Physical education teaching management system based on big data

Abstract

The invention relates to the field of education health management, and discloses a physical education teaching management system based on big data, and the system comprises a data collection module which collects physiological, motion, emotion and environment data through a wearable device, a video monitor and a sensor; the data storage module is used for storing and managing structured, semi-structured and unstructured data by adopting a big data architecture; the data processing and analyzing module is used for performing data cleaning, preprocessing, feature extraction, dimension reduction and data fusion; the intelligent decision-making module is used for realizing health risk monitoring, exercise recommendation and teaching optimization by using a deep learning model; and the front-end display and interaction module provides data visualization, real-time feedback and personalized reports. Through real-time acquisition and hierarchical storage of multi-modal data of physiology, movement, emotion and environment, comprehensive monitoring of health and movement states of students is realized, rapid response and trend analysis are supported, and a basis is provided for safety guarantee and scientific decision in teaching.

Description

Technical Field

[0001] The present invention relates to the technical field of education and health management, and in particular to a physical education teaching management system based on big data. Background Art

[0002] As people's awareness of health and physical exercise continues to increase, physical education plays an increasingly important role in schools and communities. Traditional physical education teaching management systems mainly rely on teachers' experience and observations, lack of scientific data support, making it difficult to fully understand students' sports performance and health status. In recent years, with the development of information technology and wearable devices, data collection and analysis technology has gradually been introduced into physical education, making data-based management models possible.

[0003] In the existing technology, data collection methods usually rely on a single data source, such as monitoring physiological indicators only through wearable devices, or analyzing sports performance through video recording. This single data collection method is difficult to provide a comprehensive assessment of students' sports status and easily ignores the impact of emotions and environmental factors on sports performance. In addition, the existing system mostly uses traditional database management methods for data storage and processing, which cannot effectively process large-scale, multi-modal data, limiting the system's intelligence and real-time response capabilities;

[0004] In addition, many current sports teaching management systems lack dynamic adjustment capabilities and are unable to optimize teaching plans in a timely manner based on students’ real-time status. Health risk monitoring is often insufficient, making it difficult to accurately identify potential health problems, increasing safety hazards during exercise. Students’ personalized needs are difficult to meet, resulting in some students being unable to obtain exercise advice and support that suits them.

[0005] Therefore, a new physical education teaching management system based on big data is urgently needed to adapt to the higher requirements of education quality and student health in the new era. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a physical education teaching management system based on big data, which solves the problem of lack of comprehensive and real-time data monitoring, analysis and personalized guidance in traditional physical education teaching management.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A physical education teaching management system based on big data, comprising:

[0008] Data collection module, used to collect physiological data, sports performance data, emotional data and environmental data through wearable devices, video monitoring equipment and environmental sensors;

[0009] A data storage module is used to store and manage the above data, and adopts a big data storage architecture to support the storage of structured, semi-structured and unstructured data;

[0010] Data processing and analysis module, used for data cleaning, preprocessing, feature extraction and dimensionality reduction, and feature-level and decision-level fusion of multimodal data;

[0011] Intelligent decision-making module, which is used to conduct real-time health risk monitoring, personalized exercise recommendations and teaching strategy optimization based on multimodal deep learning models;

[0012] The front-end display and interaction module is used to provide data visualization, real-time feedback and personalized report generation.

[0013] Preferably, the data processing and analysis module includes a dynamic weight allocation submodule, which can dynamically adjust the fusion weights of physiological data, sports performance data, emotional data and environmental data according to the real-time status of the student, and specifically includes the following steps:

[0014] a. State recognition: through real-time data analysis, identify the student's current state as high exercise load, low exercise load or psychological tension;

[0015] b. Weight adjustment: Dynamically assign weights to each modality data based on the state of recognition, increase the weights of physiological data and sports performance data under high exercise load conditions, and increase the weights of emotional data and environmental data under low exercise load conditions;

[0016] c. Fusion feature calculation: Calculate the fusion feature vector F according to the adjusted weight value weighted , the fusion formula is:

[0017] F weighted =αf phys +βf motion +γf emo +δf env

[0018] where f phys 、f motion 、f emo 、f env are physiological characteristics, motion characteristics, emotional characteristics, and environmental characteristics, respectively. α, β, γ, and δ are weights adjusted according to the real-time status;

[0019] The physiological data includes heart rate, cadence, body temperature, blood oxygen level and heart rate variability; the sports performance data includes movement speed, movement trajectory, jump height, displacement and joint angle; the emotional data includes facial expression characteristics, voice emotion characteristics and emotion intensity; and the environmental data includes temperature, humidity, air quality and light intensity.

