Exercise training data monitoring method and device based on Internet of Things

Abstract

The invention relates to the field of exercise data monitoring and analysis, in particular to an exercise training data monitoring method and device based on the Internet of Things. The method comprises the following steps: deploying a multi-modal sensor array to collect use data of sports equipment in a public ice and snow place in real time, performing data analysis, and generating equipment dynamic use information of different scenes; using intensity quantification is carried out according to the equipment dynamic using information, and a using intensity time sequence curve is generated; periodic fatigue prediction is carried out based on the use intensity time sequence curve, and state fatigue prediction values of all the devices are generated; according to the multi-modal sensor array, carrying out scene exercise people flow statistics to obtain an exercise behavior evaluation report; and carrying out equipment maintenance and equipment resource configuration decision based on the motion behavior evaluation report and the state fatigue prediction value of each equipment. The service life of equipment in public ice and snow places is prolonged, and the user satisfaction degree and safety are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of sports data monitoring and analysis, and in particular to a sports training data monitoring method and device based on the Internet of Things. BACKGROUND

[0002] With the rapid development of the Internet of Things (IoT) technology, smart devices and sensors are gradually becoming an important means to improve service quality and operational efficiency in various public places. In particular, in ice and snow sports training places, as people's interest in winter sports continues to increase, the number of participants in public ice and snow sports continues to rise, bringing new demands for equipment management, venue operation, and athlete training effectiveness evaluation. In this context, the ice and snow sports training data monitoring method based on the Internet of Things has emerged as the times require, and has become an important tool for improving venue management and training effectiveness.

[0003] Unlike traditional manual recording and analysis, the ice and snow sports training data monitoring method based on the Internet of Things achieves real-time monitoring of various sports equipment in ice and snow venues by deploying a variety of sensors (such as temperature, humidity, pressure, acceleration, etc.) and smart devices, thereby improving the accuracy and timeliness of data collection. These sensors can collect various data during the training process of athletes in real time, including but not limited to the use frequency of sports equipment, movement trajectory, movement intensity, environmental parameters, etc., thereby providing scientific decision support for venue managers, coaches, and athletes themselves. With the widespread deployment of these sensors and devices, traditional monitoring methods face some new challenges. First, due to the wide variety of equipment involved in public ice and snow sports, the complex and variable environment of the sports venue, and the diverse training needs and movement behaviors of participants, a single sensor data is often difficult to fully reflect the equipment status and athlete performance. Second, ice and snow venues often have extreme weather and complex environmental conditions, which pose higher requirements for the stability and data accuracy of sensors. Traditional monitoring methods often lack precise adaptation to different scenarios and real-time warning capabilities for abnormal situations, and cannot effectively respond to various problems that may occur in the venue, such as equipment failure, abnormal athlete behavior, etc. SUMMARY

[0004] To solve the above technical problems, the present application provides a sports training data monitoring method and device based on the Internet of Things to solve at least one of the above technical problems.

[0005] To achieve the above purpose, the present application provides a sports training data monitoring method based on the Internet of Things, comprising the following steps: Step S1: deploying a multi-modal sensor array to collect sports equipment usage data in public ice and snow venues in real time, performing data analysis, and generating dynamic equipment usage information for different scenarios; Step S2: quantifying the use intensity according to the device dynamic use information, generating a use intensity time curve; Step S3: predicting periodic fatigue based on the use intensity time curve, generating a device state fatigue prediction value; Step S4: scene motion crowd statistics according to the multi-modal sensor array, obtaining a motion behavior evaluation report; Step S5: making device maintenance and device resource configuration decisions based on the motion behavior evaluation report and the device state fatigue prediction value.

[0006] In this specification, a kind of based on Internet of Things's motion training data monitoring device for executing based on Internet of Things's motion training data monitoring method as described above, including: Device use analysis module, for deploying multi-modal sensor array to collect the motion equipment use data of public ice and snow place in real time, data analysis is carried out, and the device dynamic use information of different scenes is generated; Intensity quantification module, for quantifying the use intensity according to the device dynamic use information, generating a use intensity time curve; Fatigue prediction module, for predicting periodic fatigue based on the use intensity time curve, generating a device state fatigue prediction value; Crowd statistics module, for carrying out scene motion crowd statistics according to the multi-modal sensor array, obtaining a motion behavior evaluation report; Resource configuration module, for making device maintenance and device resource configuration decisions based on the motion behavior evaluation report and the device state fatigue prediction value.

[0007] The beneficial effects of the present application are as follows: Through the multi-modal sensor array (such as temperature sensors, pressure sensors, acceleration sensors, etc.), the use of various types of ice and snow sports equipment can be monitored in real time. These sensors can capture key information such as equipment state changes, exercise intensity, and usage frequency, ensuring comprehensive and accurate data collection. Different sensors can monitor different types of equipment usage scenarios. For example, the performance of skis, sleds, and ice vehicles in different environments can help analyze the usage frequency and intensity of different equipment, providing specific dynamic usage information. After obtaining this data, reports and scenario information on equipment dynamic usage can be generated, providing data support for subsequent maintenance and optimization and reducing errors caused by manual estimation. Through quantitative analysis of equipment dynamic usage data, the usage intensity of equipment at each moment can be accurately determined. For example, the pressure changes of skis in different sliding time periods can directly reflect the frequency of equipment use. By converting usage intensity data into a time series curve, the usage of equipment in different time periods and the change trend can be clearly presented. Visualization of this time series data helps to understand the usage patterns of equipment and predict equipment fatigue. The usage intensity time series curve provides the necessary data basis for periodic fatigue prediction. This step provides quantitative indicators for subsequent equipment maintenance decisions. Based on the usage intensity time series curve, the accumulation of fatigue in equipment during long-term use can be analyzed. Through comprehensive analysis of equipment usage frequency and intensity, the fatigue condition of equipment can be predicted to avoid equipment damage before reaching the critical fatigue state. By predicting periodic fatigue in advance, potential equipment failures can be warned, effectively avoiding equipment sudden failures and improving equipment service life. Reasonable inspection and maintenance cycles for equipment can be established to ensure that equipment operates in the best state. Avoiding excessive fatigue of equipment can affect the normal operation of the site. Through comprehensive data collection by the sensor array, the number of people exercising in different areas, activity types, and their trends can be counted. This data can provide exercise behavior analysis reports for venue managers to help analyze the efficiency and demand distribution of the venue. By analyzing the flow of people and the use of scenarios, the resource allocation of the venue can be optimized to improve the utilization rate of the venue facilities. For example, if the equipment in some areas is too dense or has a high usage frequency, the venue manager can make adjustments or allocate more resources. Through the evaluation of exercise behavior, data support can be provided for the arrangement and guidance of venue activities. For example, when the demand for certain sports is high, the time can be reasonably arranged or the venue resources can be adjusted to improve user experience. Combining equipment state fatigue prediction and exercise behavior evaluation reports can intelligently determine when to maintain or replace equipment. This not only avoids equipment failure due to overloading, but also reduces unnecessary maintenance costs. According to the flow of people and equipment usage intensity data in different areas, equipment resource allocation can be performed.For example, some devices are frequently used in areas that can be arranged more devices or improve the frequency of device maintenance, while less used areas can reduce device investment and save resources. Through real-time data monitoring and analysis, the venue management decision-making can be more forward-looking and scientific, and the overall operation level and service quality of public places can be improved. Reduce equipment idle or overuse, improve overall efficiency and user satisfaction. BRIEF DESCRIPTION OF DRAWINGS

[0008] Fig. 1 A schematic diagram of the step flow of the method of the present application is shown in Figure 1. Fig. 2 A schematic diagram of the detailed implementation step flow of step S1 is shown in Figure 2. Fig. 3 A schematic diagram of the detailed implementation step flow of step S2 is shown in Figure 3. DETAILED DESCRIPTION

[0009] It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application.

