Middle-distance running training management and data analysis system

Abstract

The invention provides a middle and long distance running training management and data analysis system, and relates to the technical field of physical training, the system comprises an acquisition module used for acquiring motion information of a target athlete, the motion information comprising motion data and physiological indexes; the processing module is used for performing segmentation processing on the motion data based on a preset segmentation algorithm to obtain a segmentation result; and the management module is used for matching the segmentation result and the physiological indexes with preset training courses to obtain a matching result, and adjusting the corresponding preset training courses according to the matching result. According to the method, the exercise data is segmented, namely, the whole exercise process is divided into different exercise stages, so that coaches and athletes can obtain feedback in time in the training process, the training progress and the training effect are known, the training strategy is adjusted in time according to the feedback, and then the training efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sports training, and in particular to a middle- and long-distance running training management and data analysis system. Background Art

[0002] With the rise of middle- and long-distance running and marathons, sports science and athletic coaching face enormous research demands. The continuous advancement and widespread adoption of smart wearable device technology has enabled the widespread application of biometric data such as heart rate, cadence, and speed in running training. This data holds immense value and potential, which urgently needs to be fully explored. Accurate performance prediction can help runners optimize their training and improve their competitive level, while also contributing to the general public's physical fitness.

[0003] However, current market research and development is still in its infancy, relying primarily on empirical formulas and lacking in-depth, data-driven research and development. Furthermore, there is a lack of convenient solutions for aggregating and analyzing team training information. Summary of the Invention

[0004] The problem solved by the present invention is one or more of the above-mentioned problems in the prior art.

[0005] To solve the above problems, the present invention provides a middle- and long-distance running training management and data analysis system.

[0006] In a first aspect, the present invention provides a middle- and long-distance running training management and data analysis system, comprising: A collection module, used to collect sports information of target athletes, wherein the sports information includes sports data and physiological indicators; A processing module, configured to perform segmentation processing on the motion data based on a preset segmentation algorithm to obtain a segmentation result; The management module is used to match the segmentation results and the physiological indicators with the preset training courses to obtain matching results, and adjust the corresponding preset training courses according to the matching results.

[0007] Optionally, the processing module is specifically configured to include: performing smoothing processing on the motion data to obtain temporary motion data; Determining the corresponding preset segmentation algorithm based on the temporary motion data according to a preset determination method; The motion data is segmented according to the corresponding preset segmentation algorithm to obtain the segmentation result.

[0008] Optionally, the motion data includes a speed change sequence, and the preset determination method includes a first determination method; and determining the corresponding preset segmentation algorithm based on the temporary motion data according to the preset determination method includes: When the first discrimination method is adopted, the speed change sequence is discriminated based on a preset classifier to obtain a discrimination result, and the matching preset segmentation algorithm is determined according to the discrimination result; The speed change sequence is segmented according to the corresponding preset segmentation algorithm to obtain the segmentation result.

[0009] Optionally, the preset segmentation algorithm includes a Bayesian online change point detection algorithm, and the segmentation processing of the speed change sequence according to the corresponding preset segmentation algorithm to obtain a segmentation result includes: determining an initial mean and an initial variance, and calculating a posterior probability distribution of a speed run length in the temporary motion data based on the initial mean, the initial variance, and the speed change sequence; According to the comparison between the difference between adjacent data in the posterior probability distribution and a preset threshold, the data points whose difference is greater than the preset threshold are used as segmentation points, and the segmentation result is determined based on all the segmentation points.

[0010] Optionally, the preset segmentation algorithm further includes a preset threshold discrimination algorithm, and the segmentation processing of the speed change sequence according to the corresponding preset segmentation algorithm to obtain the segmentation result further includes: Get the preset high-speed threshold and low-speed threshold; The speed change sequence is segmented using the preset high-speed threshold and the preset low-speed threshold to obtain the segmentation result.

[0011] Optionally, the preset determination method further includes a second determination method, and determining the corresponding preset segmentation algorithm based on the temporary motion data according to the preset determination method further includes: When the second discrimination method is adopted, data calculation is performed on the speed change sequence to obtain a calculation result, which includes a mean, a median and a standard deviation; A corresponding distribution form is determined according to the speed change sequence and the calculation result, and the matching preset segmentation algorithm is determined according to the distribution form.

[0012] Optionally, the sports information includes real sports environment data and basic information; the middle and long distance running training management and data analysis system also includes a load prediction module and an evaluation module, The load prediction module is used to determine the target motion data set for the target athlete in the preset training course, and obtain corresponding heart rate prediction data based on a preset load prediction model according to the target motion data, the sports environment data and the basic information; wherein, the preset load prediction model is constructed based on a long short-term memory network.

[0013] Optionally, the physiological index includes actual heart rate data; the middle and long distance running training management and data analysis system also includes an evaluation module, The evaluation module is used to evaluate the actual heart rate data and the corresponding predicted heart rate data to obtain corresponding evaluation data, and the evaluation data is used to evaluate the performance of this training; The evaluation module is further configured to calculate a training load based on the actual heart rate data, and the training load is used to evaluate the burden on the target athlete's body.

[0014] Optionally, the middle- and long-distance running training management and data analysis system further includes a visualization module, a system end module and a storage module. The visualization module is used to query the data during the training process and present it in the form of charts; The system end module is used to coordinate and integrate the operation and data transmission between the modules in the middle and long distance running training management and data analysis system; The storage module is used to store the operation data between the modules in the middle and long distance running training management and data analysis system.