[0020] Preferably, the data acquisition module includes:

[0021] The physiological data collection submodule is used to collect students' heart rate, cadence, body temperature and blood oxygen through wearable devices;

[0022] The sports performance data collection submodule is used to collect students' sports speed, step frequency and posture through video monitoring equipment;

[0023] The emotional data collection submodule is used to collect students’ emotional states through facial recognition and voice emotion analysis;

[0024] The environmental data acquisition submodule is used to collect data such as temperature, humidity and air quality through environmental sensors.

[0025] Preferably, the data storage module includes a data lake architecture, which is used to store real-time data, historical data and cold data in layers and supports batch processing and stream processing, specifically including:

[0026] Real-time data partition: used to store real-time physiological data, sports performance data, emotional data and environmental data obtained from the data acquisition module, supporting fast query and incremental update of real-time data;

[0027] Historical data partition: used to store processed and feature-extracted historical data, support periodic analysis and trend mining, and optimize data storage and reading efficiency;

[0028] Cold data partition: used to store historical data that has not been called for a long time, reducing storage resource consumption, and is only called for low-frequency analysis;

[0029] Partition data alignment and fusion: The consistency of each partition data is achieved through timestamp alignment, and the time alignment of real-time data is completed based on the following formula:

[0030]

[0031] Among them, f(τ) represents the aligned data at time τ, and t1 and t2 are adjacent time points.

[0032] Preferably, the data processing and analysis module includes a data cleaning submodule, and the data cleaning submodule cleans the collected data by an outlier detection method based on Mahalanobis distance, and the outlier detection formula is:

[0033]

[0034] Among them, x is the data sample, μ is the sample mean, Σ is the covariance matrix, Σ- 1 is the inverse of the covariance matrix.

[0035] Preferably, the data processing and analysis module includes a time synchronization and alignment submodule, and linear interpolation is used to time align data of different modalities, and the interpolation formula is:

[0036]

[0037] Among them, f(τ) is the interpolation result at time τ, which means the result of aligning the data sampled at different times through interpolation. t1 and t2 are two adjacent actual sampling time points, and the data value at a specific time point is generated by interpolation. f(t1) and f(t2) represent the actual data values ​​collected at time t1 and t2 respectively. τ is the time point to be aligned, and the estimated data at this time point is generated by interpolation.

[0038] Preferably, the multimodal data fusion module combines physiological features, motion features, emotional features and environmental features into a comprehensive feature vector through feature-level fusion and decision-level fusion. The feature-level fusion formula is:

[0039] F=[f phys ,f motion ,f emo ,f env ]

[0040] Among them, F is the comprehensive feature vector, f phys is the feature combination of the collected physiological data, f motion is the feature combination of sports performance data, f emo is the feature combination of sentiment data, f env It is a feature combination of environmental data.

[0041] Preferably, the multimodal data fusion module further comprises a dynamic weight allocation submodule, which adjusts the weight according to the importance of different modal data, and the final fused feature representation is:

[0042] F weighted =αf phys +βf motion +γf emo +δf env

[0043] Among them, F weighted is the feature vector after weighted fusion, αf phys is the weight coefficient of physiological characteristics, βf motion is the weight coefficient of motion features, γf emo is the weight coefficient of emotional features, δf env is the weight coefficient of environmental characteristics.

[0044] Preferably, the intelligent decision-making module includes a health risk monitoring submodule, which is based on the multimodal fusion feature F fusion Calculate the health risk probability P(risk) and generate warning information when it exceeds the preset threshold.

[0045] Preferably, the front-end display and interaction module includes a personalized report generation submodule for generating a visual report of athletic performance, health status and emotional changes based on the student's historical data.

[0046] The present invention provides a physical education teaching management system based on big data. It has the following beneficial effects:

[0047] 1. The present invention realizes comprehensive monitoring of students' health and exercise status through real-time collection and hierarchical storage of multimodal data on physiology, movement, emotion and environment, supports rapid response and trend analysis, and provides a basis for safety assurance and scientific decision-making in teaching.

[0048] 2. The present invention combines dynamic weight allocation with intelligent decision-making to adjust the exercise plan according to the real-time status and individual differences of students, making exercise recommendations and teaching strategies more personalized and adaptable, thereby effectively improving teaching results.

