A cycling platform data collection method and a cycling platform parameter adjustment method

Abstract

This application relates to a method for collecting data from a cycling trainer, belonging to the field of fitness equipment technology. The method includes: in a detection mode, acquiring operational information when a user uses the cycling trainer, including wheel speed; determining, based on the operational information and a computational model, whether the user should stop cycling; determining whether the user's cycling time exceeds a preset time; and / or whether the user's cycling power exceeds a preset power; and / or whether the wheel speed is within a preset range; if none of the above conditions are met, the detection steps are repeated; if the conditions are met, the system enters a storage mode. In the storage mode, multiple storage areas are set up, and according to the different operating conditions of the cycling trainer, each operating condition corresponds to a storage area, and the operational data under that condition is stored in the corresponding storage area. This application can automatically complete multi-condition data collection during the use of the cycling trainer, reducing the burden of manual operation and improving data collection efficiency and automation.

Description

Technical Field

[0001] This application relates to the technical field of fitness equipment, and more particularly to a method for collecting data from a cycling trainer. Background Technology

[0002] To ensure the accuracy of power data, existing fitness equipment such as smart cycling trainers generally require a despinning calibration procedure. By measuring the device's natural deceleration characteristics under no-load conditions, power measurement deviations caused by factors such as transmission system friction, belt errors, temperature drift, and component aging are accurately calculated and corrected, thereby ensuring the reliability of training data.

[0003] In existing technologies, this process largely relies on manual triggering by the user. The user must actively enter the calibration process within the software or device and accelerate to a specified speed before stopping. However, the entire calibration process heavily depends on the user's active cooperation and precise operation. If the user fails to reach the target speed, stops too early, or restarts the process midway, calibration failure or timeout is likely. This high sensitivity to manual operation not only increases the user's burden but also affects the calibration success rate and the accuracy of device power calculations. Summary of the Invention

[0004] This application addresses, to at least some extent, one of the technical problems in the related art.

[0005] Therefore, this application aims to provide a method for collecting data from a cycling trainer. By combining detection and storage modes, the method collects data from the cycling trainer under different operating conditions during user use, thereby eliminating the need for user cooperation in data collection and improving user experience.

[0006] To achieve the above objectives, this application provides a method for collecting data from a cycling trainer, the steps of which include: In detection mode, when a user is using the cycling trainer, the system acquires the trainer's operating information, including wheel speed. Based on this information, a calculation model is used to determine whether the user should stop cycling and whether the user's cycling time has exceeded a preset time. And / or, whether the user's riding power exceeds the preset power; And / or, whether the wheel speed is within the preset range; If not, repeat the above steps; if yes, enter storage mode. In storage mode, multiple storage areas are set up so that each storage area corresponds to a specific operating condition of the cycling train, and the data information of the cycling train under that operating condition is stored in the corresponding storage area.

[0007] In this technical solution, a detection mode is used to avoid relying on users to manually trigger the calibration process, and the data collection start conditions are set within the device. During the natural process of using the cycling trainer, data from multiple operating conditions is automatically collected, reducing the burden of manual operation, improving the user experience, and increasing the automation and coverage of data collection, thus laying a data foundation for subsequent accurate calibration.

[0008] In some embodiments of this application, during the storage mode, the system continuously determines whether the user has stopped riding based on the operation information through a calculation model. When it is determined that the user is riding, the system re-enters the detection mode.

[0009] In the technical solution, during data collection, the user may resume pedaling midway, which disrupts the physical decay characteristics expected under no-load gliding conditions. If data continues to be recorded in the same storage area, data from non-gliding states will be mixed in, thus contaminating the validity of the collected data. Continuously monitoring the riding status during storage mode ensures that the data stored in each storage area accurately reflects the pure gliding decay characteristics under the corresponding operating conditions, improving the accuracy and reliability of data collection and avoiding the problem of data loss due to user mid-journey pedaling requiring re-collection.

[0010] In some embodiments of this application, during the storage mode process, it is determined whether the storage area stores data information. If so, a scoring function is used to score each storage area based on the data information, the operating condition of the cycling train is adjusted to the operating condition corresponding to the storage area with the lowest score, and the data information of the cycling train operating under that operating condition is stored in the corresponding storage area.