[0015] Optionally, the management module is further configured to send the adjusted information of the preset training course to the corresponding target athlete.

[0016] The beneficial effects of the middle and long distance running training management and data analysis system of the present invention are: First, the acquisition module collects the motion information of the target athlete, which usually includes motion data (speed change sequence and corresponding event type, exercise mileage, total exercise time, etc.), physiological indicators (such as actual heart rate data, etc.), and operating environment data (such as humidity, temperature, and altitude data); providing comprehensive data support for the evaluation of training effects.

[0017] The processing module pre-processes the received data (such as motion data), including data cleaning, smoothing and denoising, and then segments the processed data according to the preset segmentation algorithm to obtain segmentation results. This segmentation process divides the motion process into different stages, such as jogging stage, sprinting stage, rest stage, etc. By accurately identifying different motion stages, excessive training or ineffective training in inappropriate stages is avoided. That is, by processing the data through a scientific segmentation algorithm, the evaluation of training effects is more accurate and detailed.

[0018] Finally, the management module matches the segmented results with the pre-set training program. Pre-set training programs typically include detailed metrics such as mileage, duration, and target speed ranges for each phase (e.g., warm-up, steady-pace run, sprint, etc.), as well as heart rate zones. The matching rules are based on the degree of similarity between the actual mileage, duration, average speed, and average heart rate of each phase and the pre-set values. A match is considered successful when the similarity reaches a certain threshold (e.g., 80%). The corresponding pre-set training program is then adjusted based on the matching results. For example, if the athlete's speed in a certain phase of the actual segmented results exceeds the pre-set program requirements while their heart rate is within a reasonable range, the intensity of subsequent training in that phase can be increased. Conversely, if the speed is below the target and the heart rate is too high, the speed target for that phase can be lowered or the duration can be shortened. In other words, the management module combines the matching of segmented results and physiological indicators with the pre-set program, combined with statistical analysis of key metrics (e.g., total distance, total time, average pace, etc.), to provide coaches and athletes with an objective and quantitative basis for evaluating training effectiveness. This helps to scientifically judge the athletes' training level and progress, and provides a strong reference for subsequent training arrangements.

[0019] Therefore, the present invention is to divide the entire motion process into different motion phases by carrying out segmentation processing to motion data, so that coaches and athletes can obtain feedback in time during the training process, thereby understanding the training progress and effect, and adjusting the training strategy in time accordingly, thereby improving training efficiency. This segmentation processing method can accurately identify different motion phases, effectively avoiding overtraining or ineffective training in inappropriate stages. For example, in the warm-up stage, it is possible to find out in time whether the athlete has fully moved his body, avoiding the risk of injury caused by insufficient warm-up, and also preventing the warm-up time from being too long and wasting training time. In addition, accurately dividing the motion phase can ensure that the training time and intensity of each stage are reasonably allocated. For example, the time and intensity of the warm-up, steady-state exercise and sprint phase are reasonably arranged so that the athlete's body can adapt to different exercise loads, thereby reducing the probability of occurrence of sports injuries such as muscle strain and joint sprain.

[0020] At the same time, by matching segmented results and physiological indicators (such as heart rate data) with preset training courses, it is possible to accurately understand the difference between an athlete's actual performance in each training phase and the preset goals. For example, if an athlete's speed during a steady-pace run is lower than the preset speed, or their heart rate data is too high, the system can adjust the training intensity of that phase accordingly, providing the athlete with personalized training guidance to meet the individual differences and training needs of different athletes. Moreover, by timely adjusting the preset training courses based on the matching results, the training plan can be more closely aligned with the athlete's actual ability level and development needs, helping athletes gradually improve their athletic performance and avoid overtraining or undertraining.

[0021] Furthermore, by monitoring exercise data in real time and adjusting training sessions appropriately, athletes can avoid prolonged periods of high-intensity exercise. For example, if heart rate monitoring indicates an athlete's heart rate is too high for an extended period, the system can promptly remind them to reduce their exercise intensity, thereby preventing sports injuries such as cardiovascular disease caused by excessive fatigue. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a structural diagram of a middle- and long-distance running training management and data analysis system according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a segmentation result according to an embodiment of the present invention; Figure 3 This is a second schematic diagram of the segmentation result of an embodiment of the present invention; Figure 4 This is a second structural diagram of a middle- and long-distance running training management and data analysis system according to an embodiment of the present invention; Figure 5 A schematic diagram of the structure of a preset load forecasting model according to an embodiment of the present invention; Figure 6 Schematic diagram of the prediction effect of the preset load prediction model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0023] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings. Although certain embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0024] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0025] The term "including" and its variations used in this document are open inclusions, that is, "including but not limited to"; the term "based on" means "based at least in part on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one other embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "optional embodiments". The relevant definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0026] It should be noted that the modifications of "one" and "multiple" mentioned in the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0027] The names of the messages or information exchanged between multiple devices in the embodiments of the present invention are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0028] like Figure 1 As shown, an embodiment of the present invention provides a middle- and long-distance running training management and data analysis system, including: The acquisition module is used to acquire the sports information of the target athlete, wherein the sports information includes sports data and physiological indicators.