[0049] 3. The present invention uses real-time health risk monitoring to dynamically evaluate students' health risks based on multimodal fusion features. Once the safety threshold is exceeded, an immediate warning is issued to help teachers and students adjust exercise plans in a timely manner to ensure safety during exercise. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a system module flow chart of the present invention;

[0051] Figure 2 It is a schematic diagram of a data acquisition module of the present invention;

[0052] Figure 3 It is a schematic diagram of the data processing and analysis module of the present invention;

[0053] Figure 4 It is a schematic diagram of the data processing and analysis module of the present invention;

[0054] Figure 5 It is a schematic diagram of a multimodal data fusion module of the present invention;

[0055] Figure 6 It is a schematic diagram of the intelligent decision-making module of the present invention;

[0056] Figure 7 It is a schematic diagram of the front-end display and interaction module of the present invention. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] Please refer to the attached Figure 1 The embodiment of the present invention provides a physical education teaching management system based on big data, including:

[0059] Data collection module, used to collect physiological data, sports performance data, emotional data and environmental data through wearable devices, video monitoring equipment and environmental sensors;

[0060] A data storage module is used to store and manage the above data, and adopts a big data storage architecture to support the storage of structured, semi-structured and unstructured data;

[0061] Data processing and analysis module, used for data cleaning, preprocessing, feature extraction and dimensionality reduction, and feature-level and decision-level fusion of multimodal data;

[0062] Intelligent decision-making module, which is used to conduct real-time health risk monitoring, personalized exercise recommendations and teaching strategy optimization based on multimodal deep learning models;

[0063] The front-end display and interaction module is used to provide data visualization, real-time feedback and personalized report generation.

[0064] Specifically, first, the data acquisition module collects students' physiological data, sports performance data, emotional data and environmental data in a multi-modal manner through wearable devices, video surveillance equipment and environmental sensors, providing a comprehensive and real-time data input basis for the system. The results of data acquisition are transmitted to the data storage module in real time. The data storage module adopts a big data storage architecture, and manages real-time data, historical data and cold data reasonably by storing structured, semi-structured and unstructured data in layers, ensuring that the system can meet the needs of fast access while supporting long-term storage and periodic analysis of data. Next, the data processing and analysis module cleans, preprocesses, extracts features and reduces dimensions of the data, and fuses multi-modal data at the feature level and decision level, ensuring data quality while integrating multi-modal data into a unified comprehensive feature to provide deep correlation data for intelligent analysis and decision-making. Based on the multi-modal deep learning model, the intelligent decision-making module performs real-time health risk monitoring, personalized exercise recommendations and teaching strategy optimization on the fused data, generates personalized exercise plans suitable for each student, and timely detects and warns potential health risks to ensure the personalization and security of the system. Finally, the front-end display and interaction module displays data analysis results and personalized feedback through a visual interface, providing real-time health and sports guidance information, so that teachers, students and parents can understand and adjust exercise plans in a timely manner. The close cooperation between the modules enables the system to achieve a complete closed loop from data collection, storage, analysis to decision-making feedback, while improving the efficiency of teaching management, and providing personalized support and safety for students' sports health.

[0065] Please refer to the attached Figure 4 ,The data processing and analysis module includes a dynamic weight allocation submodule, which can dynamically adjust the fusion weights of physiological data, sports performance data, emotional data and environmental data according to the real-time status of the students, specifically including the following steps:

[0066] a. State recognition: through real-time data analysis, identify the student's current state as high exercise load, low exercise load or psychological tension;

[0067] b. Weight adjustment: Dynamically assign weights to each modality data based on the state of recognition, increase the weights of physiological data and sports performance data under high exercise load conditions, and increase the weights of emotional data and environmental data under low exercise load conditions;

[0068] c. Fusion feature calculation: Calculate the fusion feature vector F according to the adjusted weight value weighted , the fusion formula is:

[0069] F weighted =αf phys +βf motion +γf emo +δf env

[0070] where f phys 、f motion 、f emo 、f env are physiological characteristics, motion characteristics, emotional characteristics, and environmental characteristics, respectively. α, β, γ, and δ are weights adjusted according to the real-time status;

[0071] Physiological data include heart rate, cadence, body temperature, blood oxygen level and heart rate variability; sports performance data include movement speed, movement trajectory, jump height, displacement and joint angle; emotional data include facial expression characteristics, voice emotion characteristics and emotion intensity; environmental data include temperature, humidity, air quality and light intensity.