[0010] The present application provides a kind of based on Internet of Things's sports training data monitoring method and device. The execution subject of the based on Internet of Things's sports training data monitoring method and device includes but is not limited to the system carried: mechanical equipment, data processing platform, cloud server node, network upload equipment etc. It can be regarded as the general computing node of the present application, and the data processing platform includes but is not limited to: audio image management system, information management system, cloud data management system at least one kind.

[0011] Please refer to Figs. 1 to 3 The present application provides a kind of based on Internet of Things's sports training data monitoring method, including the following steps: Step S1: deploy multi-modal sensor array to collect the sports equipment use data of public ice and snow place in real time, carry out data analysis, generate equipment dynamic use information of different scenes; Step S2: according to the equipment dynamic use information, use intensity quantization is generated, and use intensity time series curve is generated; Step S3: based on use intensity time series curve, periodic fatigue prediction is generated, and each equipment state fatigue prediction value is generated; Step S4: according to the multi-modal sensor array, scene sports flow statistics are carried out, and sports behavior evaluation report is obtained; Step S5: based on sports behavior evaluation report and each equipment state fatigue prediction value, equipment maintenance and equipment resource configuration decision are made.

[0012] In the embodiment of the present application, please refer to Fig. 1A flowchart of the steps of the motion training data monitoring method based on the Internet of Things is shown in the present example. The steps of the motion training data monitoring method based on the Internet of Things include: Step S1: Deploy a multi-modal sensor array to collect motion equipment usage data in public ice and snow venues in real time, analyze the data, and generate equipment dynamic usage information in different scenarios; In this embodiment, a multi-modal sensor array is deployed to collect the usage state of the motion equipment in all directions in real time. The sensor array is composed of pressure sensors, accelerometers, gyroscopes, temperature and humidity sensors, infrared thermal imaging modules, and high-resolution video acquisition units. Various sensors are arranged according to the functional areas of the venue, such as the sliding channel, the curve training area, and the equipment storage area. The pressure sensors are embedded in the lower layer of the ice surface to monitor the stress distribution; the accelerometers and gyroscopes are fixed on the equipment to record the motion direction, speed, and attitude changes; the infrared thermal imager is deployed at a high place to monitor the heat distribution and usage density changes. The sensor nodes use 5G or Wi-Fi 6 wireless communication and use a unified time synchronization protocol (such as NTP) to ensure sampling consistency. The data acquisition frequency is set to 100Hz to 500Hz to balance real-time performance and accuracy. After noise filtering and time synchronization, the raw data is analyzed by multi-modal fusion to identify the equipment stress state, temperature change rate, and vibration characteristics. By calculating indicators such as pressure distribution entropy, temperature gradient, and vibration spectrum concentration, equipment dynamic usage information in different scenarios can be generated, laying a data foundation for subsequent intensity quantification and fatigue prediction.

[0013] Step S2: Quantify the usage intensity based on the equipment dynamic usage information to generate a usage intensity time series curve; In this embodiment, the key parameters such as equipment pressure distribution, vibration amplitude, and temperature change are normalized to ensure comparability between different types of equipment. Then, the time slice method is used to segment the data for statistical analysis, dividing the continuous time stream into fixed time windows (such as 5 minutes or 10 minutes) to calculate the average load value and fluctuation range of the equipment in each time window. According to the usage environment and scene characteristics, the usage intensity of different regions is weighted and corrected, for example, the high-speed sliding area has a higher weighting coefficient, and the rest area has a lower weighting coefficient. Through cumulative analysis, the usage intensity trend of the equipment in a certain period of time can be obtained. The vertical axis of the usage intensity time series curve represents the equipment usage load level, and the horizontal axis represents the time sequence. This curve reflects the high and low peak cycles and long-term wear trends of the equipment, which can be used to identify high-intensity running phases and equipment overload risks. Through continuous monitoring, periodic patterns of equipment usage can be formed, providing quantitative basis for fatigue modeling and maintenance strategies.

[0014] Step S3: generating fatigue prediction values of each device state based on periodic fatigue prediction using the intensity time series curve; In this embodiment, the intensity curve is periodically decomposed to extract the short-term and long-term fluctuation rules for judging the stability of device operation and the periodic load characteristics. Then the intensity curve is fused with the physical state parameters of the device (such as vibration concentration, pressure balance, temperature fluctuation rate) to construct a multi-dimensional feature vector. Based on the time series modeling method, the fatigue rate growth can be predicted by using the cumulative change trend of the sliding time window. For different types of devices, different fatigue threshold intervals can be set according to the material characteristics and use mode, for example, ice skate type devices focus on structural stress analysis, and ski equipment focuses on thermal strain fluctuation characteristics. Through continuous analysis, the fatigue prediction value of the device can be generated, which reflects the health degree and remaining use potential of the device. When the prediction value approaches the set warning line, it means that the device is about to enter a high fatigue stage, and maintenance or maintenance should be carried out in advance. The final output fatigue prediction result is saved in the form of time series, providing quantitative basis for subsequent maintenance and resource allocation decisions.

[0015] Step S4: scene motion crowd statistics based on the multi-modal sensor array, and obtaining a motion behavior evaluation report; In this embodiment, the same sensor array is used to realize the statistics and behavior recognition of the crowd in the venue. Through the joint collection of visual cameras, millimeter wave radars and infrared thermographs, the number of crowds and activity trajectories can be recognized under different light and climate conditions. After the video stream is processed by target detection and tracking algorithms, individual position and speed information is obtained; millimeter wave radar supplements the detection results under low visibility conditions; infrared sensors are used to identify hot signal aggregation areas to calculate the crowd density distribution. After spatial fusion of the above multi-source data, real-time crowd distribution and motion trajectory features can be obtained. Further analysis of the average stay time, activity density and device interaction frequency of the crowd can generate a motion behavior feature set. Combined with historical data, the crowd change law in different time periods and different regions can be identified, such as peak hours, dense areas and high-risk areas. The final motion behavior evaluation report includes crowd distribution map, activity trend curve and behavior type statistics, providing human factor analysis support for device maintenance and resource allocation.

[0016] Step S5: device maintenance and device resource allocation decision based on the motion behavior evaluation report and the fatigue prediction values of each device state.

[0017] In this embodiment, after obtaining the equipment fatigue prediction result and the motion behavior evaluation report, comprehensive decision is made for equipment maintenance and resource allocation. First, the equipment health level is identified according to the fatigue prediction value, and the state is divided into three categories: normal, early warning and high fatigue. High fatigue equipment is included in the maintenance plan, and early warning equipment is arranged for periodic detection. Then, combined with the people flow distribution and the site use heat in the motion behavior evaluation report, the space configuration of the equipment is matched and optimized. In the area with high people flow density and high use frequency, the equipment in good condition is preferentially allocated, and standby resources are arranged to prevent sudden failure; in the area with low people flow, the number of equipment can be moderately reduced to reduce energy consumption and maintenance load. The resource allocation decision adopts a dynamic adjustment mechanism, which is based on real-time data, and is updated periodically according to the people flow trend and the remaining value of equipment life. The maintenance strategy and the configuration strategy form a closed-loop decision logic, realizing the whole process management from equipment state perception, behavior evaluation to resource optimization. Through this process, the use efficiency of equipment can be maximized, the safety of site operation can be improved, and the maintenance cost can be controlled, providing technical support for the intelligent operation of public ice and snow sports places.