[0029] Specifically, the acquisition module can use mobile sports devices or wearable devices (such as sports bracelets and running watches) to collect the target athlete's motion information at a certain sampling frequency, such as once per second. This information includes motion data such as event type (such as middle- and long-distance running), mileage, total exercise duration, and the sequence of position and speed changes during exercise. The device transmits this data to the system using wireless communication methods such as Bluetooth and WiFi. This motion information typically includes motion data (such as event type (such as middle- and long-distance running), mileage, total exercise duration, and the sequence of position and speed changes during exercise), physiological indicators (such as actual heart rate data and heart rate variability (HRV)), operating environment data (such as humidity, temperature, and altitude), and basic information such as the target athlete's corresponding user ID, gender, sport type, and age, providing comprehensive data support for evaluating training effectiveness.

[0030] The processing module is used to perform segmentation processing on the motion data based on a preset segmentation algorithm to obtain a segmentation result.

[0031] Specifically, the processing module is responsible for in-depth processing of the collected motion data, with its core focus on achieving precise segmentation based on a pre-set segmentation algorithm. Specifically, after the data acquisition module acquires the athlete's motion information, such as speed, heart rate, and HRV, this raw data is transmitted to the processing module. The module's built-in pre-set segmentation algorithm scans and analyzes the data sequence based on established rules, such as heart rate fluctuation amplitude and speed change thresholds, to identify different motion phases and automatically segment the data. This segmentation processing method accurately delineates different motion phases based on the athlete's actual training performance, such as changes in heart rate and speed. This not only improves training efficiency but also provides data support for the development of personalized training plans. This allows coaches to gain a deeper understanding of the athlete's training progress at each stage, enabling them to develop more targeted training plans. This not only helps athletes optimize training results and improve their competitive level, but also effectively prevents sports injuries caused by inappropriate training schedules.

[0032] The management module is used to match the segmentation results and the physiological indicators with the preset training courses to obtain matching results, and adjust the corresponding preset training courses according to the matching results.

[0033] Specifically, the management module obtains the segmentation results from the processing module and physiological indicators (such as heart rate, HRV, etc.) from the acquisition module. Preset training courses contain detailed training goals and plans, such as exercise mileage, duration, specific target speed ranges for each stage (such as warm-up, steady-pace running, sprinting, etc.), heart rate ranges, etc. The system compares the segmentation results and physiological indicators with the preset training course and calculates similarity. That is, the degree of match is evaluated based on the proximity of the actual mileage, duration, average speed, average heart rate, and other indicators of each stage to the preset values. The preset training course is adjusted based on the matching results. For example, if the athlete's speed in a certain stage is faster than the preset speed and the heart rate is reasonable, the intensity of subsequent training in that stage will be increased; conversely, if the speed does not meet the standard and the heart rate is too high, the intensity will be reduced or the duration will be shortened.

[0034] By monitoring exercise data and physiological indicators in real time, training results are promptly fed back. If the segment results show that the athlete's speed is not up to standard, the system can issue a reminder, allowing the coach to immediately adjust the course, optimize the training plan, improve efficiency, and shorten the time to reach the target.

[0035] The system also collects a wealth of motion data and physiological indicators, providing comprehensive data support for evaluating training effectiveness. By analyzing this data, athletes' training effects can be quantitatively assessed. For example, by comparing changes in physiological indicators before and after adjustments, a clearer understanding of the impact of training on the athlete's physical function can be obtained. Physiological indicators can also be monitored in real time, allowing for appropriate adjustments to training intensity to avoid overtraining. For example, if the heart rate monitor shows that an athlete's heart rate is too high and persists for an extended period, the system will promptly remind them to reduce exercise intensity to prevent sports injuries such as cardiovascular disease caused by excessive fatigue.

[0036] It should be noted that when adjusting the corresponding preset training course according to the matching result, the preset training course can be adjusted in real time during the training process according to the matching result, and the training course for the next training can also be adjusted according to the matching result.

[0037] In this embodiment, first, the acquisition module collects the motion information of the target athlete. This motion information generally includes motion data (such as speed change sequence and corresponding event type, exercise mileage, total exercise time, etc.), physiological indicators (such as actual heart rate data, etc.) and operating environment data (such as humidity, temperature and altitude data), providing comprehensive data support for the evaluation of training effects.

[0038] Next, the processing module preprocesses the received data (such as exercise data), including data cleaning, smoothing, and denoising. It then segments the processed data according to a preset segmentation algorithm to produce segmented results. This segmentation process divides the exercise process into distinct phases, such as jogging, sprinting, and resting. By accurately identifying different exercise phases, overtraining or ineffective training during inappropriate periods is avoided. In other words, data processing using a scientific segmentation algorithm enables more accurate and detailed assessment of training effectiveness.

[0039] Finally, the management module matches the segmented results with pre-set training programs. Pre-set training programs typically include detailed metrics such as mileage, duration, and target speed ranges for each phase (e.g., warm-up, steady-pace run, sprint, etc.), as well as heart rate zones. The matching rules are based on the degree of similarity between the actual mileage, duration, average speed, and average heart rate of each phase and the pre-set values. A match is considered successful when the similarity reaches a certain threshold (e.g., 80%). The corresponding pre-set training program is then adjusted based on the matching results. For example, if the athlete's speed in a certain phase of the segmented results exceeds the pre-set program requirements while their heart rate is within a reasonable range, the intensity of subsequent training in that phase can be increased. Conversely, if the speed falls short of the target and the heart rate is too high, the speed target for that phase can be lowered or the duration can be shortened. In other words, the management module combines the matching of segmented results and physiological indicators with the pre-set program, combined with statistical analysis of key metrics (e.g., total distance, total time, average pace, etc.), to provide coaches and athletes with an objective and quantitative evaluation of training effectiveness. This helps scientifically assess an athlete's training level and progress, providing a strong basis for subsequent training arrangements.