[0072] Specifically, the data processing and analysis module includes a dynamic weight allocation submodule, which dynamically adjusts the fusion weight of each modal data by analyzing the real-time status of the students, so as to more accurately reflect the comprehensive influence of the students' physical, emotional and external environment in different states; first, the dynamic weight allocation submodule will perform state recognition based on real-time data. The recognition process will continuously monitor and analyze physiological, motor, emotional and environmental data to determine whether the students are currently in a state of high exercise load, low exercise load or psychological tension; this recognition step provides a scientific basis for weight adjustment to ensure that the dynamic allocation of weights can reflect the actual state of the students. After state recognition, the module will adjust the weights according to the judgment results; the weight adjustment step uses the relative importance of different modal data to assign corresponding weights to each modal data; for example, when the student is in a state of high exercise load, the weights of physiological data and sports performance data will be increased, so that the system pays more attention to the impact of physical load on exercise; in a state of low exercise load, the weights of emotional data and environmental data will be increased to more carefully capture the impact of external environment and psychological factors on the student's state; through this state-based The weight adjustment of the system can more accurately capture the changes of students in different situations, reduce the deviation of single data on the analysis results, and improve the adaptability and flexibility of the system analysis. Finally, the fusion feature calculation step calculates the comprehensive feature vector according to the adjusted weight value; in this step, the dynamically assigned weight value will be multiplied by each modal feature to generate a fusion feature vector; the calculation formula of the fusion feature multiplies the physiological characteristics, movement characteristics, emotional characteristics and environmental characteristics by the adjusted weight coefficient respectively, and adds the weighted characteristics to form a unified comprehensive feature; the fusion feature vector can reflect the current overall state of the student, ensuring that the system is based on the optimal feature combination after dynamic adjustment when making intelligent decisions, thereby improving the accuracy and reliability of functions such as health risk monitoring and personalized exercise recommendations. Among them, the specific data items contained in each modal data, such as heart rate, cadence, body temperature, etc. in physiological data, movement speed, displacement, etc. in sports performance data, and specific indicators of emotional data and environmental data, enable the system to comprehensively and meticulously capture the real situation of students in different states, thereby realizing more personalized and data-driven physical education teaching management.

[0073] Please refer to the attached Figure 2 , the data acquisition module includes:

[0074] The physiological data collection submodule is used to collect students' heart rate, cadence, body temperature and blood oxygen through wearable devices;

[0075] The sports performance data collection submodule is used to collect students' sports speed, step frequency and posture through video monitoring equipment;

[0076] The emotional data collection submodule is used to collect students’ emotional states through facial recognition and voice emotion analysis;

[0077] The environmental data acquisition submodule is used to collect data such as temperature, humidity and air quality through environmental sensors.

[0078] Specifically, the physiological data acquisition submodule is used to collect students' physiological status information through wearable devices, including heart rate, cadence, body temperature and blood oxygen level; wearable devices can monitor these physiological indicators in real time, which is particularly critical under high exercise load conditions, to ensure that they can capture and feedback in time when students have abnormal physiological conditions; heart rate and cadence are used to analyze students' exercise intensity and endurance level, body temperature monitoring can evaluate their body temperature regulation ability in high temperature or high-intensity exercise environments, and blood oxygen level helps monitor students' oxygen supply, especially during strenuous exercise to evaluate their blood oxygen changes; the data input of this submodule can provide key physiological reference indicators for the system's health risk monitoring; the sports performance data acquisition submodule is used to collect students' physiological status information through wearable devices. The video monitoring equipment records the students' sports performance data, including sports speed, cadence and posture; this module combines computer vision technology to analyze video data, capture speed changes, sports posture and body posture during exercise, and support the evaluation of students' sports techniques and movement norms; sports speed can reflect students' sports rhythm and sports performance, cadence is used to judge their sports stability, and posture monitoring can identify bad sports habits or postures to reduce the risk of sports injuries; through these data, the system can analyze students' sports techniques, coordination and sports efficiency, and provide basic data for the optimization of personalized sports plans; the emotional data collection submodule is used to collect students' emotional states through facial recognition and voice emotion analysis technology; this The module uses image analysis to obtain facial expression features, and uses voice emotion recognition to determine students' psychological state during exercise, including emotional tension, pleasure, and concentration. Facial expression recognition determines students' emotional reactions by capturing micro-expressions and changes in expression, while voice emotion analysis evaluates students' emotions through factors such as voice intonation and rhythm. Emotional data can reflect students' attitudes and psychological acceptance towards exercise, which helps the system optimize exercise plans and teaching strategies, and enhance students' exercise experience and psychological comfort. The environmental data acquisition submodule uses environmental sensors to collect external environmental data, mainly including temperature, humidity, and air quality. These data are used to analyze the impact of environmental factors on students' exercise status, such as high temperature. The workload of exercise in a high-humidity environment is higher, and air quality also has an important impact on students' respiratory system; temperature and humidity monitoring ensure that the system can dynamically evaluate the comfort of the sports venue, and air quality monitoring helps to determine the potential impact of external air pollution on students' sports performance, especially in outdoor venues; the introduction of environmental data enables the system to fully consider the actual impact of the external environment on students when making exercise recommendations or health risk assessments; through multimodal collection of four types of data: physiological, sports performance, emotional and environmental, the data acquisition module provides the system with multi-dimensional, real-time student status information, ensuring that the system has comprehensive data support in subsequent analysis and decision-making, and achieving accurate health monitoring and personalized exercise recommendations.