[0018] In this embodiment, reference is made to Fig. 2 For the detailed implementation step flowchart of step S1, in this embodiment, the detailed implementation steps of step S1 include: Deploying a multi-modal sensor array to collect real-time motion equipment usage data of a public ice and snow place; Identifying sensor abnormal jitter and filtering abnormal noise for the motion equipment usage data to obtain standardized equipment usage samples; Identifying equipment position based on the multi-modal sensor array and classifying usage scene types to obtain multiple usage scenes; According to the multiple usage scenes, the scene type of the standardized equipment usage samples is labeled to generate scene label samples; Based on the scene label samples, the pressure distribution entropy of the equipment, the temperature gradient change rate, the frequency concentration degree of the vibration spectrum and the personnel gathering heat density are calculated to generate equipment dynamic usage information in different scenes.

[0019] In this embodiment, in the ice and snow sports public place, multi-modal sensors are deployed on each sports equipment to realize the construction of sensor array distribution in the public place, so as to realize the real-time monitoring of the use state of the sports equipment. The array is usually composed of pressure sensors, acceleration sensors, temperature and humidity sensors, infrared thermal imaging sensors, and high frame rate visual cameras. Each sensor is arranged according to the functional area, for example, MEMS pressure sensors can be embedded under the sliding path in the ice training area to measure the contact force distribution of the equipment (such as ice skate, snowboard) and the ice surface; temperature and humidity and infrared sensors are installed in the ski equipment storage area to judge the preheating and use state change of the equipment; and high frame rate cameras (≥120 fps) and laser radar modules are deployed above and on both sides of the sports channel to realize the equipment trajectory and speed monitoring. The data acquisition frequency is suggested to be controlled in the interval of 100 Hz-500 Hz to balance the real-time performance and the computing load. All sensor nodes access the edge computing gateway through 5G or Wi-Fi 6 network and use MQTT protocol for data transmission. In order to ensure the synchronization of multi-modal data, the system uses NTP time synchronization mechanism and caches 0.5 seconds of time window data locally to realize time alignment, thereby constructing continuous and reconfigurable equipment motion data stream. After multi-modal data acquisition, due to the complex environment of outdoor low temperature and ice surface reflection, the sensor output often contains random jitter and noise. In order to ensure data quality, first, the signals of each channel need to be identified and filtered. The jitter identification uses an improved Z-score algorithm combined with a sliding time window. When the standard deviation of a certain sensor signal in the adjacent 100 sampling points exceeds 3 times the historical mean value, it is determined as jitter anomaly. At the same time, Kalman filter and wavelet threshold denoising technology are used for signal smoothing. The temperature and pressure channels use a first-order Kalman filter to suppress slow noise, and the acceleration and vibration signals use Daubechies wavelet (db6) for multi-scale decomposition to remove high-frequency interference. For optical image data, a frame jitter correction algorithm based on optical flow dense detection is used to ensure the consistency of visual data and inertial signals. After the above processing, all signals are standardized and normalized (Z-score standardization) to eliminate the influence of different sensor dimensions. The final standardized sample data contains time sequence index, equipment ID, three-axis acceleration, force vector, temperature, humidity and spatial coordinates fields, and the sampling frequency is unified to 200 Hz, providing high-quality input for subsequent scene analysis.

[0020] After data quality assurance, the system enters the spatial position recognition and scene classification phase. Position recognition is achieved by fusing inertial measurement unit (IMU) and visual SLAM algorithms. IMU provides instantaneous acceleration and angular velocity information of the device, and the visual sensor (camera + laser radar) provides environmental feature point coordinates. Extended Kalman filter (EKF) is used for multi-sensor fusion to reduce drift error, and finally the device positioning error can be controlled within 10 cm. Scene classification is achieved based on clustering and machine learning algorithms: by extracting motion trajectory features (rate of change of speed, turning radius, force mode), combined with temperature and heat distribution features, the K-Means++ clustering algorithm is used to unsupervised clustering of different regions and motion patterns, to obtain primary scene categories such as "high-speed sliding area", "bend training area", "rest area", etc. Then introduce convolutional neural network (CNN) to perform scene semantic recognition on image data, and use random forest model for multi-feature fusion classification to generate the final multi-level scene label system. The classification accuracy can reach more than 93% in the experiment, thereby realizing the automatic recognition of the device use scene. After completing the scene recognition, the system automatically maps the recognized scene label to the standardized data sample to form a training sample set with scene annotation. This process is completed through a space-time matching algorithm: first, align according to device ID and timestamp to ensure data continuity; then use the spatial relationship between position coordinates (x, y, z) and scene area boundaries to determine the scene to which the device belongs at that time. If the spatial position of a sample is located in the intersection area of multiple scene boundaries, the nearest neighbor probability matching method (Nearest Scene Matching, NSM) is used to determine the most likely scene according to the device motion trend and heat density priority. In order to prevent mislabeling, the system will perform secondary screening when generating scene label samples, and remove sample points with trajectory drift exceeding the set threshold (>0.5 m). The final "scene label sample set" contains fields: timestamp, device number, motion feature vector, sensor fusion signal, scene type ID. Through sample balancing and SMOTE oversampling method to adjust the proportion of different scene samples, ensure the balance of training data, provide a high reliable data basis for subsequent dynamic feature analysis.

[0021] Physical feature calculation and dynamic usage information generation on labeled data. First, the pressure distribution entropy (PDE) of the equipment is calculated. After normalizing the pressure distribution matrix, the information entropy value is calculated to reflect the uneven degree of force on the equipment in different areas. Second, the temperature gradient change rate is calculated by calculating the partial derivative of the temperature spatial field within the time window Δt = 5s to evaluate the stability of the equipment heat conduction. The spectral concentration coefficient (SCC) is based on the FFT frequency spectrum energy distribution to calculate the proportion of main frequency band energy to determine whether the equipment vibration state is concentrated in a specific frequency band (for example, 15-30 Hz is the normal motion interval). The personnel gathering heat density is generated by infrared thermal imaging data to generate a heat map, and then the Gaussian kernel density estimation (KDE) is used to calculate the distribution of human heat signal intensity per unit area to reflect the dynamic changes of people flow density in different scenes. Finally, the system integrates and normalizes the four indicators to generate a "dynamic usage information matrix of equipment", which contains the running stability of the equipment, the environmental thermal characteristics, the load intensity and the human flow interaction characteristics in each scene, providing accurate data support for equipment scheduling, safety warning and energy consumption optimization in ice and snow sports venues.

[0022] In this embodiment, referring to Fig. 3 For the detailed implementation step flowchart of step S2, in this embodiment, the detailed implementation steps of step S2 include: Extracting the data collection timestamp of the sports equipment usage data; Performing inter-sensor time sequence synchronization processing on the data collection timestamp to obtain a synchronized timestamp; Performing day and night and season recognition based on the synchronized timestamp to generate periodic data; Calculating the equipment usage frequency and usage duration of the dynamic usage information of the equipment; According to the periodic data, the equipment usage frequency and usage duration are analyzed to obtain the equipment usage mode in different periods; Quantifying the usage intensity of the equipment usage mode and performing intensity distribution statistics in multiple time windows to generate a usage intensity time sequence curve.

