Physical training risk assessment method and system based on artificial intelligence

Abstract

The invention relates to a physical training risk assessment method and system based on artificial intelligence, and belongs to the field of physical training, and the method comprises the following steps: (a) collecting the physiological and motion data of soldiers in the training process in real time; (b) the collected data are processed and then transmitted to a central server; (c) analyzing the collected data and evaluating a training risk; (d) when the training risk exceeds a preset threshold value, sending out an early warning signal; and (e) displaying a risk assessment result to a coach and soldiers. The method has the advantages that the accuracy of risk assessment is improved, the training risk can be assessed more accurately by comprehensively considering various factors such as physiological data, motion data, environmental data and individual differences, and misjudgment is reduced. Data are collected and analyzed in real time in the training process, risks can be found in time, early warning can be given out, coaches and soldiers can take corresponding measures, and the risk occurrence possibility is reduced.

Description

Technical Field

[0001] The invention relates to a physical training risk assessment method based on artificial intelligence, and belongs to the field of physical training. Background Art

[0002] Exercise load is one of the main indicators for evaluating the risk of physical training personnel. Through heart rate information, exercise time, weight information and individual exercise history records, a five-level monitoring system is carried out, and real-time monitoring is carried out according to the maximum heart rate percentage. At the same time, the load is divided into aerobic recovery, aerobic endurance, mixed oxygen training, anaerobic endurance, maximum capacity and other levels. At the same time, the training effect is evaluated.

[0003] In existing physical training, the assessment of training risks currently relies mainly on experience and simple physiological indicator monitoring, such as simply observing the soldier's complexion, breathing, etc., or monitoring limited physiological parameters such as heart rate. However, this approach has the following problems:

[0004] (1) Limited accuracy: Experience-based judgment is highly subjective and prone to misjudgment. Simple heart rate monitoring cannot fully reflect the training risk. Different individuals may have different exercise loads and risks at the same heart rate.

[0005] (2) Lack of systematicity: data processing capabilities are limited, and it is impossible to comprehensively consider the impact of multiple factors on training risks, such as environmental factors (temperature, humidity, altitude, etc.), soldiers' physical condition (past medical history, recent fatigue, etc.), training intensity and type, etc.;

[0006] (3) Lack of real-time performance: Failure to timely and dynamically assess risks and issue early warnings during training may result in measures being taken only after risks occur, affecting soldiers’ health and training effectiveness. Summary of the invention

[0007] In order to overcome the defects of the prior art, the present invention provides a physical training risk assessment method and system based on artificial intelligence. The present invention aims at the problems that the previous sports load monitoring is highly subjective, the data processing capacity is limited, and the need to consider multiple physical sign data and possible cross-influences increases the complexity of the assessment. A more accurate, comprehensive and real-time physical training risk assessment method and system are provided to reduce training risks and ensure the physical health and training effect of soldiers. The technical solution of the present invention is:

[0008] A physical training risk assessment method based on artificial intelligence comprises the following steps:

[0009] (a) Real-time collection of soldiers’ physiological and motion data during training;

[0010] (b) processing the collected data and transmitting it to a central server;

[0011] (c) Analyze the collected data and assess training risks;

[0012] (d) Send out warning signals when the training risk exceeds a preset threshold;

[0013] (e) Present risk assessment results to trainers and Soldiers.

[0014] The step (a) is specifically:

[0015] 1.1 Use wearable devices to collect physiological indicators such as heart rate, blood pressure, and body temperature;

[0016] 1.2 Collect motion trajectory, speed, and acceleration data through GPS tracking devices, Beidou satellite navigation system, or motion sensors;

[0017] 1.3 Deploy environmental sensors to collect temperature, humidity, and altitude environmental parameters;

[0018] The heart rate data is collected by using a heart rate belt to collect photoplethysmography (PPG) signals, formula: Where V(t) is the PPG signal amplitude at time t, Vmax is the maximum amplitude, and τ is the time constant;