[0079] Please refer to the attached Figure 3The data storage module includes a data lake architecture, which is used to store real-time data, historical data, and cold data in layers, and supports batch processing and stream processing, including:

[0080] Real-time data partition: used to store real-time physiological data, sports performance data, emotional data and environmental data obtained from the data acquisition module, supporting fast query and incremental update of real-time data;

[0081] Historical data partition: used to store processed and feature-extracted historical data, support periodic analysis and trend mining, and optimize data storage and reading efficiency;

[0082] Cold data partition: used to store historical data that has not been called for a long time, reducing storage resource consumption, and is only called for low-frequency analysis;

[0083] Partition data alignment and fusion: The consistency of each partition data is achieved through timestamp alignment, and the time alignment of real-time data is completed based on the following formula:

[0084]

[0085] Among them, f(τ) represents the aligned data at time τ, and t1 and t2 are adjacent time points.

[0086] Specifically, the data storage module adopts a data lake architecture to perform hierarchical storage and efficient management of different types of data in the system; through the hierarchical storage mechanism of real-time data partitions, historical data partitions, and cold data partitions, the module not only meets the system's demand for rapid access to real-time data, but also supports batch processing and long-term storage of large-scale historical data, providing a stable data management foundation for the system's periodic analysis and trend mining; the real-time data partition is used to store real-time physiological data, sports performance data, emotional data, and environmental data obtained from the data acquisition module; this partition is designed with efficient query and fast incremental update as the goal, ensuring that real-time data can be obtained and used by the system in a timely manner to meet the system's health and risk requirements. The real-time needs of functions such as risk monitoring, status identification and personalized recommendations can be met; real-time data can be directly called by the intelligent decision-making module through the fast query interface to achieve instant warning and dynamic adjustment of health risks; in a high-frequency data collection environment, the fast response characteristics of the real-time data partition effectively support the system's efficient processing and instant response needs for real-time data; the historical data partition is used to store historical data that has undergone data processing and feature extraction, which facilitates the system to analyze and mine trends in long-term data; this partition retains the key data features processed in the system, such as heart rate fluctuations in physiological data, speed change trends in sports performance, and long-term changes in emotional data; this partition can not only store the extracted core features It can not only perform batch processing and periodic analysis operations on these historical data, but also support long-term tracking of students' physical health, mood swings and sports performance. In addition, historical data partitions improve data storage efficiency and reading speed through data compression and index optimization, ensuring that data query needs can still be efficiently responded to in the case of large-scale data. Cold data partitions are used to store historical data that has not been called for a long time, reducing the system's storage resource usage and only calling it when low-frequency analysis is needed. The design of this partition is based on the low access frequency of cold data. Through data storage optimization strategies, cold data is archived and compressed to save system storage costs and computing resources. The data in the cold data partition is mainly used for long-term The cold data partition can be used for historical retrospective analysis of archived and low-frequency data, such as a comprehensive review of students' long-term health data, or to generate statistical reports based on historical data accumulated in the cold data partition; the low-cost and low-occupancy design of the cold data partition ensures that the system can still run efficiently and stably as the data accumulation increases. The partition data alignment and fusion are used to maintain the consistency of the data in each partition while storing it in layers; through timestamp alignment, the real-time data, historical data and cold data are ensured to be consistent in time; and the calculation of the formula ensures that the data in different partitions are seamlessly connected in time, providing time-consistent input data for subsequent multimodal data fusion and intelligent analysis, thereby improving the data integrity and analysis accuracy of the system.