[0023] In this embodiment, in the ice and snow sports training monitoring system, the data generated by all sensor nodes carries timestamp information, which is written by the data acquisition module in the edge computing node, for identifying the sampling time and data source identity. To realize subsequent multi-modal fusion processing, the system first extracts and organizes the timestamp of the original data stream. This step is performed in the edge node, by analyzing the metadata of each data packet, extracting information including sensor ID, data type, sampling time (Unix time format) and sampling frequency, etc. Since different sensors (such as pressure, temperature, acceleration, infrared) have different sampling frequencies, there are also millisecond-level deviations in the collection interval, so a unified time index table needs to be established when extracting. For example, in the experimental field, the sampling frequency of pressure and acceleration sensors is set to 200 Hz, the sampling frequency of temperature sensors is 10 Hz, and the sampling frequency of infrared imaging data is 30 fps. The system uses time window division method in the extraction process, maps all timestamps to a unified time axis (Δt = 0.01 s), and generates a serial number index for each data packet, so as to facilitate subsequent interpolation and synchronization. The final output timestamp sequence file is saved in a structured form (such as JSON or Parquet), providing a standardized time basis for subsequent time synchronization and cycle identification. Due to the existence of multi-sensor asynchronous sampling and network transmission delay, there is a microsecond to millisecond level drift between the original timestamps, so time synchronization must be performed. The system uses a dual method of hierarchical clock calibration and multi-modal time synchronization. First, at the sensor node level, the network time protocol (NTP) or precise time protocol (PTP, accuracy up to 1 μs) is used for network time unification. Secondly, at the data fusion level, the Time Drift Estimation Model (TDEM) is used to dynamically correct the sampling offset between different sensors. This model uses a sliding window (window length 1 second) to calculate the correlation coefficient of multi-sensor signals, when the correlation coefficient reaches the maximum value at time delay τ, τ is the synchronization offset. After the system estimates the offset of all sensors in pairs, the offset is used for time resampling, so that all signals are mapped to the reference time axis. Experiments show that through this method, the time error can be reduced from the initial ±15 ms to within ±1 ms. The synchronized timestamp sequence is the "synchronized timestamp", which lays the foundation for the time consistency of subsequent day-night cycle identification and device usage frequency analysis.

[0024] After time synchronization, the system needs to identify the diurnal and seasonal cycles of data collection based on timestamp information, so as to reveal the time regularity of device use. Diurnal identification is completed by timestamp analysis and environmental light intensity data fusion. The system first analyzes the hour field in the timestamp (UTC+8), defines 00:00-06:00 as night, 06:00-18:00 as day, and 18:00-24:00 as transition period. At the same time, light sensors or infrared brightness data are introduced to assist in judgment: if the light intensity value is lower than the threshold L<30 lx, it is automatically determined as night. Seasonal identification is based on the date information of the timestamp and the average statistical values of environmental temperature and humidity. In the experiment, months with average temperature below 0°C and humidity greater than 70% (usually November to March) are defined as winter period, and the rest are summer or transition season. The system combines diurnal and seasonal information to generate cycle labels (such as "winter-night" and "summer-day") and writes them into the data index table in the form of cycle encoding (CycleID). The generated cycle data will be used as an important grouping basis for subsequent device use frequency and duration statistics, supporting cross-season and cross-time period motion pattern research. After completing time and cycle identification, the system performs statistical analysis on the device's use state based on synchronized time series, and calculates the use frequency and duration. The use state is determined by key indicators in device dynamic use information: if the pressure distribution entropy (PDE) is greater than the set threshold 0.15, the vibration spectrum frequency concentration (SCC) is in the 15-30 Hz interval, and the thermal density rising rate is >5% / s, it is considered as "use state"; otherwise, it is "idle state". The system counts the number of "use state" occurrences in each time period as the use frequency (times / hour), and the total duration of the state as the use duration. To avoid short-time misjudgment, the system introduces a minimum use threshold Δt_min=3 s, i.e. signals with continuous activation time less than 3 seconds are not counted as valid use. Experimental data show that this determination logic can effectively exclude false triggers and sensor noise. By using a time sliding window (length 30 minutes, overlap rate 50%), the use state is counted to generate high-resolution frequency and duration curves. These data reflect the use activity of different devices in different time periods, providing a quantitative basis for subsequent pattern analysis. The system combines device use frequency and duration with cycle data to analyze periodic use patterns. First, the data is grouped according to the diurnal and seasonal labels, and then the average use frequency μ_f and average use duration μ_t of each cycle are calculated. By analyzing the time series spectrum through Fourier transform (FFT), the periodic characteristics of device use can be identified, such as 24-hour cycle (diurnal cycle) or 168-hour cycle (weekly cycle).Further, the K-means clustering algorithm is used to cluster the usage feature vector {μ_f, μ_t, PDE mean, SCC mean} of each period, and automatically divide multiple typical usage modes, such as “high-frequency short-time training type”, “low-frequency long-time maintenance type”, “night high-density usage type”, etc. In the experimental scene, different devices (snowboard, ice skate, ice surface trimming machine) show obvious differences in periodic use. The system finally outputs a “periodic usage mode table” containing the period label, usage mode category and mode feature parameters, which is used to reveal the device usage rules in different seasons and time periods.

[0025] The device usage mode in different periods is intensively quantified and time series statistical analysis is performed. The usage intensity (Usage Intensity, UI) is defined as a multi-factor weighted index: ; Where α, β, γ, δ are weight coefficients (0.3, 0.3, 0.2, 0.2 in the experiment), f_norm and t_norm are the normalized usage frequency and duration respectively. The system calculates the UI mean and standard deviation in multiple time windows (1 hour, 3 hours, 6 hours), and generates the intensity distribution curve through sliding window statistics. To enhance interpretability, a smoothing filter (Savitzky-Golay filter, window length 9) is used to denoise the curve and obtain a smooth usage intensity time series change graph. The results show that the usage intensity significantly increases during the daytime peak period and decreases at night, while the overall intensity in winter increases by about 25% compared to summer. The generated “usage intensity time series curve” intuitively reflects the dynamic usage state of the device in different periods, providing a quantitative basis for site operation and personnel scheduling.

[0026] In this embodiment, step S3 includes the following steps: Detecting device physical state indicators based on a multi-modal sensor array; Performing dynamic change analysis based on the device physical state indicators to generate state change data; Constructing a device fatigue assessment model according to the state change data; Inputting the usage intensity time series curve into each device fatigue assessment model to perform device state health score and generate real-time device fatigue score value; Performing curve evolution trend analysis on the real-time device fatigue score value to extract the device fatigue evolution trend; Performing fatigue rate calculation based on the device fatigue evolution trend to generate fatigue rate change law; Performing periodic fatigue prediction based on the fatigue rate change law to generate device state fatigue prediction value.

[0027] In this embodiment, real-time detection of device physical state indicators is achieved through a multi-modal sensor array, including a three-axis accelerometer, a strain gauge pressure sensor, a temperature and humidity sensor, an infrared thermal imager, and a vibration monitoring module. The acceleration sensor is installed on the key structural nodes of the device (such as the center axis of the ski board, the support beam of the ice skate), with a sampling frequency of 500 Hz to monitor the vibration and impact response. The strain sensor has a sampling frequency of 200 Hz, which is used to measure the force change of the device under different loads. The infrared thermal imaging module captures the temperature distribution field of the device surface at a frame rate of 30 fps, and the temperature and humidity sensor monitors the influence of environmental conditions on the device. All sensor nodes are connected to the edge computing node through LoRa or 5G communication link, and the unified timestamp synchronization mechanism (error <1 ms) is used to ensure the time sequence alignment of multi-channel data. The system extracts physical state indicators from these sensor outputs, including device overall force mean, temperature distribution standard deviation, vibration spectrum peak frequency, deformation rate change Δε, and thermal gradient change rate ΔT / Δt, etc. These indicators together constitute the physical state vector S(t) of the device, providing basic physical data support for subsequent dynamic change analysis and fatigue modeling. A sliding time window (Δt = 10 s, overlap rate 50%) is used to calculate the time change rate and standard deviation of each indicator, and a dynamic change feature vector D(t) = {Δε / Δt, ΔT / Δt, ΔA / Δt, ΔP / Δt} is constructed. Where ΔA is the acceleration change amplitude, and ΔP is the pressure distribution change. To eliminate short-term noise, the system introduces a Savitzky-Golay filter (window length = 9) for smoothing, and uses principal component analysis (PCA) to extract the main change pattern. Secondly, through time series difference analysis (Time Differencing Analysis), the device state mutation points are identified, such as vibration peak transition, strain overrun or thermal stress anomaly. Experimental data shows that in the ski board use scenario, the vibration spectrum energy is concentrated in the 20-35 Hz range under normal conditions, and when the device fatigue tends to intensify, the energy in this interval decreases by about 18%, and the low-frequency component significantly increases. The system records such change trends as "state change events" and stores them in a structured form as a state change dataset. This data contains timestamp, change amplitude, trend direction (enhancement or weakening), and characteristic energy distribution, which becomes the dynamic input basis for fatigue modeling.