[0019] The blood pressure data is collected by measuring the blood pressure using a sphygmomanometer by oscillometric method or auscultatory method, according to the formula: P = ρ·g·h; wherein P is the blood pressure, ρ is the blood density, g is the acceleration of gravity, and h is the height of the blood column;

[0020] The body temperature data collection uses a temperature patch or an ear thermometer to measure body temperature, the formula is: T = T 环境 +ΔT, where T is body temperature, Tambient is ambient temperature, and ΔT is the difference between body temperature and ambient temperature;

[0021] The motion trajectory data collection is performed by using a GPS tracker or Beidou satellite navigation system to collect location data, the formula: location (t) = (x (t), y (t)); wherein (x (t), y (t)) is the geographic location coordinate at time t;

[0022] The velocity and acceleration data are collected by using an accelerometer and a gyroscope: Acceleration formula: Where a is acceleration, Δv is velocity change, and Δt is time interval; velocity formula: Among them, v is velocity, g is gravitational acceleration, and h is displacement;

[0023] The data types collected by the environmental data collection include temperature, humidity, altitude and atmospheric pressure;

[0024] The temperature and humidity data collection uses a temperature and humidity sensor;

[0025] Temperature formula: T = Tsensor; where T is the ambient temperature and Tsensor is the temperature measured by the sensor;

[0026] Altitude data collection uses a barometer to measure altitude, and the altitude formula is: Among them, h is the altitude, P is the current atmospheric pressure, P0 is the standard atmospheric pressure, and h0 is the standard altitude.

[0027] The step (b) is specifically:

[0028] 2.1 Data denoising: Gaussian filtering algorithm is used to denoise the collected multi-source data, specifically:

[0029]

[0030] Among them, I′(x, y) is the denoised data value, I(x, y) is the original data value, and μ x , μ y is the center point of the data, σ is the standard deviation of the Gaussian distribution;

[0031] 2.2 Missing value filling: Missing values ​​in multi-source data are filled by linear interpolation. Where x1, x2 are known data points, y1, y2 are the corresponding known data values, and y is the estimated value of the missing data point;

[0032] 2.3 Outlier Identification: Use Z-score algorithm anomaly detection algorithm to identify outliers.

[0033] The step (c) is specifically:

[0034] 3.1 Feature extraction: Extract risk assessment features from preprocessed data.

[0035]

[0036] Where xi is the data point, μ is the mean, and σ is the standard deviation;

[0037] Heart Rate Variability (HRV): Among them, RRi is the interval between adjacent heartbeats;

[0038] 3.2 Build and apply the risk assessment model to calculate the training risk based on the extracted features. The specific algorithm is:

[0039] Support Vector Machine (SVM):

[0040] where αi is the Lagrange multiplier, yi is the class label, and xi is the support vector.

[0041] Deep learning model: Recurrent Neural Network (RNN): h t =σ(W ih x t +b ih +W hh h t-1 +b hh );

[0042] Where ht is the hidden state at time t, xt is the input, W and b are weights and biases. 3.3 Calculate the risk index based on the model output, formula:

[0043] Risk Index: Risk Index = w1·F1+w2·F2+...+w n ·F n , where Fn is the risk score corresponding to the feature and wn is the weight;

[0044] 3.4 Set the warning threshold of risk index:

[0045] Threshold setting: Threshold = μ Risk Index +k·σ Risk Index ;

[0046] Where μ and σ are the mean and standard deviation of the risk index, respectively, and k is the coefficient.

[0047] The step (d) is specifically:

[0048] 4.1 Real-time monitoring of training risk index:

[0049] The average risk index in the preset time period is calculated in real time through the sliding window algorithm. The formula is: Where Risk is the risk index of the i-th time window, and n is the size of the time window;

[0050] 4.2 Evaluate whether the current risk index exceeds the preset safety threshold. If the current risk index exceeds the threshold, trigger an early warning;

[0051] 4.3. Early warning signal generation;

[0052] 4.4 Send early warning notifications through communication channels;

[0053] 4.5. Record warning events and record warning events for future analysis and audit;

[0054] 4.6 Database record: store detailed information of warning events in the database;

[0055] 4.7 Feedback loop: Collect responses and feedback to warnings to improve risk assessment models;

[0056] Update the risk assessment model based on the collected feedback data. new =Model old +ΔModel;

[0057] Among them, ΔModel is the model parameter adjusted according to the feedback.

