training duration, frequency, number of training groups, interval time and other arrangements were the same. In addition, the experimental subjects were consistent in the content of technical and tactical training, and all athletes did not receive other training content except technical and tactical training. The testers, testing methods, and models of testing instruments and equipment were the same before and after the intervention, and the temperature and humidity of the venues were the same before and after the test.

3.2.3

Physical training optimization results

Combining the experience of coaches and the research on sports, this paper sets the test indexes of athletes’ physical training as speed, sensitivity and strength. Among them, the speed test indexes are 30m sprint run and 5×30m sprint run, the sensitivity test index is Illinois run, and the strength test indexes are vertical jump and pull-up. 1)

Speed test (1)

30m sprint run

The results of the comparative analysis of the intervention effect of 30 m sprint performance are shown in Figure 5, (a) and (b) represent the results of the comparative analysis of the pre-intervention and post-intervention of the athletes in the optimized training group and the traditional training group, respectively. Before the intervention, there was no significant difference between the 30-m sprint speed of the athletes in the optimized training group and the traditional training group (P=0.152>0.05). Comparison between groups revealed that athletes in the optimized training group improved their 30-m sprint run significantly more than those in the traditional training group after the intervention (P=0.001<0.01). This implies that the physical training optimization program specified based on the sports biomechanics data and systematic decision-making can effectively improve the athletes’ 30-m sprint performance, and the magnitude of improvement is better than that of the conventional physical training program.

(2)

5 x 30-meter sprint run

The intervention effects of repeated sprint performance are shown in Table 3, where fatigue index = (5th sprint speed - 1st sprint speed)/1st sprint speed. Prior to the intervention, there was no significant difference in the mean speed and fatigue index of the 5×30 m sprints between the athletes in the optimized and traditional training groups (P=0.825, 0.558>0.05). Comparison between the groups revealed that the mean speed and fatigue index of repeated sprints of the athletes in the optimized training group were significantly improved after the intervention (P<0.01), and the mean time required for the 5×30 m sprint run was reduced from 4.98 s to 4.51 s. The athletes in the traditional training group had only a significant improvement in the mean speed (P=0.048<0.05), and there was a non-significant improvement in fatigue index (P=0.084>0.05). 0.084>0.05). Comparison between the groups revealed that the average speed of repeated sprints of the athletes in the optimized training group was significantly higher than that of the traditional training group after the intervention (P=0.000), and there was also a significant difference in the fatigue index (P=0.000). This implies that the physical exercise training system can effectively improve the performance of repetitive sprinting and is better than the conventional physical training program in terms of the improvement of the average speed of repetitive sprinting.

2)

Results and analysis of sensitivity quality

The results and analysis of Illinois running indexes before and after the intervention of the two groups of athletes are shown in Table 4. In the paired-sample t-test of the two groups after the intervention, the mean value of Illinois in the optimized training group was improved from 16.58±1.54s before the intervention to 15.24±0.59s after the intervention, T=2.596, P=0.032 (P<0.05), and the Illinois indexes showed a significant difference after intervention, and the Illinois in the traditional training group did not have any significant difference after the intervention. It indicates that intelligent physical training has more positive effects on improving the quality of sensitivity than traditional training. The independent samples t-test of the two groups after the experimental intervention yielded that there was a significant difference between the Illinois sensitivity of the optimized training group (15.24±0.59s) and the Illinois sensitivity of the traditional training group (16.30±0.87s) (T=4.526, P=0.005, P<0.05).

3)

Strength quality results and analysis

The results of the comparison between the two groups of athletes before and after the experimental intervention of longitudinal jump and pull-up data are shown in Figure 6, (a) and (b) represent the comparison results before and after the intervention, respectively. From the results of vertical jump analysis, the average value of vertical jump of the optimized training group increased from 41.47cm before the experiment to 48.11cm after the experiment, P=0.000<0.01, and the vertical jump was highly significant difference after the experiment. It indicates that the intelligent assisted training system has a significant effect on improving athletes’ strength quality. Through the independent sample t-test of the two groups after the experimental intervention, it was concluded that the longitudinal jump of the optimized training group (48.11cm) showed a highly significant difference with the longitudinal jump of the traditional training group (45.18cm) (T=6.529, P=0.000, P<0.01). In the analysis of pull-up data, it was found that after the intervention experiment the optimized training group improved by about 3 while the traditional training group improved by about 1, indicating that intelligent training based on sports biomechanics had a greater effect on the optimized training group in pull-ups than the traditional training group in traditional training.