[0028] On the basis of dynamic state change data, the system constructs fatigue assessment models for different types of equipment (such as snowboards, ice skate cars, and ice surface finishing machines). The model design follows the principle of multi-physical quantity coupling and multi-time scale fusion. First, the key fatigue characteristic parameters are determined by combining statistical regression and physical modeling. Taking strain energy density (SED) as the core index, the characteristic selection is carried out using long-term monitoring data, and the variables that have a significant impact on fatigue (such as temperature gradient change rate, vibration concentration SCC, and pressure distribution entropy PDE) are selected through correlation analysis. The model training uses a time series prediction structure based on LSTM (Long Short-Term Memory Network) to capture the time evolution law of equipment fatigue. The training samples are taken from 30 consecutive days of equipment operation data, with about 5x10 6 records sampled per day, 200 rounds of training, and a learning rate of 0.001. The model output is the fatigue health index (HI), ranging from 0 to 1, with a value closer to 1 indicating a healthier device. The model achieved a fatigue trend prediction accuracy of 92.7% in experimental verification, providing a high-precision evaluation basis for real-time state scoring.

[0029] The UI curve generated in the previous step is used as one of the input features, along with the physical state vector S(t), to input into the model for real-time fatigue scoring calculation. The system runs model inference in the edge computing node, updating the score result every 5 minutes. The model input feature dimensions include: usage frequency f_norm, usage duration t_norm, temperature gradient ΔT / Δt, strain rate Δε / Δt, and UI average value. The model output is the device fatigue score F(t), with a numerical range of 0-100, and a threshold set as: F<40 is determined as "high fatigue risk", 40≤F<70 is "moderate fatigue", and F≥70 is "healthy state". To verify the real-time performance, the experimental system uses NVIDIA Jetson Orin NX edge node for inference, with a single calculation delay controlled within 0.25 s. The generated real-time fatigue score is presented in curve form through the visualization platform, and is superimposed and compared with historical data to achieve continuous health monitoring of the device operating state. The score curve is decomposed into long-term trend, periodic fluctuation and random disturbance, eliminating the interference of incidental noise. Then the change slope and acceleration of the score curve are calculated to judge the dynamic stage of fatigue aggravation or recovery. When the score continues to decline and remains for a certain time span, the system automatically marks it as the fatigue acceleration stage; if the score is stable or slightly rebounds, it is determined as the recovery stage. To ensure the reliability of trend determination, the system uses non-parametric statistical test method to verify the significance of score change. In the experimental scenario, the ski equipment shows a significant downward trend in the score curve after four hours of high-intensity continuous use, while it shows a slow rebound during the rest and maintenance period. The system organizes such trend information into fatigue evolution trend data, including change direction, stability analysis results and possible turning point position, to support subsequent fatigue rate and prediction model calculation. Further calculation of fatigue rate and analysis of its change law. Fatigue rate represents the decline amplitude of health score per unit time, which can be used to measure the speed of device performance degradation. The system performs rate statistics on one hour, three hours and twelve hours time scales, and draws the trend curve of fatigue rate change over time through polynomial fitting and smoothing algorithm. According to the rate size, the system divides the device operating state into three categories: stable zone, acceleration zone and decline zone. Experimental results show that the snowboard enters the accelerated fatigue stage after twelve hours of continuous high-frequency use, with a fatigue rate increase of about 30%. After smoothing the rate curve, the system can clearly see the fatigue growth characteristics in different stages. The final generated fatigue rate change law reflects the fatigue evolution pattern of the device under different usage intensity, environmental temperature and periodic conditions, providing a scientific basis for device maintenance planning and life management. A periodic fatigue prediction model is established to calculate the future health status of the device. The prediction model is based on time series learning method, taking historical fatigue score, fatigue rate change curve and usage intensity periodic features as the main input.The model automatically identifies diurnal, weekly, and seasonal usage patterns and predicts the fatigue trend of the equipment in the next few hours or days. At least one month of historical data is used during the training process to optimize the model parameters through multiple iterations, ensuring the stability and accuracy of the prediction results. Experiments show that in typical scenarios at ski resorts, the prediction model can provide a 24-hour advance warning of potential fatigue overload risks, with an average prediction error of less than 5%. The system displays the prediction results in the form of a curve, with possible fatigue overload periods and equipment recovery cycles marked. The final fatigue prediction value provides a basis for early intervention for management personnel, making equipment maintenance more scientific and efficient, and effectively reducing the rate of sudden failures.

[0030] In this embodiment, step S4 includes the following steps: According to the multi-modal sensor array, the scene motion crowd flow statistics are performed to obtain motion crowd monitoring data; According to the motion crowd monitoring data, user average stay time, activity trajectory heat distribution, equipment and personnel interaction frequency, and high-risk operation frequency are calculated to generate crowd behavior feature data; Based on the crowd behavior feature data, user identity type clustering is performed to identify motion user types; Based on the motion user types, crowd flow distribution calculation is performed to generate multi-type crowd flow distribution maps; Based on the crowd behavior feature data, crowd behavior evaluation is performed on the multi-type crowd flow distribution maps to generate a motion behavior evaluation report.

[0031] In this embodiment, a multi-modal sensor array is used for crowd counting and motion behavior recognition. The array includes high-resolution visual cameras, millimeter-wave radar sensors, infrared thermal imagers, and environmental acoustic sensors. The visual cameras capture video images at a rate of 60 frames per second and perform pedestrian detection algorithms through edge computing nodes to identify and count moving targets in real time. Millimeter-wave radar is used to capture human motion trajectories in low-light or snow and fog environments, supplementing the blind area information of visual sensors; the infrared thermal imaging module can detect human thermal signals at night or under temperature difference conditions to distinguish between real people and equipment movements. To prevent duplicate counting, the system uses a multi-sensor data fusion method to match and compare the identification results of the same target, with a matching accuracy of more than 98%. The data of all sensor nodes is synchronized with a unified timestamp and uploaded to the edge server through a 5G communication protocol. Finally, the system generates "motion crowd monitoring data" containing time, spatial location, individual trajectory number, and instantaneous speed, providing the original basis for subsequent behavior analysis. The activity behavior of users in the ice and snow place is quantitatively analyzed. First, the system calculates the individual's stay duration in a specific area according to the start and end time of the personnel trajectory, and then calculates the average value of all samples to obtain the average stay duration index. Second, the trajectory heat distribution is calculated by dividing the site into fixed grids (for example, each 10 square meters is a cell), calculating the cumulative time and frequency of personnel appearing in each grid, and generating a two-dimensional heat map to reflect the intensity of the crowd. The system also compares equipment usage data and personnel trajectory data to count the number of interactions and duration of equipment and personnel, and thus obtains the interaction frequency index. For high-risk operation behaviors, such as high-speed sliding, out-of-bound sliding, and simultaneous occupation of equipment, the system combines acceleration threshold, trajectory curvature, and collision detection algorithms for recognition and statistics. After all feature data is time-aligned and normalized, a "crowd behavior feature data set" is formed, which includes average stay duration, trajectory density value, interaction intensity index, and risk frequency, providing support for crowd clustering and behavior evaluation.