[0058] A system for implementing a physical training risk assessment method based on artificial intelligence, comprising:

[0059] Data acquisition module, used to collect soldiers' physiological and motion data in real time during training;

[0060] A data transmission module, used to transmit the collected data to a central server;

[0061] AI analysis module, used to analyze collected data and assess training risks;

[0062] An early warning module is used to issue an early warning signal when the training risk exceeds a preset threshold;

[0063] The result display module is used to display the risk assessment results to trainers and soldiers.

[0064] The advantages of the present invention are:

[0065] Improve the accuracy of risk assessment: By comprehensively considering multiple factors, such as physiological data, exercise data, environmental data and individual differences, training risks can be assessed more accurately and misjudgments can be reduced.

[0066] Enhanced real-time performance: Real-time data collection and analysis during training can promptly identify risks and issue early warnings, allowing coaches and soldiers to take appropriate measures to reduce the possibility of risks.

[0067] Personalized assessment: Taking individual differences into consideration, we provide each soldier with personalized risk assessment and training suggestions to improve training effectiveness while ensuring the physical health of soldiers.

[0068] Improve training management level: The system can provide coaches with detailed risk assessment reports and training suggestions to help coaches better formulate training plans and manage the training process, thereby improving the scientificity and effectiveness of training management. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is the main structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0070] The present invention will be further described below in conjunction with specific embodiments, and the advantages and features of the present invention will become clearer as the description proceeds. However, these embodiments are exemplary only and do not constitute any limitation to the scope of the present invention. It should be understood by those skilled in the art that the details and forms of the technical solution of the present invention may be modified or replaced without departing from the spirit and scope of the present invention, but these modifications and replacements all fall within the scope of protection of the present invention.

[0071] See also Figure 1 The present invention relates to a physical training risk assessment method based on artificial intelligence, comprising the following steps:

[0072] (a) Real-time collection of soldiers’ physiological and motion data during training;

[0073] (b) processing the collected data and transmitting it to a central server;

[0074] (c) Analyze the collected data and assess training risks;

[0075] (d) Send out warning signals when the training risk exceeds a preset threshold;

[0076] (e) Present risk assessment results to trainers and Soldiers.

[0077] Based on the settings in the above steps, the following are achieved:

[0078] Improve safety: Real-time monitoring of soldiers' physiological status and athletic performance can provide timely warnings when training risks are too high, thereby reducing training injuries.

[0079] Enhanced training effectiveness: Through personalized risk assessment, coaches are able to adjust training plans to accommodate individual differences in soldiers, thereby improving training effectiveness.

[0080] Objectivity: Using artificial intelligence algorithms to analyze data reduces the bias of human subjective judgment and makes risk assessment more objective.

[0081] Comprehensiveness: It takes into account physiological data, exercise data and environmental factors to provide a comprehensive evaluation result.

[0082] Real-time: Real-time data collection and processing ensure the timeliness and accuracy of evaluation results.

[0083] Reduced manpower requirements: Automated risk assessment reduces reliance on manual monitoring and saves human resources.

[0084] The step (a) is specifically:

[0085] 1.1 Use wearable devices (such as smart bracelets, heart rate belts, etc.) to collect physiological indicators such as heart rate, blood pressure, and body temperature;

[0086] 1.2 Collect motion trajectory, speed, and acceleration data through GPS tracking devices, Beidou satellite navigation system, or motion sensors;

[0087] 1.3 Deploy environmental sensors to collect temperature, humidity, and altitude environmental parameters;

[0088] The heart rate data is collected by using a heart rate belt to collect photoplethysmography (PPG) signals, formula: Where V(t) is the PPG signal amplitude at time t, Vmax is the maximum amplitude, and τ is the time constant;