[0011] In the technical solution, during long-term use, the data richness and timeliness of storage areas corresponding to different operating conditions may be uneven due to differences in users' riding habits. Some operating conditions may lack effective data for a long time, or existing data may become invalid due to the passage of time. By introducing a scoring function to quantitatively evaluate the status of each storage area, the relative value or degree of data loss in each storage area can be comprehensively reflected. Without user intervention, the data completeness and timeliness of each operating condition can be balanced, thereby improving the completeness and usability of data information and providing more accurate data support for subsequent calibration based on this data system.

[0012] In some embodiments of this application, during the storage mode, when adjusting the operating conditions of the cycling train, a neglect period is set, and during the neglect period, the calculation model is stopped from determining whether the user has stopped cycling. Alternatively, improve the criteria for the computational model to determine if a user is riding a bicycle.

[0013] In the technical solution, when the method adjusts the operating conditions of the cycling train to collect data under specific operating conditions, this adjustment itself affects the wheel speed. This effect is similar to the user's active pedaling or stopping behavior, which can easily trigger misjudgments in the computational model. By setting an ignore duration, the sensitivity to triggering the exit from storage mode can be temporarily shielded or increased, preventing the method's own control actions from interfering with the data collection process. This enhances the stability and anti-interference capability of the data collection method, ensuring a smooth transition to a stable data collection phase after operating condition adjustments, and improving the success rate of data collection.

[0014] In some embodiments of this application, the scoring function calculates the score of each storage area based on the time freshness, interval integrity, sampling score and data variance in the data information. The time freshness is calculated by a time function, and the time freshness is inversely proportional to the time since the last storage of information in the storage area.

[0015] In this technical solution, a single-dimensional evaluation metric cannot fully reflect the true usable value of the data stored in a storage area. By using four dimensions for scoring, the method can more accurately determine the priority of storage areas, improving the overall efficiency and effectiveness of data maintenance. Over time, physical factors such as wear and tear on the internal mechanical structure and lubrication conditions gradually alter the response characteristics of the cycling train under different operating conditions. Therefore, the older the data, the lower its reference value for guiding current calibration. A monotonically decreasing time function is used as the conversion rule to ensure that the stored data has good overall timeliness.

[0016] In some embodiments of this application, during the detection process, the acquired operating information also includes the angular acceleration of the cycling platform. When the calculation model makes a judgment, if the angular acceleration is not less than zero; And / or, when the difference between the angular acceleration at this moment and the angular acceleration at the previous moment exceeds the preset angular acceleration amplitude; And / or, when the change in angular acceleration per unit time exceeds the preset angular acceleration, the calculation model determines that the user is riding a bicycle.

[0017] In this technical solution, when a user actively pedals the trainer, energy is continuously input into the wheel system, at which point the angular acceleration of the wheels is necessarily greater than or equal to zero. Furthermore, the force exerted by the user's pedaling action has a periodic discontinuity, resulting in a significantly higher instantaneous rate of change of angular acceleration than the natural decay under pure inertial gliding. Therefore, by monitoring the absolute value and time-varying rate of angular acceleration and comparing the monitored values ​​with preset angular acceleration amplitude or preset angular acceleration, it is possible to distinguish between actively driven and undriven free gliding states. This improves the sensitivity and accuracy of identifying the user's true riding intentions under complex riding postures such as weak pedaling and intermittent pedaling, preventing invalid data from being incorrectly included in the collection process due to missed user pedaling behavior, thereby ensuring the reliability of data collection trigger conditions.

[0018] In some embodiments of this application, during the detection mode, the acquired operating information also includes the resistance value of the cycling platform, and the preset angular acceleration is calculated based on the wheel speed and resistance value using an angular acceleration function.

[0019] In the technical solution, the load torque borne by the wheel system of the cycling train varies under different resistance settings. Under higher resistance, even without user pedaling, angular acceleration decreases rapidly due to the resistance torque; conversely, under low resistance, the decay of angular acceleration is very gradual. Using a fixed angular acceleration threshold cannot adapt to working environments with significant resistance variations, easily misjudging gliding as pedaling under high resistance or pedaling as gliding under low resistance. Dynamically generating a judgment threshold matching the current working condition using an angular acceleration function ensures consistent recognition accuracy across the entire working range from zero resistance to high resistance, increasing the adaptability of the data collection method to complex cycling environments.