[0032] The feature data is standardized and dimensionally reduced to eliminate dimensional differences and improve clustering efficiency. Then, an unsupervised learning algorithm is used for clustering analysis. In the experiment, an improved K-means algorithm combined with a density clustering method (DBSCAN) is used, and the optimal number of clusters is determined by the silhouette coefficient. Generally, it can be divided into four to six types of user types. Typical clustering results include: short-time experience users (short stay time, limited range of trajectory), medium training users (frequent interaction, regional concentration), professional competitive users (high-speed trajectory, strong interaction, high-frequency use of equipment), and recreational users (wide range of movement but less use of equipment). Clustering is completed on a GPU server, and the time consumption of a single calculation is less than two minutes. After clustering, the system labels each user sample as the corresponding type and records it in the user behavior profile, providing semantic user category information for subsequent crowd distribution and behavior evaluation. The entire ice and snow field is divided into several functional areas, such as the sliding area, training area, rest area, and equipment debugging area. The system calculates the proportion of the number of users, activity density, and time period distribution of different types of users in each area, and generates dynamic crowd distribution data at the hourly and daily levels combined with timestamps. Then, the data is continuously processed by a spatial interpolation algorithm to form a smooth crowd density distribution map. The visualization module uses heat rendering to distinguish between high and low crowd density, and different user types are displayed with different identifiers, such as red for high-intensity sports users and blue for recreational users. Experimental data shows that the number of short-time experience users significantly increases during weekends and at night, while the proportion of professional training users is higher during weekdays. The system finally generates a "multi-type crowd distribution map" to dynamically monitor different user groups in different time and spatial dimensions, providing quantitative basis for site scheduling, safety management, and service optimization. After completing the multi-type crowd distribution calculation, the overall crowd behavior is comprehensively evaluated to form a sports behavior evaluation report. The evaluation process combines crowd feature indicators, crowd distribution characteristics, and equipment interaction to analyze from three dimensions of safety, efficiency, and site carrying capacity. First, the system calculates the average stay time and equipment interaction intensity in high-density areas to determine whether there is a risk of congestion or overload use; if the stay time in high-density areas exceeds twice the average value, the system automatically marks it as a potential safety hazard area. Second, by analyzing the activity distribution of different user types, the system evaluates whether the site usage structure is reasonable, such as whether the proportion of professional users in the training area meets the expected value. The system also compares historical data to identify periodic behavior patterns, such as the surge in crowd during holidays or the increase in high-risk operation frequency at night. All evaluation results are generated in the form of charts and text in a comprehensive report, which includes key indicator curves, hotspot area distribution, and optimization suggestions. The "sports behavior evaluation report" can be used by site operators for scientific planning, risk prevention and control, and service improvement, achieving intelligent management and safe operation of ice and snow sports public places.

[0033] In this embodiment, the specific steps of step S5 are: Based on the motion behavior evaluation report and the fatigue prediction value of each device state, the service life of the device is quantified, and device service life data is generated; Analyze the service life data of the device for abnormal use, and identify abnormal use states and abnormal devices; Generate an automatic warning report based on the abnormal use state and the abnormal device; Based on the automatic warning report, analyze the device maintenance, and make device resource configuration decisions based on the motion behavior evaluation report.

[0034] In this embodiment, the fatigue prediction value, the use intensity curve, the environmental load parameter and the behavior evaluation result are standardized, and the dimension and time scale are unified. Subsequently, according to the working stress, temperature fluctuation amplitude and use frequency of the device under different motion scenes, the life attenuation trend is calculated. The life quantification adopts a cumulative analysis method based on time segments, which weights and superimposes the fatigue rate and use intensity of the device in a continuous time window to calculate the life consumption ratio. The stress distribution balance degree, temperature gradient stability and vibration concentration degree are used as life loss factors to participate in weight distribution. After data fusion, a life information set containing device number, cumulative use time, life consumption rate, remaining use period and health level can be generated. The life data can reflect the wear state and performance degradation trend of the device under different use conditions, and provide a basis for subsequent abnormal identification and maintenance analysis. A reference baseline for device life attenuation is established, and the average life curve of the same type of device and the attenuation rate under normal use conditions are selected as the comparison standard. Compare the real-time life consumption data, and when the life decline rate of a single device significantly exceeds the baseline value, or there is a sharp fluctuation in a short period of time, it is determined to be abnormal use. In order to enhance the accuracy of judgment, multi-dimensional data such as temperature change, vibration intensity, device load rate and people flow use characteristics can be referred to at the same time, and the abnormal type is identified through feature cross comparison, such as mechanical fatigue overuse, structure loosening or thermal stress concentration. Each abnormal state will be recorded as an event sample, including time, device number, abnormal level and possible causes. After continuous monitoring, a statistical result of device abnormal distribution can be formed, which provides a basis for subsequent risk warning and maintenance decision.

[0035] After identifying the abnormal use state and equipment, it is necessary to generate corresponding automatic warning report according to the abnormal degree and characteristics. The warning process includes three links of abnormal grading, response triggering and report generation. Abnormal grading is evaluated according to comprehensive parameters such as life attenuation rate, temperature abnormal amplitude and vibration intensity change, which can be divided into three levels of mild, moderate and severe. When the duration of abnormal event exceeds the set threshold, or the same equipment repeatedly appears multiple times in a short time, the warning mechanism is automatically triggered. The response measures include on-site sound and light warning, data platform alarm prompt and remote maintenance port notification. The warning report records the equipment identification, location coordinates, abnormal type, occurrence period and recommended treatment measures, and the report information is pushed to the management platform and maintenance personnel terminal in real time. Through this process, potential hazards can be found in the early stage, and equipment failure or safety accidents can be avoided, realizing the change from passive maintenance to active prevention. All warning records are classified and summarized, and the maintenance priority is determined according to the abnormal level, life remaining proportion and equipment type. For equipment with serious life consumption or frequent abnormality, it is preferentially included in the maintenance list and arranged for fixed point detection or replacement. Then, combined with the people flow distribution, equipment use density and site activity characteristics provided in the motion behavior evaluation report, the equipment resources are dynamically configured. The equipment quantity or standby resources can be appropriately increased in the area with high use frequency, and maintenance scheduling and energy consumption optimization can be carried out in the low load area. Maintenance analysis considers seasonal changes, temperature environment and use cycle differences to form a complete maintenance plan and equipment scheduling strategy. The final equipment maintenance and configuration decision results can provide quantitative basis for site operation management, realize the goals of equipment life extension, operation safety improvement and fine management of resource configuration.

[0036] In the embodiment, the specific steps of equipment maintenance analysis based on automatic warning report and equipment resource configuration decision based on motion behavior evaluation report are as follows: Perform equipment maintenance analysis based on the automatic warning report to generate an equipment maintenance strategy; Perform scene people flow density growth trend analysis based on the motion behavior evaluation report to generate a people flow growth trend; Perform equipment resource adaptability evaluation based on the people flow growth trend to obtain an equipment-people flow adaptation value; Perform equipment resource configuration decision based on the equipment-people flow adaptation value to generate an equipment configuration strategy; Drive the equipment monitoring and motion data analysis jobs of the public ice and snow place based on the equipment maintenance strategy and the equipment configuration strategy.