[0040] Therefore, the present invention is to divide the entire motion process into different motion phases by carrying out segmentation processing to motion data, so that coaches and athletes can obtain feedback in time during the training process, thereby understanding the training progress and effect, and adjusting the training strategy in time accordingly, thereby improving training efficiency. This segmentation processing method can accurately identify different motion phases, effectively avoiding overtraining or ineffective training in inappropriate stages. For example, in the warm-up stage, it is possible to find out in time whether the athlete has fully moved his body, avoiding the risk of injury caused by insufficient warm-up, and also preventing the warm-up time from being too long and wasting training time. In addition, accurately dividing the motion phase can ensure that the training time and intensity of each stage are reasonably allocated. For example, the time and intensity of the warm-up, steady-state exercise and sprint phase are reasonably arranged so that the athlete's body can adapt to different exercise loads, thereby reducing the probability of occurrence of sports injuries such as muscle strain and joint sprain.

[0041] At the same time, by matching segmented results and physiological indicators (such as heart rate data) with preset training courses, it is possible to accurately understand the difference between an athlete's actual performance in each training phase and the preset goals. For example, if an athlete's speed during a steady-pace run is lower than the preset speed, or their heart rate data is too high, the system can adjust the training intensity of that phase accordingly, providing the athlete with personalized training guidance to meet the individual differences and training needs of different athletes. Moreover, by timely adjusting the preset training courses based on the matching results, the training plan can be more closely aligned with the athlete's actual ability level and development needs, helping athletes gradually improve their athletic performance and avoid overtraining or undertraining.

[0042] Furthermore, by monitoring exercise data in real time and adjusting training sessions appropriately, athletes can avoid prolonged periods of high-intensity exercise. For example, if heart rate monitoring indicates an athlete's heart rate is too high for an extended period, the system can promptly remind them to reduce their exercise intensity, thereby preventing sports injuries such as cardiovascular disease caused by excessive fatigue.

[0043] Optionally, the processing module is specifically configured to include: performing smoothing processing on the motion data to obtain temporary motion data; Determining the corresponding preset segmentation algorithm based on the temporary motion data according to a preset determination method; The motion data is segmented according to the corresponding preset segmentation algorithm to obtain the segmentation result.

[0044] Specifically, the raw motion data is preprocessed, including data cleaning, smoothing, and denoising. For example, the heart rate curve is smoothed using a moving average method, and small amounts of missing position data are filled with interpolation to account for errors, missing data, or outliers that may occur during the wearable device sampling process.

[0045] Optionally, the motion data includes a speed change sequence, and the preset determination method includes a first determination method; and determining the corresponding preset segmentation algorithm based on the temporary motion data according to the preset determination method includes: When the first discrimination method is adopted, the speed change sequence is discriminated based on a preset classifier to obtain a discrimination result, and the matching preset segmentation algorithm is determined according to the discrimination result; The speed change sequence is segmented according to the corresponding preset segmentation algorithm to obtain the segmentation result.

[0046] In some embodiments, the preset classifier is trained based on a large amount of historical motion data. This historical data includes speed change sequences of different types of motion. By learning these historical data through machine learning algorithms (such as decision trees, support vector machines, neural networks, etc.), the classifier can identify the speed change patterns of different types of motion stages, that is, the classifier distinguishes the speed change sequence in real time and outputs the distinction result. According to the distinction result output by the preset classifier, the system matches the corresponding preset segmentation algorithm. For example, if the classifier determines that the current stage is a uniform running stage, the system will select an algorithm suitable for uniform running segmentation; if it is an interval running stage, it will select an algorithm suitable for interval running segmentation. For example: the preset classifier is built based on the xgboost network, and the input is the features of each motion data (such as a speed change sequence). The distinction result it outputs can be 0 or 1 label data, and the corresponding preset segmentation algorithm is determined by the label data.

[0047] In some preferred embodiments, the preset classifier can also distinguish the movement phase, rather than simply the type of movement or the way of movement. The movement phase is a division of different intensities and characteristics during the movement process, such as the steady-speed running phase, the interval running phase, the sprint phase, the warm-up phase, the rest phase, etc. The division of these phases helps to more finely manage and adjust the training plan to meet the needs of athletes under different training goals. For example, the steady-speed running phase focuses on the training of stable speed and lasting endurance, while the sprint phase focuses on high-speed explosive power training in a short period of time. By accurately identifying these movement phases, the system can match and apply the most suitable segmentation algorithm to achieve accurate analysis and processing of motion data.

[0048] Optionally, the preset segmentation algorithm includes a Bayesian online change point detection algorithm, and the segmentation processing of the speed change sequence according to the corresponding preset segmentation algorithm to obtain a segmentation result includes: determining an initial mean and an initial variance, and calculating a posterior probability distribution of a speed run length in the temporary motion data based on the initial mean, the initial variance, and the speed change sequence; According to the comparison between the difference between adjacent data in the posterior probability distribution and a preset threshold, the data points whose difference is greater than the preset threshold are used as segmentation points, and the segmentation result is determined based on all the segmentation points.