[0087] Please refer to the attached Figure 4The data processing and analysis module includes a data cleaning submodule. The data cleaning submodule cleans the collected data through an outlier detection method based on the Mahalanobis distance. The outlier detection formula is:

[0088]

[0089] Among them, x is the data sample, μ is the sample mean, Σ is the covariance matrix, Σ -1 is the inverse matrix of the covariance matrix;

[0090] The data processing and analysis module includes a time synchronization and alignment submodule, which uses linear interpolation to time align data of different modalities. The interpolation formula is:

[0091]

[0092] Among them, f(τ) is the interpolation result at time τ, which means the result of aligning the data sampled at different times through interpolation. t1 and t2 are two adjacent actual sampling time points, and the data value at a specific time point is generated by interpolation. f(t1) and f(t2) represent the actual data values ​​collected at time t1 and t2 respectively. τ is the time point to be aligned, and the estimated data at this time point is generated by interpolation.

[0093] Specifically, the data cleaning submodule in the data processing and analysis module effectively identifies and removes abnormal data that deviates from the normal distribution through an outlier detection method based on the Mahalanobis distance, thereby ensuring the accuracy and consistency of multimodal data; this method uses the mean and covariance characteristics of the data to automatically detect outliers, making the data quality more stable and laying a high-quality data foundation for subsequent feature extraction and analysis; the time synchronization and alignment submodule uses linear interpolation to align the sampling times of different modal data to generate estimated data on a unified time axis; by seamlessly connecting the temporal aspects of each modal data, the system can achieve temporal consistency of multimodal data, ensuring the integrity and synchronization of data during the analysis and fusion process; the combination of data cleaning and time alignment improves the credibility and temporal consistency of the data, providing high-quality data support for the execution of functions such as health risk monitoring and personalized exercise recommendations.

[0094] Please refer to the attached Figure 5 In the multimodal data fusion module, physiological features, motion features, emotional features and environmental features are combined into a comprehensive feature vector through feature-level fusion and decision-level fusion. The feature-level fusion formula is:

[0095] F=[f phys ,f motion ,f emo ,f env ]

[0096] Among them, F is the comprehensive feature vector, f phys is the feature combination of the collected physiological data, f motion is the feature combination of sports performance data, f emo is the feature combination of sentiment data, f env is a feature combination of environmental data;

[0097] The multimodal data fusion module further includes a dynamic weight allocation submodule, which adjusts the weights according to the importance of different modal data. The final fused feature representation is:

[0098] F weighted =αf phys +βf motion +γf emo +δf env

[0099] Among them, F weighted is the feature vector after weighted fusion, αf phys is the weight coefficient of physiological characteristics, βf motion is the weight coefficient of motion features, γf emo is the weight coefficient of emotional features, δf env is the weight coefficient of environmental characteristics.

[0100] Specifically, the multimodal data fusion module provides comprehensive input for the system's intelligent decision-making and personalized analysis; the design of feature-level fusion aims to preliminarily fuse data features of different modalities to eliminate data deviations that may be caused by a single modality, so that the system can obtain all-round data support; through the feature-level fusion formula, the combination of features of each modality constitutes a comprehensive feature vector, which represents the overall state of the student; this fusion method combines the core information extracted from physiological characteristics, motion characteristics, emotional characteristics and environmental characteristics, ensuring that the system can rely on multi-dimensional state characteristics in subsequent analysis, thereby improving the depth and accuracy of comprehensive analysis; in addition, the multimodal data fusion module further includes a dynamic weight allocation submodule, which is used to adjust the weight according to the importance of each modal data to generate the final weighted fusion feature representation; dynamic weight allocation enables the system to flexibly adjust the importance of each modal feature under different states by setting a weight coefficient for each feature; for example, under high exercise load conditions, the weights of physiological characteristics and motion characteristics can be increased, Reduce the proportion of emotional characteristics and environmental characteristics so that the analysis focuses on the changes in the state of physical load; in the case of large emotional fluctuations or low exercise load, increase the weight of emotional characteristics and environmental characteristics to capture the influence of external environment and psychological state; this weight adjustment mechanism ensures the adaptability and flexibility of the system in different situations and can more accurately reflect the current state of students; through the dynamic weighted fusion formula, the system weightedly combines physiological characteristics, movement characteristics, emotional characteristics and environmental characteristics according to the weight coefficient to generate a weighted fused feature vector, which integrates the optimal combination of each modal feature; this fusion feature can not only reflect the comprehensive situation of the student in the current state, but also ensure that the system can rely on the most relevant information to make judgments when making intelligent decisions; the multimodal data fusion module combines feature-level fusion with dynamic weight allocation to achieve deep fusion and adaptive adjustment of multimodal data, enhance the accuracy of the system in functions such as health risk monitoring and personalized recommendation, and provide flexible and comprehensive input for intelligent analysis.