[0037] In this embodiment, the abnormal equipment data in the early warning report is aggregated and classified, and a maintenance priority matrix is established according to the abnormal type, severity level, occurrence frequency and life remaining proportion. For high-frequency or high-risk equipment, priority is given to recent maintenance plans; for mild abnormal or low-usage equipment, periodic inspection plans are included. In the maintenance analysis process, by comparing the life attenuation trend, fatigue rate change and use intensity distribution, it is judged whether the equipment is in the performance boundary interval or there is potential structural fatigue. Combined with the physical state data of the equipment (such as temperature, pressure, vibration balance) and the operating environment parameters (such as ice surface temperature, humidity, use density), the maintenance period can be adaptively adjusted. In generating the maintenance strategy, a hierarchical decision principle is adopted, i.e. divided into three categories of emergency repair, preventive maintenance and periodic overhaul, and each type of maintenance scheme contains task priority, estimated working hours, spare parts demand and scheduling period, etc. The final equipment maintenance strategy can be used as the core basis for subsequent resource allocation and operation control, realizing the scientific and fine management of equipment. Based on the people flow distribution, activity trajectory and use time data provided in the motion behavior evaluation report, first, the time series modeling of the people flow change rate in different areas is carried out. The sliding time window method is used to divide the peak and trough periods of each day into fixed intervals (for example, 30 minutes), and the growth rate and change amplitude of people flow density are calculated. Through the comparison of consecutive time periods, the regularity of people flow growth trend can be identified, such as holiday peak, weekend concentrated flow or night increase amplitude mode. In the trend identification process, combined with historical people flow data and real-time monitoring results, the time weighted smoothing method is used to extract the main trend item of the growth curve, and the growth critical point is marked. When the growth rate continuously exceeds the set threshold, it can be determined as the rapid rising stage of people flow. The final "people flow growth trend data" contains time interval, region number, growth rate, peak prediction period and growth intensity level, which provides the basis for subsequent equipment resource matching and capacity optimization.

[0038] The matching degree between the equipment and the flow of people is calculated by comparing the current running state, usage intensity, available quantity of the equipment and the predicted growth rate of the flow of people. First, the available performance interval of the equipment is determined according to the remaining life value, health level and workload range in the equipment maintenance strategy. Then, the usage pressure of each type of equipment in different scene areas is analyzed in combination with the density distribution data and usage intensity index in the flow of people growth trend. The adaptability evaluation obtains the "equipment-flow adaptability value" by weighted calculation of the ratio of the available rate of each area equipment to the flow demand, which is used to measure whether the resource distribution is balanced. When the adaptability value is close to 1, it indicates that the supply and demand are balanced; lower than 0.7, it means that the equipment is insufficient; higher than 1.3, it represents resource redundancy. The evaluation results reflect the spatial location and time period characteristics of the equipment overload or idling. Through continuous monitoring and periodic correction, a dynamic adaptability curve can be formed, providing a quantitative reference for equipment scheduling and site operation optimization. After obtaining the equipment-flow adaptability value, equipment resource allocation decisions need to be made based on it to achieve the optimal allocation of resources in each area. The allocation decision takes the adaptability value, equipment health level and site usage type as the main input parameters. First, identify the areas where the equipment resources are insufficient, and prioritize scheduling or allocation from areas with higher adaptability values (resource surplus). For areas with high usage frequency but fast equipment life consumption, equipment rotation and intermittent use strategies can be adopted to avoid excessive wear and tear. At the same time, according to the prediction of the flow of people growth trend, standby equipment can be deployed in peak periods or activity-intensive areas in advance to ensure the continuity of site operation. The equipment configuration strategy not only includes spatial distribution adjustment, but also includes the coordinated planning of maintenance time and operating load, such as scheduling concentrated maintenance tasks during low-flow periods to reduce the impact on user experience. The final generated equipment configuration strategy file contains equipment number, deployment path, target area, configuration time and priority, etc., which constitutes the core data basis of intelligent scheduling of equipment resources. After the development of maintenance strategy and configuration strategy, the two types of strategies together constitute the execution drive of equipment operation management and sports data analysis. The equipment maintenance strategy ensures the safe and controllable operation of each equipment, while the equipment configuration strategy ensures the overall resource distribution of the site to match the dynamic flow of people. Combined with the two, the monitoring linkage and data feedback are realized through the Internet of Things platform. The real-time running state of the equipment, the progress of the maintenance plan execution and the configuration adjustment record are synchronized to the data center for dynamic analysis and optimization. Through continuous data flow collection, the equipment state model can be corrected in real time, the flow of people behavior pattern can be updated, and the threshold setting of the monitoring algorithm can be adjusted. The monitoring platform displays key indicators such as equipment health score, regional load level, usage duration distribution and flow density change curve through a visual interface, forming a closed-loop monitoring-analysis-decision system. The ultimate goal is to realize an intelligent, data-based and self-adaptive comprehensive operation mechanism in the aspects of safety monitoring, energy consumption control, personnel scheduling and equipment management in public ice and snow venues.

[0039] In the embodiment, the device resource configuration decision based on the device-person flow adaptation value generates a device configuration strategy, which specifically comprises: The device-person flow adaptation value is compared with a preset threshold value of devices per person in a place. When the preset threshold value of devices per person in a place is greater than the device-person flow adaptation value, device redundancy optimization is performed. When the preset threshold value of devices per person in a place is less than the device-person flow adaptation value, a device shortage procurement decision is made.

[0040] In this embodiment, after completing the equipment-person flow adaptability evaluation, it is necessary to judge whether the equipment resource configuration is within a reasonable range through threshold comparison. First, set the per capita equipment threshold of the place, which can be determined according to the type of the place, functional partition and management standard. For example, in public places such as skating rinks and skiing experience areas, the standard equipment quantity corresponding to every hundred users can be set according to historical data, such as 15 sets of skis for every hundred people and 20 pairs of ice skates for every hundred people. Then, compare the preset threshold value with the equipment-person flow adaptation value region by region. When the threshold value is greater than the adaptation value, it means that the number of equipment in this area is relatively excessive, and there is a redundancy phenomenon. At this time, the equipment resources need to be optimized, mainly including equipment parking area adjustment, spare equipment collection, equipment rotation interval extension and other measures. In order to improve the utilization rate of resources, trend smoothing analysis can be performed on multi-period data to determine whether the redundancy exists continuously. If the equipment idle rate exceeds 30% in continuous multiple time periods, the equipment redundancy optimization strategy is started. This strategy includes equipment dispatching, early maintenance of low-usage equipment, standby equipment energy consumption control and resource redistribution. In this way, energy consumption and maintenance costs can be effectively reduced, while avoiding performance degradation caused by long-term idling of equipment, and achieving dynamic balance management of place resources. In the threshold comparison process, when the per capita equipment threshold is lower than the equipment-person flow adaptation value, it means that the equipment supply is insufficient, and the person flow density has exceeded the service capacity that the current equipment can support, and the equipment shortage response mechanism needs to be started. First, identify the shortage area, determine the shortage level according to the real-time person flow density distribution and the deviation of the adaptation value. If the adaptation value exceeds the threshold value by 15% to 30%, it is determined as moderate shortage; if it exceeds 30% or more, it is determined as serious shortage. Then carry out analysis of the type of equipment in short supply, determine the type and specification of the equipment that needs to be supplemented most, such as ski poles, protective gear, balance training equipment, etc. The procurement decision not only considers the short-term supplement demand, but also combines the trend of person flow growth and seasonal change characteristics to establish a prediction model to estimate the future equipment demand. According to the prediction results, generate a procurement plan, including equipment model, quantity, budget range and delivery cycle, etc. At the same time, temporary loan scheme can be set in the equipment scheduling strategy, that is, temporarily allocate equipment from low-traffic areas or areas with low training intensity to relieve peak pressure. After the procurement decision is made, the new equipment needs to be included in the monitoring and maintenance plan to realize the coordinated management of new and old equipment. Through this step, it can ensure the sufficient supply of equipment resources in the high-load period, and maintain the safety, comfort and operation continuity of the ice and snow sports place.