Table 3.

Analysis of the effectiveness of the 5×30 meter sprint

Repeat sprint (s)
Group	Before intervention		After intervention		T	P
Group	Mean	SD	Mean	SD	T	P
Optimize training	4.98	0.52	4.51	0.25	7.425	0.000
Traditional training	4.96	0.58	4.88	0.51	1.263	0.048
T	0.042		4.526		——	——
P	0.825		0.001		——	——

Fatigue index
Group	Before intervention		After intervention		T	P
Group	Mean	SD	Mean	SD	T	P
Optimize training	7.85	2.64	4.29	1.48	6.523	0.000
Traditional training	7.94	2.54	7.64	2.93	0.623	0.084
T	0.623		1.263		——	——
P	0.558		0.045		——	——

Table 4.

The results and analysis of the Illinois trail index were conducted (s)

Group	Before intervention		After intervention		T	P
Group	Mean	SD	Mean	SD	T	P
Optimize training	16.58	1.54	15.24	0.59	2.596	0.032
Traditional training	16.48	0.85	16.30	0.87	1.326	0.095
T	0.524		4.526		——	——
P	0.775		0.005		——	——

4

Conclusion

Based on the theories of sports biomechanics and human body multi-rigid body dynamics, this paper simulates and analyzes human posture in physical training by constructing a coordinate system of athletes’ human skeletal posture. And then equipped with the athlete muscle surface electromyography data acquisition program design system to monitor and evaluate the physical fitness of each athlete, to support the coaches to carry out physical training optimization program decision-making. The analysis of the accuracy of the system motion data analysis and the effect of system application revealed that: 1)

The error values between the results of the exercise posture angle data collected by the Vicon device and the results of the system analysis were small, and the monitoring results of the subject’s exercise residual force (0.068N~2.892N) and residual moment (0.231Nm~ 18.539Nm) were all within reasonable ranges, which indicated that the system possessed good accuracy.

2)

After 10 weeks of sports training intervention practice, the athletes in the optimized training group showed significantly higher improvement in the 30-m sprint run than those in the traditional training group (P=0.001<0.01), and there was also a significant difference between the Illinois Sensitivity (15.24±0.59s) and the Illinois Sensitivity (16.30±0.87s) in the traditional training group (T=4.526, P= 0.005, p<0.05). In addition, in terms of strength quality improvement, the effect of sports biomechanics-based intelligent training on athletes in the optimized training group was greater than that of traditional training in the traditional training group.

The sports biomechanics-based physical training monitoring and evaluation system can assist athletes to carry out efficient and reasonable training, and effectively improve the level of physical fitness. It is suggested that coaches should widely apply this system to sports training and design scientific and efficient physical training programs to improve the physical fitness level of athletes.

Lingua:: Inglese

Frequenza di pubblicazione:: 1 volte all'anno
Argomenti della rivista:: Scienze biologiche, Scienze della vita, altro, Matematica, Matematica applicata, Matematica generale, Fisica, Fisica, altro

Feed RSS della rivista

A study on the optimization of athletes’ fitness assessment and training program based on sports biomechanics

Yali Wang

Pubblicato online: 26 set 2025

Ricevuto: 05 gen 2025

Accettato: 30 apr 2025

DOI: https://doi.org/10.2478/amns-2025-1035

Parole chiaveSports biomechanics, Sports coordinate transformation, Electromyographic information, Physical fitness assessment, Training optimization

© 2025 Yali Wang, published by Sciendo.

This work is licensed under the Creative Commons Attribution 4.0 International License.

Parole chiave
Sports biomechanics, Sports coordinate transformation, Electromyographic information, Physical fitness assessment, Training optimization