[0020] In some embodiments of this application, when the calculation model makes a judgment, if the actual decrease in wheel speed within a preset time exceeds a preset decrease value; And / or, if the actual wheel speed is lower than the preset speed, the calculation model will determine that the user should stop riding.

[0021] In the technical solution, when the user stops inputting energy to the trainer, the wheel system is only subjected to resistance load, and its rotational speed will inevitably experience a clear and measurable cumulative decrease within a preset time. Furthermore, regardless of the initial speed, when the wheel system decelerates to an extremely low speed range, its remaining kinetic energy is insufficient to maintain an effective riding posture, which is objectively equivalent to the cessation of riding. By comparing the physical parameters corresponding to the above two phenomena, a criterion for determining whether the user has stopped riding is provided. This method allows the user to confirm that they have stopped riding based on rotational speed even when the angular acceleration signal is interfered with, ensuring the accuracy of subsequent storage mode triggering.

[0022] In addition, this application also proposes a method for adjusting the parameters of a cycling train, which adjusts the parameters of the cycling train based on the data information collected by the above-mentioned cycling train data collection method, and adjusts the torque parameters of the cycling train under the corresponding working conditions by means of a torque function based on the torque stored in the storage area.

[0023] In this technical solution, the torque data stored in each storage area during the data collection phase directly reflects the combined effect of all resistance sources, including mechanical friction, electromagnetic resistance, and transmission losses. When parameter adjustments are needed, the torque data from the corresponding storage area is directly used as the input variable. The data is then processed through a preset torque function, enabling the riding platform to be precisely calibrated based on data collected during actual use. This effectively eliminates deviations caused by individual differences, assembly tolerances, and other factors, improving the consistency of parameter adjustments under different operating conditions.

[0024] In some embodiments of this application, when no torque is stored in the storage area corresponding to a certain working condition, the simulated torque under the working condition corresponding to the storage area without stored torque is calculated by using the torque already stored in other storage areas and the corresponding working conditions, and the torque parameters under the corresponding working condition are adjusted by using a torque function based on the simulated torque.

[0025] In the technical solution, due to the unevenness of users' riding habits, some operating conditions may lack sufficient data samples for a long time, resulting in a gap in the corresponding storage area. If the torque parameters under such operating conditions cannot be adjusted, the overall calibration system will have blind spots, affecting the parameter consistency of the equipment across the entire operating range. A calculation model is constructed using the mapping relationship between the operating conditions and torque of the stored data. This model reasonably estimates the torque under unsampled operating conditions, thereby supplementing the missing data without relying on additional user cooperation. This allows the cycling trainer to effectively adjust for various operating conditions even when data is incomplete, ensuring that the equipment obtains relatively accurate parameter configurations under different usage modes.

[0026] As can be seen from the above technical solutions, additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0027] Figure 1 This is a flowchart of a cycling trainer data collection method and a cycling trainer parameter adjustment method according to an embodiment of this application; Figure 2 This is a flowchart of a cycling train data collection method and a cycling train parameter adjustment method according to an embodiment of this application. Detailed Implementation

[0028] In the description of this application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0029] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection, an electrical connection, or a connection that allows communication between components; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication between two components or the interaction between two components, unless otherwise expressly limited. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0030] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "over," and "on top" of the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0031] In this application, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0032] The present application will now be described in detail through exemplary embodiments. However, it should be understood that, without further description, elements, structures, and features in one embodiment may be advantageously incorporated into other embodiments.

[0033] In the following, embodiments of this application will be described in detail with reference to the accompanying drawings.

[0034] Please refer to all the accompanying drawings. In one illustrative embodiment of the cycling trainer data collection method of this application, the cycling trainer data collection method includes: In detection mode, when a user uses the cycling trainer, the system acquires the trainer's operating information, including wheel speed. Based on this information, a calculation model is used to determine whether the user should stop cycling. If all conditions are met—that the user has stopped cycling, the user's cycling time has exceeded a preset time, the user's cycling power has exceeded a preset power, and the wheel speed is within a preset range—the system enters storage mode. Otherwise, the above steps are repeated.

[0035] In another embodiment of this application, the detection mode acquires the operating information of the cycling trainer when the user uses it. The operating information includes the wheel speed. Based on the operating information, the system uses a calculation model to determine whether the user should stop cycling. When the user stops cycling and at least one of the following conditions is met: the user's cycling time exceeds a preset time, the user's cycling power exceeds a preset power, or the wheel speed is within a preset range, the system enters the storage mode. Otherwise, the above steps are repeated.