[0041] In this embodiment, a motion training data monitoring device based on Internet of Things is provided for executing the motion training data monitoring method based on Internet of Things as described above, comprising: The device usage analysis module is configured to deploy a multi-modal sensor array to collect real-time sports device usage data of a public ice and snow venue, analyze the data, and generate device dynamic usage information of different scenes. The intensity quantification module is configured to quantize usage intensity according to the device dynamic usage information, and generate a usage intensity time curve. The fatigue prediction module is configured to perform periodic fatigue prediction based on the usage intensity time curve, and generate fatigue prediction values of each device state. The people flow statistics module is configured to perform scene sports people flow statistics according to the multi-modal sensor array, and obtain a sports behavior evaluation report. The resource configuration module is configured to make device maintenance and device resource configuration decisions based on the sports behavior evaluation report and the fatigue prediction values of each device state.

[0042] Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting, the scope of the present application being defined by the appended claims and not by the above description, and all changes falling within the meaning and range of equivalence of the essential features of the application are intended to be embraced therein.

[0043] The above description is merely that of specific embodiments of the present application, enabling a person skilled in the art to understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will accord with the widest scope consistent with the principles and novel features developed herein.

Claims

1. An Internet of Things-based sports training data monitoring method, characterized in that, The method comprises the following steps: Step S1: deploying a multi-modal sensor array to collect real-time sports equipment usage data of public ice and snow venues, performing data analysis, and generating equipment dynamic usage information in different scenarios; Step S2: quantifying the usage intensity according to the equipment dynamic usage information, and generating a usage intensity time curve; Step S3: predicting periodic fatigue based on the usage intensity time curve, and generating fatigue prediction values of each equipment state; Step S4: performing scene sports crowd statistics according to the multi-modal sensor array, and obtaining a sports behavior evaluation report; Step S5: making equipment maintenance and equipment resource allocation decisions based on the sports behavior evaluation report and the fatigue prediction values of each equipment state. 2.The Internet of Things based sports training data monitoring method according to claim 1, characterized in that, The specific steps of step S1 are: deploying a multi-modal sensor array to collect real-time sports equipment usage data of public ice and snow venues; performing sensor abnormal jitter identification on the sports equipment usage data, and filtering abnormal noise to obtain standardized equipment usage samples; performing equipment position identification based on the multi-modal sensor array, and classifying the usage scene types to obtain multiple usage scenes; labeling the standardized equipment usage samples according to the multiple usage scenes to generate scene label samples; calculating the pressure distribution entropy, temperature gradient change rate, frequency concentration of vibration spectrum, and personnel gathering heat density based on the scene label samples to generate equipment dynamic usage information in different scenes. 3.The Internet of Things based sports training data monitoring method according to claim 1, characterized in that, The specific steps of step S2 are: extracting the data collection time stamp of the sports equipment usage data; performing inter-sensor time synchronization processing on the data collection time stamp to obtain a synchronized time stamp; performing day and night and season identification based on the synchronized time stamp to generate periodic data; calculating the equipment usage frequency and usage time length based on the equipment dynamic usage information; performing periodic usage mode analysis on the equipment usage frequency and usage time length based on the periodic data to obtain equipment usage modes in different periods; quantifying the equipment usage modes and performing multiple time window intensity distribution statistics to generate a usage intensity time curve. 4.The Internet of Things based sports training data monitoring method according to claim 1, characterized in that, The specific steps of step S3 are: detecting equipment physical state indicators based on a multi-modal sensor array; performing dynamic change analysis based on the equipment physical state indicators to generate state change data; constructing a fatigue evaluation model for each equipment based on the state change data; inputting the usage intensity time curve into the fatigue evaluation model for each equipment to perform equipment state health degree scoring and generate real-time equipment fatigue score values; performing curve evolution trend analysis on the real-time equipment fatigue score values to extract equipment fatigue evolution trends; performing fatigue rate calculation based on the equipment fatigue evolution trends to generate fatigue rate change rules; performing periodic fatigue prediction based on the fatigue rate change rules to generate fatigue prediction values of each equipment state. 5.The Internet of Things based sports training data monitoring method according to claim 1, wherein, The specific steps of step S4 are: performing scene sports crowd statistics based on the multi-modal sensor array to obtain sports crowd monitoring data; calculating user average stay time, activity trajectory heat distribution, equipment and personnel interaction frequency, and high-risk operation frequency based on the sports crowd monitoring data to generate crowd behavior feature data; Clustering user identity types based on human flow behavior characteristic data to identify sports user types; Calculating human flow distribution based on sports user types to generate multi-type human flow distribution maps; Evaluating human flow behavior based on human flow behavior characteristic data to generate sports behavior evaluation reports. 6.The Internet of Things based sports training data monitoring method according to claim 1, wherein, The specific steps of step S5 are: Quantifying device service life based on sports behavior evaluation reports and device state fatigue prediction values to generate device service life data; Analyzing abnormal use of device service life data to identify abnormal use states and abnormal devices; Generating automatic warning reports based on abnormal use states and abnormal devices; Performing device maintenance analysis based on automatic warning reports and making device resource configuration decisions based on sports behavior evaluation reports. 7.The Internet of Things based sports training data monitoring method according to claim 6, characterized in that, The specific steps of performing device maintenance analysis based on automatic warning reports and making device resource configuration decisions based on sports behavior evaluation reports are: Performing device maintenance analysis based on automatic warning reports to generate device maintenance strategies; Analyzing scene human flow density growth trends based on sports behavior evaluation reports to generate human flow growth trends; Evaluating device resource adaptability based on human flow growth trends to obtain device-human flow adaptation values; Making device resource configuration decisions based on device-human flow adaptation values to generate device configuration strategies; Driving device monitoring and sports data analysis jobs of public ice and snow places based on device maintenance strategies and device configuration strategies. 8.The Internet of Things based sports training data monitoring method according to claim 7, characterized in that, The specific steps of making device resource configuration decisions based on device-human flow adaptation values to generate device configuration strategies are: Comparing device-human flow adaptation values with preset per-capita device thresholds of places to perform device redundancy optimization when the preset per-capita device thresholds of places are greater than the device-human flow adaptation values; Making device shortage procurement decisions when the preset per-capita device thresholds of places are less than the device-human flow adaptation values.

9. An Internet of Things based sports training data monitoring device, characterized by, A method for performing the Internet of Things-based sports training data monitoring method of claim 1, comprising: A device use analysis module for deploying a multi-modal sensor array to collect sports device use data of public ice and snow places in real time, performing data analysis, and generating device dynamic use information in different scenarios; An intensity quantification module for quantifying use intensity based on the device dynamic use information to generate a use intensity time series curve; A fatigue prediction module for predicting periodic fatigue based on the use intensity time series curve to generate device state fatigue prediction values; A human flow statistics module for performing scene sports human flow statistics based on the multi-modal sensor array to obtain sports behavior evaluation reports; A resource configuration module for making device maintenance and device resource configuration decisions based on sports behavior evaluation reports and device state fatigue prediction values.

Images