[0101] Please refer to the attached Figure 6 -Attached Figure 7 The intelligent decision-making module includes a health risk monitoring submodule, which is based on the multimodal fusion feature F fusion Calculate the health risk probability P(risk) and generate warning information when it exceeds the preset threshold;

[0102] The intelligent decision-making module includes a health risk monitoring submodule, which is based on the multimodal fusion feature F fusion Calculate the health risk probability P(risk) and generate warning information when it exceeds the preset threshold.

[0103] Specifically, the intelligent decision-making module includes a health risk monitoring submodule, which calculates the health risk probability of students by analyzing multimodal fusion features and generates warning information when it exceeds the preset threshold; the health risk monitoring submodule uses the physiological, motor, emotional and environmental characteristics of multimodal fusion to evaluate the current state of students through deep learning models or probabilistic calculation models to form a probabilistic estimate of health risks; the system determines whether the student is in a potential high-risk state based on the calculated health risk probability value. If the health risk probability value exceeds the set safety threshold, the system will immediately trigger the warning mechanism to notify teachers and students to take appropriate measures; the calculation of health risk probability combines the real-time feature input of multimodal data, including heart rate fluctuations, exercise intensity, emotional tension and environmental conditions, to ensure the comprehensiveness and accuracy of risk assessment; by integrating multi-dimensional features such as physiological and psychological, the health risk monitoring submodule can identify potential risks that are difficult to detect with single modal data; for example, a high heart rate When the student's physical condition is high and accompanied by emotional tension, risk monitoring will comprehensively consider the superposition effect of these two factors, identify the possibility that the student is in a high-pressure state, and generate timely warnings; this module can also dynamically adapt to different exercise intensities or environmental conditions. For example, in high-temperature or high-intensity exercise environments, it is more sensitive to students' physiological and emotional characteristics to improve the accuracy of risk detection; the warning information generated by this submodule can not only prompt the current risk, but also provide a brief risk cause and recommended measures, so that teachers and students can quickly understand the warning content and make adjustments; through real-time health risk monitoring and warning information, the system can intervene in potential health risks in a timely manner, avoid safety hazards caused by excessive physical fatigue or psychological burden, and ensure students' sports safety and health; the health risk monitoring submodule in the intelligent decision-making module provides the system with the ability to control health risks in real time, ensures the safety and scientificity of students in physical education, and makes the system more practical in realizing intelligent monitoring and personalized management.

[0104] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A physical education teaching management system based on big data, characterized in that: include: Data collection module, used to collect physiological data, sports performance data, emotional data and environmental data through wearable devices, video monitoring equipment and environmental sensors; A data storage module is used to store and manage the above data, and adopts a big data storage architecture to support the storage of structured, semi-structured and unstructured data; Data processing and analysis module, used for data cleaning, preprocessing, feature extraction and dimensionality reduction, and feature-level and decision-level fusion of multimodal data; Intelligent decision-making module, which is used to conduct real-time health risk monitoring, personalized exercise recommendations and teaching strategy optimization based on multimodal deep learning models; The front-end display and interaction module is used to provide data visualization, real-time feedback and personalized report generation.

2. A physical education teaching management system based on big data according to claim 1, characterized in that: The data processing and analysis module includes a dynamic weight allocation submodule, which can dynamically adjust the fusion weights of physiological data, sports performance data, emotional data and environmental data according to the real-time status of the student, and specifically includes the following steps: a. State recognition: through real-time data analysis, identify the student's current state as high exercise load, low exercise load or psychological tension; b. Weight adjustment: Dynamically assign weights to each modality data based on the state of recognition, increase the weights of physiological data and sports performance data under high exercise load conditions, and increase the weights of emotional data and environmental data under low exercise load conditions; c. Fusion feature calculation: Calculate the fusion feature vector F according to the adjusted weight value weighted , the fusion formula is: F weighted =αf phys +βf motion +γf emo +δf env where f phys 、f motion 、f emo 、f env are physiological characteristics, motion characteristics, emotional characteristics, and environmental characteristics, respectively. α, β, γ, and δ are weights adjusted according to the real-time status; The physiological data includes heart rate, cadence, body temperature, blood oxygen level and heart rate variability; the sports performance data includes movement speed, movement trajectory, jump height, displacement and joint angle; the emotional data includes facial expression characteristics, voice emotion characteristics and emotion intensity; and the environmental data includes temperature, humidity, air quality and light intensity.