[0049] In some embodiments, when the segmentation algorithm selected after discrimination is made using a preset classifier is the Bayesian online change point detection algorithm, the algorithm is based on the Bayesian theorem and uses a recursive method to update the posterior probability distribution of the speed run length (i.e., the number of seconds the current speed mode is continuously maintained) in the motion data in real time. By analyzing the posterior probability distribution, the change point position is determined, thereby dividing the motion process into different stages.

[0050] The speed run length refers to the duration, measured in seconds, of the current speed pattern. In middle- and long-distance running, an athlete's speed varies depending on factors such as physical condition and training schedule. The Bayesian online change point detection algorithm accurately identifies these speed variations, updates the posterior probability distribution of the speed run length in real time, and determines the change point location by analyzing whether the difference between adjacent data exceeds a preset threshold. This effectively divides the movement into stages and provides data support for personalized training guidance and optimization.

[0051] A change point refers to the point in a speed change sequence where the speed pattern changes. Specifically, by comparing the difference between adjacent data in the posterior probability distribution with a preset threshold, the data points where the difference is greater than the threshold are determined as segmentation points. These segmentation points are the change point locations, indicating that the speed pattern has changed significantly at this point. For example, when an athlete enters the sprint phase from the steady-state running phase, the speed will increase significantly, and a change point location will appear at this time. By determining the locations of these change points, the entire exercise process can be divided into different stages, such as the warm-up phase, the stable exercise phase, the sprint phase, etc., thereby achieving refined analysis and processing of sports data.

[0052] The specific process includes: First, define the hazard function and set the prior probability of the change point. For example, the average change point interval is 100 seconds, that is, the probability of a change point per second is 0.01. The speed change sequence is sampled every second.

[0053] Initial parameter calculation: From the collected motion data, for example, you can first take the running speed data of the first 20 seconds, calculate the initial mean and variance of the 20-second speed data, and use these two statistics as the starting parameters to provide a basis for subsequent recursive calculations.

[0054] Recursive update process: For each time point t (starting at 1 and continuing until all time points T), obtain the speed value x for that second. Then, for each possible speed run length r (ranging from 0 to t-1), calculate and update the posterior probability distribution of the speed run length r in the temporary motion data in real time according to the Bayesian recursive formula.

[0055] During the recursive update process, for each possible speed run length r, the posterior mean and accuracy need to be updated. Specifically, the new accuracy Equal to the old precision Add 1 divided by the variance σ²; the new mean μ new It is calculated by multiplying the old mean by the old precision plus the current speed value x divided by σ², and then dividing by the new precision get.

[0056] Based on the updated parameters and related settings, the posterior probability is calculated for each time point and corresponding speed run length. This results in a logarithmic posterior probability matrix, logR, where logR[t, r] represents the logarithmic posterior probability for a speed run length of r at time t. The calculation of logR differs for different speed run lengths. When the speed run length r is 0 (indicating a change point), the calculation method is to take the logarithm of the cumulative sum, where the cumulative term involves a combination of operations on different parameters. When the speed run length r is greater than or equal to 1 (indicating no change point), the calculation method is to calculate the logarithm of the relevant parameters and probabilities at the previous time point.

[0057] Finally, points where the difference in the logarithmic posterior probability matrix (logR) between two adjacent points is greater than a preset threshold (for example, 30) are identified as outliers (segmentation points). This is used to determine the location of the change point, thereby accurately segmenting the motion data. In this way, the entire exercise process can be divided into multiple stages such as warm-up, steady-state movement, and sprint based on changes in characteristics such as speed, providing accurate data for subsequent training analysis and curriculum adjustments.

[0058] In some preferred embodiments, the posterior probability update formula of the speed run length r is as follows: ; in, , marginal prediction distribution Calculated by setting Gaussian distribution; is the velocity change sequence from time 1 to t, r t is the speed running length at the current time t.

[0059] H: hazard function.

[0060] ; P gap (g=τ) is the probability density of a change point at time τ, if P gap (g) is a discrete exponential (geometric) distribution with time scale λ, then the process is memoryless and the hazard function is is constant.

[0061] Substitute the change point probability into: ; π r : Prior probability of state r.

[0062] Probability of no change point: ; For each speed run length r, update the posterior mean and accuracy: ; Finally, we get the logarithmic posterior probability matrix logR[t,r] which represents log .

[0063] .

[0064] The segmentation results are as follows Figure 2 As shown in the figure, it is a schematic diagram of the segmented results. This diagram can effectively detect changes in motion status in real time, allowing coaches to instantly capture important turning points in training and improve the efficiency and accuracy of training data analysis.

[0065] Optionally, the preset segmentation algorithm further includes a preset threshold discrimination algorithm, and the segmentation processing of the speed change sequence according to the corresponding preset segmentation algorithm to obtain the segmentation result further includes: Get the preset high-speed threshold and low-speed threshold; The speed change sequence is segmented using the preset high-speed threshold and the preset low-speed threshold to obtain the segmentation result.

[0066] Specifically, based on past experience and professional knowledge, two key speed indicators are determined, namely the preset high-speed threshold and low-speed threshold. For example, the high-speed threshold is set to 3.0m / s and the low-speed threshold is set to 1.5m / s. These two thresholds will be used to divide the athletes' exercise intensity stages and clearly distinguish between high-speed, medium-speed and low-speed exercise intervals.