[0089] The blood pressure data is collected by measuring the blood pressure using a sphygmomanometer by oscillometric method or auscultatory method, the formula: P = ρ·g·h; where P is the blood pressure, is the blood density, g is the acceleration due to gravity, and h is the height of the blood column;

[0090] The body temperature data collection uses a temperature patch or an ear thermometer to measure body temperature, the formula is: T = T 环境 +ΔT, where T is body temperature, Tambient is ambient temperature, and ΔT is the difference between body temperature and ambient temperature;

[0091] The motion trajectory data collection is performed by using a GPS tracker or Beidou satellite navigation system to collect location data, the formula: location (t) = (x (t), y (t)); wherein (x (t), y (t)) is the geographic location coordinate at time t;

[0092] The velocity and acceleration data are collected by using an accelerometer and a gyroscope: Acceleration formula: Where a is acceleration, Δv is velocity change, and Δt is time interval; velocity formula: Among them, v is velocity, g is gravitational acceleration, and h is displacement;

[0093] The data types collected by the environmental data collection include temperature, humidity, altitude and atmospheric pressure;

[0094] The temperature and humidity data collection uses a temperature and humidity sensor;

[0095] Temperature formula: T = Tsensor; where T is the ambient temperature and Tsensor is the temperature measured by the sensor;

[0096] Altitude data collection uses a barometer to measure altitude. Altitude formula: Among them, h is the altitude, P is the current atmospheric pressure, P0 is the standard atmospheric pressure, and h0 is the standard altitude.

[0097] The above steps achieve:

[0098] Comprehensiveness: By using multiple types of sensors and devices, the soldiers' physiological indicators, athletic performance and environmental conditions can be comprehensively collected to provide comprehensive data support for risk assessment.

[0099] Real-time: Real-time data collection can capture soldiers’ health status and athletic performance in a timely manner, making risk assessment more timely and accurate.

[0100] High precision: Using specialized sensors and devices (such as heart rate belts, GPS trackers, environmental sensors) can provide high-precision data and improve the accuracy of risk assessment.

[0101] Objectivity: Automatic data collection by equipment reduces interference from human factors and makes the data more objective and reliable.

[0102] The step (b) is specifically:

[0103] 2.1 Data denoising: Gaussian filtering algorithm is used to denoise the collected multi-source data, specifically:

[0104]

[0105] Among them, I′(x, y) is the denoised data value, I(x, y) is the original data value, and μ x , μ y is the center point of the data, σ is the standard deviation of the Gaussian distribution;

[0106] 2.2 Missing value filling: Missing values ​​in multi-source data are filled by linear interpolation. Where x1, x2 are known data points, y1, y2 are the corresponding known data values, and y is the estimated value of the missing data point;

[0107] 2.3 Outlier Identification: Use Z-score algorithm anomaly detection algorithm to identify outliers.

[0108] The above steps achieve:

[0109] Data denoising: Gaussian filtering improves the signal-to-noise ratio of the signal, making the data smoother and providing more accurate input for subsequent analysis.

[0110] Missing value filling: Missing values ​​are filled by linear interpolation to avoid the impact of data incompleteness on analysis results and maintain data continuity.

[0111] Outlier identification: Use the Z-score algorithm to effectively identify and process outliers, reducing the interference of abnormal data on training risk assessment.

[0112] The step (c) is specifically:

[0113] 3.1 Feature extraction: Extract risk assessment features from preprocessed data.

[0114]

[0115] Where xi is the data point, μ is the mean, and σ is the standard deviation;

[0116] Heart Rate Variability (HRV): Among them, RRi is the interval between adjacent heartbeats;

[0117] 3.2 Build and apply the risk assessment model to calculate the training risk based on the extracted features. The specific algorithm is:

[0118] Support Vector Machine (SVM):

[0119] where αi is the Lagrange multiplier, yi is the class label, and xi is the support vector.

[0120] Deep learning model: Recurrent Neural Network (RNN): h t =σ(W ih x t +b ih +W hh h t-1 +b hh );

[0121] Where ht is the hidden state at time t, xt is the input, and W and b are the weights and biases.