3. A physical education teaching management system based on big data according to claim 1, characterized in that: The data acquisition module comprises: The physiological data collection submodule is used to collect students' heart rate, cadence, body temperature and blood oxygen through wearable devices; The sports performance data collection submodule is used to collect students' sports speed, step frequency and posture through video monitoring equipment; The emotional data collection submodule is used to collect students’ emotional states through facial recognition and voice emotion analysis; The environmental data acquisition submodule is used to collect data such as temperature, humidity and air quality through environmental sensors.

4. A physical education teaching management system based on big data according to claim 1, characterized in that: The data storage module includes a data lake architecture, which is used to store real-time data, historical data and cold data in layers, and supports batch processing and stream processing, specifically including: Real-time data partition: used to store real-time physiological data, sports performance data, emotional data and environmental data obtained from the data acquisition module, supporting fast query and incremental update of real-time data; Historical data partition: used to store processed and feature-extracted historical data, support periodic analysis and trend mining, and optimize data storage and reading efficiency; Cold data partition: used to store historical data that has not been called for a long time, reducing storage resource consumption, and is only called for low-frequency analysis; Partition data alignment and fusion: The consistency of each partition data is achieved through timestamp alignment, and the time alignment of real-time data is completed based on the following formula: Among them, f(τ) represents the aligned data at time τ, and t1 and t2 are adjacent time points.

5. A physical education teaching management system based on big data according to claim 1, characterized in that: The data processing and analysis module includes a data cleaning submodule, which cleans the collected data by using an outlier detection method based on Mahalanobis distance. The outlier detection formula is: Among them, x is the data sample, μ is the sample mean, Σ is the covariance matrix, Σ -1 is the inverse of the covariance matrix.

6. A physical education teaching management system based on big data according to claim 1, characterized in that: The data processing and analysis module includes a time synchronization and alignment submodule, which uses linear interpolation to time align data of different modalities. The interpolation formula is: Among them, f(τ) is the interpolation result at time τ, which means the result of aligning the data sampled at different times through interpolation. t1 and t2 are two adjacent actual sampling time points, and the data value at a specific time point is generated by interpolation. f(t1) and f(t2) represent the actual data values ​​collected at time t1 and t2 respectively. τ is the time point to be aligned, and the estimated data at this time point is generated by interpolation.

7. A physical education teaching management system based on big data according to claim 1, characterized in that: In the multimodal data fusion module, physiological features, motion features, emotional features and environmental features are combined into a comprehensive feature vector through feature-level fusion and decision-level fusion. The feature-level fusion formula is: F=[f phys ,f motion ,f emo ,f env ] Among them, F is the comprehensive feature vector, f phys is the feature combination of the collected physiological data, f motion is the feature combination of sports performance data, f emo is the feature combination of sentiment data, f env It is a feature combination of environmental data.

8. A physical education teaching management system based on big data according to claim 1, characterized in that: The multimodal data fusion module further includes a dynamic weight allocation submodule, which adjusts the weights according to the importance of different modal data. The final fused feature representation is: F weighted =αf phys +βf motion +γf emo +δf env Among them, F weighted is the feature vector after weighted fusion, αf phys is the weight coefficient of physiological characteristics, βf motion is the weight coefficient of motion features, γf emo is the weight coefficient of emotional features, δf env is the weight coefficient of environmental characteristics.

9. A physical education teaching management system based on big data according to claim 1, characterized in that: The intelligent decision-making module includes a health risk monitoring submodule, which is based on the multimodal fusion feature F fusion Calculate the health risk probability P(risk) and generate warning information when it exceeds the preset threshold.

10. A physical education teaching management system based on big data according to claim 1, characterized in that: The front-end display and interaction module includes a personalized report generation submodule, which is used to generate a visual report on sports performance, health status and emotional changes based on the student's historical data.

Images