[0067] The acquisition module captures complete speed change sequence data to ensure data integrity and accuracy. The speed change sequence is segmented using preset high-speed and low-speed thresholds. Specifically, the system compares each speed data point against these two thresholds and, based on the comparison results, divides the speed change sequence into multiple distinct stages.

[0068] For example, when the speed data is higher than the high-speed threshold, it is classified as the high-speed stage.

[0069] When the speed data is between the low speed threshold and the high speed threshold, it is divided into the medium speed stage.

[0070] When the speed data is lower than the low-speed threshold, it is divided into the low-speed stage.

[0071] Integrate the data from each stage to form a complete segmented result, including key information such as the speed range, start time, end time, and duration of each stage, providing a data basis for subsequent training analysis and guidance.

[0072] In some embodiments, the upper quartile and lower quartile of the speed data are selected as the thresholds for high speed and low speed. The speed sequence data is traversed, and the data is divided into different stages according to the set speed thresholds (preset high speed threshold and low speed threshold), and the starting index and duration of each stage are recorded. At the same time, the identified periodic intervals may overlap or be adjacent, and are merged. The process includes sorting the periodic intervals by starting position, traversing the sorted list, and merging overlapping or adjacent intervals into larger intervals. Finally, the merged periodic intervals are output, and each interval represents a complete motion cycle. Figure 3 As shown, the numbers are the detected cycle numbers, and each cycle is separated by a gray solid line.

[0073] Optionally, the preset determination method further includes a second determination method, and determining the corresponding preset segmentation algorithm based on the temporary motion data according to the preset determination method further includes: When the second discrimination method is adopted, data calculation is performed on the speed change sequence to obtain a calculation result, which includes a mean, a median and a standard deviation; A corresponding distribution form is determined according to the speed change sequence and the calculation result, and the matching preset segmentation algorithm is determined according to the distribution form.

[0074] Specifically, the complete speed change sequence data of the athlete is obtained from the acquisition module. The speed change sequence is calculated to obtain key statistical indicators, including the mean, median, and standard deviation. The mean reflects the average level of speed, the median reflects the middle value of speed, and the standard deviation measures the degree of dispersion of speed. The distribution of the data is analyzed by combining the histogram of the speed change sequence and the calculated statistics. If the data is symmetrical, high in the middle, low on both sides, and the mean is close to the median, it may be a normal distribution; if the data is asymmetrical or skewed, it may be another distribution form. Based on the analysis results, the distribution form of the speed change sequence is determined, such as normal distribution, skewed distribution, etc.

[0075] Match the determined distribution shape to the corresponding preset segmentation algorithm. For example, if the data conforms to a normal distribution, you can use the YES online change point detection algorithm. If the data is skewed, you can choose a segmentation method suitable for skewed data, such as a percentile-based segmentation algorithm or a preset threshold discrimination algorithm.

[0076] The speed change sequence is segmented using a matching segmentation algorithm to obtain detailed segmentation results and clarify the speed range and time interval of each motion stage.

[0077] Optionally, the motion information includes motion environment data and basic information; Figure 4As shown, the middle and long distance running training management and data analysis system also includes a load prediction module and an evaluation module. The load prediction module is used to determine the target motion data set for the target athlete in the preset training course, and obtain corresponding heart rate prediction data based on a preset load prediction model according to the target motion data, the sports environment data and the basic information; wherein, the preset load prediction model is constructed based on a long short-term memory network.

[0078] Optionally, the load prediction module is further configured to obtain corresponding heart rate prediction data based on a preset load prediction model according to the actual heart rate data, the exercise environment data and the basic information.

[0079] Specifically, if Figure 5 The following diagram shows the structure of the preset load prediction model. First, the input data is integrated: the training plan (preset training course) is converted into specific time series data, including distance series, altitude series (exercise environment data), and speed series data (preset speed series data). This data can describe the specific content of the training plan and the real-time exercise status.

[0080] Collect basic user characteristics (basic information), such as user ID, gender, sport type, age, etc. Use the Embedding Layer to convert these discrete user characteristics into continuous vector representations to characterize individual differences between athletes.

[0081] Historical performance data processing: This process organizes the athlete's historical performance data and uses it as contextual information to feed the model. This step aims to enhance the model's prediction accuracy for the current heart rate and better understand the athlete's current physiological state by drawing on their past performance.

[0082] Secondly, divide the input data into context input 1 (the most recent distance, altitude, and time series), context input 2 (the most recent speed series data), the current motion sequence (the distance, altitude, and time series from the start of the motion to the present), and user information (user embedding obtained through the user information encoding layer).

[0083] Independent Encoding and Fusion: Context Input 1 and Context Input 2 are each passed through independent contextual LSTM encoding layers to extract historical motion patterns. The resulting historical patterns are then concatenated and linearly projected to generate a contextual embedding. This contextual embedding is combined with the user embedding and concatenated with the current motion sequence temporally to form the final input for the two-layer LSTM network to capture temporal correlations during motion.

[0084] Finally, the output layer design: After the multi-layer LSTM network extracts the time series features, it outputs the prediction results through a fully connected layer and SELU activation function. The main focus is on predicting the change in heart rate from 0 to the current moment, and speed can also be inferred as needed.

[0085] If the predicted results show that the maximum heart rate or its fluctuation range exceeds the reasonable range, the coach can judge whether the current training or competition intensity is too high and make corresponding adjustments in time.