[0122] 3.3 Calculate the risk index based on the model output, formula:

[0123] Risk Index: Risk Index = w1·F1+w2·F2+...+w n ·F n , where Fn is the risk score corresponding to the feature and wn is the weight;

[0124] 3.4 Set the warning threshold of risk index:

[0125] Threshold setting: Threshold = μ Risk Index +k·σ Risk Index ;

[0126] Where μ and σ are the mean and standard deviation of the risk index, respectively, and k is the coefficient.

[0127] This step achieves the following advantages:

[0128] 3.1 Feature Extraction

[0129] Improved accuracy: By calculating statistical features (such as mean, standard deviation) and heart rate variability (HRV), the soldier's physiological state can be more accurately reflected.

[0130] Data dimensionality reduction: Extracting key features from large amounts of raw data reduces the complexity of data processing and improves analysis efficiency.

[0131] Enhanced Insight: Features such as HRV provide insight into the state of a soldier’s autonomic nervous system, helping to assess the impact of training on the body.

[0132] 3.2 Build and apply risk assessment models

[0133] Personalized assessment: By using algorithms such as SVM and RNN, a personalized risk assessment model can be constructed that takes into account the individual differences of soldiers.

[0134] Strong predictive capabilities: Algorithms such as SVM and RNN have strong predictive capabilities and can accurately predict training risks based on feature extraction results.

[0135] Adaptability: Deep learning models are able to adapt to new data and patterns, and their performance will gradually improve over time.

[0136] 3.3 Calculate the risk index based on the model output

[0137] The overall risk is:

[0138] By combining multiple risk characteristics, risk indices provide a comprehensive representation of risk.

[0139] Weight adjustment:

[0140] The weights can be adjusted according to the importance of the features to make the risk assessment more realistic.

[0141] 3.4 Setting the warning threshold of risk index Dynamic threshold: The threshold setting takes into account the statistical characteristics of the risk index, making the warning system more flexible and adaptable.

[0142] Reduce false alarms: Reasonable threshold setting can help reduce false alarms and missed alarms and improve the reliability of the early warning system.

[0143] Timely intervention: By setting early warning thresholds, high-risk situations can be identified in a timely manner so that preventive measures can be taken.

[0144] The step (d) is specifically:

[0145] 4.1 Real-time monitoring of training risk index:

[0146] The average risk index in the preset time period is calculated in real time through the sliding window algorithm. The formula is: Where Risk is the risk index of the i-th time window, and n is the size of the time window;

[0147] 4.2 Evaluate whether the current risk index exceeds the preset safety threshold. If the current risk index exceeds the threshold, trigger an early warning;

[0148] 4.3. Early warning signal generation;

[0149] 4.4 Send early warning notifications through communication channels; including various notification methods:

[0150] Choose the most appropriate notification method (SMS, email, APP push, etc.) according to the situation. Formula: Notification(method,message)Notification(method,message)

[0151] Among them, methodmethod is the notification method, and messagemessage is the warning information.

[0152] 4.5. Record warning events and record warning events for future analysis and audit;

[0153] 4.6 Database record: store detailed information of warning events in the database;

[0154] 4.7 Feedback loop: Collect responses and feedback to warnings to improve the risk assessment model; update the risk assessment model based on the collected feedback data. new =Model old +ΔModel;

[0155] Among them, ΔModel is the model parameter adjusted according to the feedback.

[0156] This step achieves:

[0157] Improve training safety: Through real-time monitoring and early warning, the safety of the training process is improved.

[0158] Enhanced responsiveness: Coaches and medical staff can respond quickly to early warnings and take necessary interventions.

[0159] Improved soldier morale: Soldiers may feel more at ease knowing that there is a system in place to keep them safe.

[0160] Reduce training interruptions: Timely warning and response can help reduce training interruptions and improve training efficiency.

[0161] The present invention also relates to a system for implementing a physical training risk assessment method based on artificial intelligence, comprising:

[0162] Data acquisition module 1, used to collect the soldiers' physiological and motion data in real time during training;

[0163] Data transmission module 2, used to transmit the collected data to the central server;

[0164] Artificial intelligence analysis module 3, used to analyze the collected data and evaluate the training risk; early warning module 4, used to issue a warning signal when the training risk exceeds a preset threshold;

[0165] The result display module 5 is used to display the risk assessment results to the trainers and soldiers.

[0166] The specific working principle of the system of the present invention is as follows:

[0167] Data collection: Use various sensors (such as smart bracelets, heart rate belts, GPS tracking devices, Beidou satellite navigation system, etc.) to collect soldiers' physiological and motion data in real time during training.

[0168] Data transmission: The collected data is transmitted to the central server via wireless network.

[0169] Feature extraction and risk assessment: After the data is preprocessed, risk assessment-related features are extracted through feature extraction algorithms (such as statistical analysis, heart rate variability calculation, etc.).

[0170] The extracted features are analyzed using machine learning models (such as support vector machines (SVMs), recurrent neural networks (RNNs), etc.) to calculate the training risk index.

[0171] Real-time monitoring: Calculate the average risk index within the current time period in real time.

[0172] Threshold comparison: Compare the current risk index with the preset safety threshold. Trigger warning: If the current risk index exceeds the threshold, the warning mechanism is triggered.

[0173] Early warning signal generation: The system generates standardized early warning signals to notify relevant personnel.

[0174] Send warning notifications: Communication channel: Send warning notifications through the most appropriate communication channel (SMS, email, APP push, etc.).

[0175] Logging of warning events: Stores detailed information of warning events in a database for future analysis and auditing.

[0176] Feedback loop: Collect responses and feedback to warnings to improve the risk assessment model. Model update: Update the risk assessment model based on the collected feedback data.

[0177] Advantages of the present invention:

[0178] Real-time: The system can monitor and evaluate training risks in real time and issue early warnings.

[0179] Accuracy: Improve the accuracy of risk assessment through advanced algorithms and models.

[0180] Flexibility: Early warning notifications can be sent through a variety of communication methods to ensure the effectiveness of information transmission.

[0181] Traceability: Records of warning events facilitate future analysis and auditing.

[0182] Adaptability: The system is able to continuously learn and improve based on feedback, improving the effectiveness of the early warning system.

[0183] Through these working principles and steps, the system provides a comprehensive risk management solution for physical training, helping to minimize the risks and injuries that may occur during training.

[0184] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A physical training risk assessment method based on artificial intelligence, characterized in that: The following steps are involved: (a) Real-time collection of soldiers’ physiological and motion data during training; (b) processing the collected data and transmitting it to a central server; (c) Analyze the collected data and assess training risks; (d) Send out warning signals when the training risk exceeds a preset threshold; (e) Present risk assessment results to trainers and Soldiers.

2. The physical training risk assessment method based on artificial intelligence according to claim 1 is characterized in that: The step (a) is specifically: 1.1 Use wearable devices to collect physiological indicators such as heart rate, blood pressure, and body temperature; 1.2 Collect motion trajectory, speed, and acceleration data through GPS tracking devices, Beidou satellite navigation system, or motion sensors; 1.3 Deploy environmental sensors to collect temperature, humidity, and altitude environmental parameters; The heart rate data is collected by using a heart rate belt to collect photoplethysmography (PPG) signals, formula: Where V(t) is the PPG signal amplitude at time t, Vmax is the maximum amplitude, and τ is the time constant; The blood pressure data is collected by measuring the blood pressure using a sphygmomanometer by oscillometric method or auscultatory method, according to the formula: P = ρ·g·g; wherein P is the blood pressure, ρ is the blood density, g is the acceleration of gravity, and h is the height of the blood column; The body temperature data collection uses a temperature patch or an ear thermometer to measure body temperature, the formula is: T = T 环境 +ΔT, where T is body temperature, T 环境 is the ambient temperature, ΔT is the difference between body temperature and ambient temperature; The motion trajectory data is collected by using a GPS tracker or Beidou satellite navigation system to collect location data. Formula: Position(t) = (x(t), y(t)); where (x(t), y(t)) is the coordinate of the geographic location at time t; The velocity and acceleration data are collected by using an accelerometer and a gyroscope: Acceleration formula: , where a is acceleration, Δv is the velocity change, and Δt is the time interval; velocity formula: ; Among them, v is velocity, g is gravitational acceleration, and h is displacement; The data types collected by the environmental data collection include temperature, humidity, altitude and atmospheric pressure; The temperature and humidity data collection uses a temperature and humidity sensor; Temperature formula: T = T 传感器 ; Where T is the ambient temperature, T 传感器 is the temperature measured by the sensor; Altitude data collection uses a barometer to measure altitude, Altitude formula: Among them, h is the altitude, P is the current atmospheric pressure, P0 is the standard atmospheric pressure, and h0 is the standard altitude.