[0086] During the training of the preset load forecasting model, the loss function and optimizer settings were as follows: Mean Squared Error (MSE) was selected as the loss function for model training, and the network parameters were iteratively updated using the RMSprop optimizer (with a learning rate set to 0.005). A dropout mechanism was incorporated into the training process, and an early stopping strategy was used to select the optimal model, prevent overfitting, and ensure the model's generalization ability in practical applications.

[0087] In summary, the LSTM-based personalized heart rate prediction model (preset load prediction model) can accurately predict the heart rate trend of athletes by integrating multi-source data, adopting multi-level coding and information fusion strategies, carefully designing the output layer, optimizing the training strategy, and building an effective data set pre-training and fine-tuning process, providing scientific support for training management and competition strategies. Figure 6 As shown in the figure, the prediction effect diagram of the preset load prediction model is shown in the figure. Figure 6 It can be observed that the comparison results of the predicted heart rate data (Predicted Heart Rate) and the actual heart rate data (Actual Heart Rate) clearly show that the model has excellent prediction performance and can accurately reflect the actual heart rate change trend.

[0088] Alternatively, as Figure 4 As shown, the physiological indicators include actual heart rate data; the middle and long distance running training management and data analysis system also includes an evaluation module, The evaluation module is used to evaluate the actual heart rate data and the corresponding predicted heart rate data to obtain corresponding evaluation data, and the evaluation data is used to evaluate the performance of this training; The evaluation module is further configured to calculate a training load based on the actual heart rate data, and the training load is used to evaluate the burden on the target athlete's body.

[0089] Alternatively, as Figure 4 As shown, the middle- and long-distance running training management and data analysis system also includes a visualization module, a system end module and a storage module. The visualization module is used to query the data during the training process and present it in the form of charts.

[0090] The system end module is used to coordinate and integrate the operation and data transmission between the modules in the middle and long distance running training management and data analysis system; The storage module is used to store the operation data between the modules in the middle and long distance running training management and data analysis system.

[0091] Optionally, the management module is further configured to send the adjusted information of the preset training course to the corresponding target athlete.

[0092] Specifically, the evaluation module compares and analyzes actual heart rate data with predicted heart rate data, calculating the degree of difference between the two, such as the mean squared error (MSE) and mean absolute error (MAE), to generate evaluation data. This data is used to measure the performance of the training session. A smaller evaluation data indicates that the training effect is closer to the expected result.

[0093] In some embodiments, the training performance percentage k is calculated using the real-time heart rate data and the corresponding predicted heart rate data by the following formula: ;in, is the heart rate prediction data, For real-time heart rate data, i is the time node, and N is the total time nodes.

[0094] By comparing actual heart rate data with predicted heart rate data, the performance of this training can be accurately evaluated, helping coaches and athletes understand whether the training effect has met expectations. For example, corresponding thresholds can be set based on historical data, and the calculated k value can be compared with the corresponding threshold to determine whether the training task needs to be modified.

[0095] Training load is also calculated based on actual heart rate data, combined with factors such as exercise time and intensity. Training load quantitatively assesses the physical strain on a target athlete during training. Calculating training load provides coaches and athletes with quantitative metrics, helping them better understand the impact of training on the body and avoid overtraining or undertraining.

[0096] Training load is calculated as follows: ; Among them, A, a, B, b are the set feature parameters, T is the total time of one training, , is the normalized data of heart rate, HR is the instantaneous heart rate (actual heart rate data or predicted heart rate data can be used), is the resting heart rate, is the maximum heart rate, , is the normalized data of velocity, V is the instantaneous velocity, V th is the threshold speed, V c is the characteristic speed.

[0097] The visualization module provides a user interface for easily accessing various training data, including motion data (speed, distance, time, etc.), physiological indicators (heart rate, HRV, etc.), and assessment data (training load, training performance evaluation results, etc.). This data is presented in intuitive charts (such as line graphs, bar graphs, and pie charts), allowing users to clearly understand the training process and results. The visualization module facilitates quick access to various training data, improving data accessibility and practicality. The intuitive charts make complex training data easy to understand, helping users quickly grasp key training information.

[0098] The system-side module serves as the hub of the entire system, responsible for coordinating and integrating the operation and data transmission between modules. It ensures seamless collaboration between the acquisition module, processing module, management module, evaluation module, visualization module, and storage module, enabling efficient data flow and sharing, and ensuring stable system operation. In other words, the system-side module ensures efficient collaboration and data transmission between modules, improving the overall operating efficiency and stability of the system, and reducing communication delays and data loss between modules. The storage module stores operational data generated by each module in the system, including raw motion data, processed segmentation results, training programs adjusted by the management module, evaluation data generated by the evaluation module, and charts generated by the visualization module. It provides secure and reliable data storage and management, supporting rapid data query, update, and backup.

[0099] The management module expands its functionality by sending adjusted pre-set training program information to the targeted athletes. Through coordination with the system-side modules, this adjusted training program information is transmitted to the athletes' mobile devices or training terminals, ensuring that athletes receive the latest training plans in a timely manner. Furthermore, the system's automated feedback mechanism reduces communication costs between coaches and athletes, improving training management efficiency.

[0100] Optionally, the middle- and long-distance running training management and data analysis system further includes a coach terminal, which can modify the training courses in the management module and view various sports data of the target athlete at any time.