3. The physical training risk assessment method based on artificial intelligence according to claim 1 is characterized in that: The step (b) is specifically: 2.1 Data denoising: Gaussian filtering algorithm is used to denoise the collected multi-source data, specifically: Among them, I′(x,y) is the denoised data value, I(x,y) is the original data value, μ x ,μ y is the center point of the data, σ is the standard deviation of the Gaussian distribution; 2.2 Missing value filling: Missing values ​​in multi-source data are filled by linear interpolation. Among them, x1, x2 are known data points, y1, y2 are the corresponding known data values, and y is the estimated value of the missing data point; 2.3 Outlier Identification: Use Z-score algorithm anomaly detection algorithm to identify outliers.

4. The physical training risk assessment method based on artificial intelligence according to claim 1 is characterized in that: The step (c) is specifically: 3.1 Feature extraction: Extract risk assessment features from preprocessed data. ; where xi is the data point, μ is the mean, and σ is the standard deviation; Heart Rate Variability (HRV): ,in, RR i The interval between adjacent heartbeats; 3.2 Build and apply the risk assessment model to calculate the training risk based on the extracted features. The specific algorithm is: Support Vector Machine (SVM): where αi is the Lagrange multiplier, yi is the class label, and xi is the support vector. Deep Learning Model: Recurrent Neural Network (RNN): h t =σ(W ih x t +b ih +W hh h t-1 +b hh ); Where ht is the hidden state at time t, xt is the input, and W and b are the weights and biases. 3.3 Calculate the risk index based on the model output, formula: Risk Index: Risk Index = w1·F1+w2·F2+...+w n ·F n , where F n is the risk score corresponding to the feature, w n is the weight; 3.4 Set the warning threshold of risk index: Threshold setting: Threshold = μ Risk Index +k·σ Risk Index ; Where μ and σ are the mean and standard deviation of the risk index, respectively, and k is the coefficient.

5. The physical training risk assessment method based on artificial intelligence according to claim 1 is characterized in that: The step (d) is specifically: 4.1 Real-time monitoring of training risk index: The average risk index within a preset time period is calculated in real time through a sliding window algorithm. formula: , where Risk is the risk index of the i-th time window, and n is the size of the time window; 4.2 Evaluate whether the current risk index exceeds the preset safety threshold. If the current risk index exceeds the threshold, trigger an early warning; 4.

3. Early warning signal generation; 4.4 Send early warning notifications through communication channels; 4.

5. Record warning events and record warning events for future analysis and audit; 4.6 Database record: store detailed information of warning events in the database; 4.7 Feedback loop: Collect responses and feedback to warnings to improve the risk assessment model; update the risk assessment model based on the collected feedback data, Model new =Model old +△Model; Among them, ΔModel is the model parameter adjusted according to the feedback.

6. A system for implementing any one of claims 1 to 5 of the method for assessing physical training risk based on artificial intelligence, comprising: Data acquisition module, used to collect soldiers' physiological and motion data in real time during training; A data transmission module, used to transmit the collected data to a central server; AI analysis module, used to analyze collected data and assess training risks; An early warning module is used to issue an early warning signal when the training risk exceeds a preset threshold; The result display module is used to display the risk assessment results to trainers and soldiers.

Images