[0101] Specifically, coaches can access the management module through the coach terminal and modify the training course according to the actual situation of the athlete (such as physical condition, training progress, competition requirements, etc.). For example, they can adjust the training distance, time, intensity, add or remove specific training phases, etc. The coach terminal also provides an interface that allows coaches to view the target athlete's various sports data at any time, including sports data (speed, distance, time, etc.), physiological indicators (heart rate, HRV, etc.), and evaluation data (training load, training performance evaluation results, etc.). When the coach modifies the training course on the coach terminal, the system-side module is responsible for synchronizing the modified training course information to the management module and updating the relevant data. When the coach views the athlete's sports data, the coach terminal obtains the latest data from the system in real time to ensure that the coach sees the athlete's current sports status and historical data.

[0102] At the same time, coaches can provide feedback to athletes through the coaching terminal based on the data they view, such as training suggestions and adjustment plans. Athletes can also view the coach's feedback on their own devices and train according to the new training course, forming a good training management closed loop.

[0103] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A middle and long distance running training management and data analysis system, characterized in that: include: A collection module, used to collect sports information of target athletes, wherein the sports information includes sports data and physiological indicators; A processing module, configured to perform segmentation processing on the motion data based on a preset segmentation algorithm to obtain a segmentation result; The management module is used to match the segmentation results and the physiological indicators with the preset training courses to obtain matching results, and adjust the corresponding preset training courses according to the matching results.

2. The middle and long distance running training management and data analysis system according to claim 1, characterized in that: The processing module is specifically configured to include: performing smoothing processing on the motion data to obtain temporary motion data; Determining the corresponding preset segmentation algorithm based on the temporary motion data according to a preset determination method; The motion data is segmented according to the corresponding preset segmentation algorithm to obtain the segmentation result.

3. The middle and long distance running training management and data analysis system according to claim 2, characterized in that: The motion data includes a speed change sequence, and the preset determination method includes a first determination method; The step of determining the corresponding preset segmentation algorithm based on the temporary motion data according to a preset determination method includes: When the first discrimination method is adopted, the speed change sequence is discriminated based on a preset classifier to obtain a discrimination result, and the matching preset segmentation algorithm is determined according to the discrimination result; The speed change sequence is segmented according to the corresponding preset segmentation algorithm to obtain the segmentation result.

4. The middle and long distance running training management and data analysis system according to claim 3, characterized in that: The preset segmentation algorithm includes a Bayesian online change point detection algorithm, and the speed change sequence is segmented according to the corresponding preset segmentation algorithm to obtain a segmentation result, including: determining an initial mean and an initial variance, and calculating a posterior probability distribution of a speed run length in the temporary motion data based on the initial mean, the initial variance, and the speed change sequence; According to the comparison between the difference between adjacent data in the posterior probability distribution and a preset threshold, the data points whose difference is greater than the preset threshold are used as segmentation points, and the segmentation result is determined based on all the segmentation points.

5. The middle and long distance running training management and data analysis system according to claim 4, characterized in that: The preset segmentation algorithm further includes a preset threshold discrimination algorithm, and the segmentation processing of the speed change sequence according to the corresponding preset segmentation algorithm to obtain the segmentation result further includes: Get the preset high-speed threshold and low-speed threshold; The speed change sequence is segmented using the preset high-speed threshold and the preset low-speed threshold to obtain the segmentation result.

6. The middle and long distance running training management and data analysis system according to claim 3, characterized in that: The preset determination method further includes a second determination method, and the determination of the corresponding preset segmentation algorithm based on the temporary motion data according to the preset determination method further includes: When the second discrimination method is adopted, data calculation is performed on the speed change sequence to obtain a calculation result, which includes a mean, a median and a standard deviation; A corresponding distribution form is determined according to the speed change sequence and the calculation result, and the matching preset segmentation algorithm is determined according to the distribution form.

7. The middle and long distance running training management and data analysis system according to claim 2, characterized in that: The sports information sports environment data and basic information; the middle and long distance running training management and data analysis system also includes a load prediction module, The load prediction module is used to determine the target motion data set for the target athlete in the preset training course, and obtain corresponding heart rate prediction data based on a preset load prediction model according to the target motion data, the sports environment data and the basic information; wherein, the preset load prediction model is constructed based on a long short-term memory network.

8. The middle and long distance running training management and data analysis system according to claim 7, characterized in that: The physiological indicators include actual heart rate data; the middle and long distance running training management and data analysis system also includes an evaluation module, The evaluation module is used to evaluate the actual heart rate data and the corresponding predicted heart rate data to obtain corresponding evaluation data, and the evaluation data is used to evaluate the performance of this training; The evaluation module is further configured to calculate a training load based on the actual heart rate data, and the training load is used to evaluate the burden on the target athlete's body.

9. The middle and long distance running training management and data analysis system according to claim 7, characterized in that: The middle and long distance running training management and data analysis system also includes a visualization module, a system end module and a storage module. The visualization module is used to query the data during the training process and present it in the form of charts; The system end module is used to coordinate and integrate the operation and data transmission between the modules in the middle and long distance running training management and data analysis system; The storage module is used to store the operation data between the modules in the middle and long distance running training management and data analysis system.

10. The middle and long distance running training management and data analysis system according to claim 1, characterized in that: The management module is further configured to send the adjusted information of the preset training course to the corresponding target athlete.

Images