[0036] By employing a detection mode, the calibration process is avoided by relying on manual user intervention, and the data collection initiation conditions are set within the device. During the use of the cycling trainer, multi-condition data collection is automatically completed, reducing the burden of manual operation, improving user experience, and enhancing the automation level of data collection, thus laying a data foundation for subsequent accurate calibration.

[0037] Furthermore, the preset time, preset power, and preset range can be set through user settings, factory settings, or cloud transmission.

[0038] In this embodiment, the preset time is 10 minutes, the preset power is 100W, and the preset range is 1200rpm to 2000rpm. This setting can prevent the method from intervening too early before the user has entered a stable riding state, thus improving the accuracy of the collected data.

[0039] In some embodiments, the difference between the upper and lower limits of the preset range is greater than the preset bandwidth. By collecting data within the preset range, torque in the undetected operating range can be simulated and predicted. However, since the total torque of the riding train varies non-linearly at different speeds, if the preset range is set too narrow, the collected data can only reflect the torque performance within that narrow range, making it difficult to construct data changes covering the entire speed range, resulting in inaccurate predicted data and affecting subsequent parameter adjustments.

[0040] Furthermore, the preset bandwidth can be set through user settings, factory settings, or cloud transmission.

[0041] In this embodiment, the preset bandwidth is 700 rpm. By requiring the speed bandwidth to be greater than 700 rpm, the data collection can capture data information over a wide range from high speed to low speed. This provides reasonably distributed sample points for subsequent calculation of torque values ​​at unsampled speed nodes, improving the accuracy of subsequent prediction data and providing data support for parameter adjustment of the riding train.

[0042] In some embodiments, in the storage mode, multiple storage areas are set up so that each storage area corresponds one-to-one with the operating conditions of the cycling trainer, and the data information of the cycling trainer operating under the corresponding operating conditions is stored in the corresponding storage area. The torque values ​​of the cycling trainer under different operating conditions exhibit non-linear differences. By classifying each set of data into an independent storage partition according to its operating condition, mutual interference between data under different operating conditions can be avoided. When subsequent parameter adjustments or calibration calculations are required, the method can extract data samples under that operating condition from the corresponding storage area, improving the accuracy of data retrieval.

[0043] In some embodiments, during the storage mode, the system continuously determines whether the user has stopped riding based on operational information using a computational model. If the system determines that the user is riding, it re-enters the detection mode. During data collection, the user may resume pedaling midway, which disrupts the physical decay characteristics expected under no-load gliding conditions. If data continues to be recorded in the same storage area, data from non-gliding states will be mixed in, thus contaminating the validity of the collected data. Maintaining continuous monitoring of the riding status during the storage mode ensures that the data stored in each storage area accurately reflects the pure gliding decay characteristics under the corresponding operating conditions, improving the accuracy and reliability of data collection and avoiding the problem of data loss due to user mid-ride pedaling requiring re-collection.

[0044] In some embodiments, during the storage mode process, it is determined whether the storage area contains data. If so, a scoring function is used to score each storage area based on the data. Determining whether a storage area contains data before scoring prevents situations where a user is using the cycling trainer for the first time and the storage area has not saved data, thus preventing scoring. Afterwards, the cycling trainer's operating condition is adjusted to correspond to the storage area with the lowest score, and the data from the cycling train's operation under that condition is stored in the corresponding storage area.

[0045] During long-term use, the data richness and timeliness of storage areas corresponding to different operating conditions may vary due to differences in user riding habits. Some operating conditions may lack effective data for a long time, or existing data may become invalid due to the passage of time. By introducing a scoring function to quantitatively evaluate the status of each storage area, the relative value or degree of data loss in each storage area can be comprehensively reflected. Without user intervention, the data completeness and timeliness of each operating condition can be balanced, thereby improving the completeness and usability of data information and providing more accurate data support for subsequent calibration based on this data system.

[0046] In some embodiments, when multiple storage areas have the same score and are all the lowest scores, the operating conditions of the cycling train are adjusted to correspond to the operating conditions of one of the random storage areas, and subsequent data information is collected and stored. This promotes the data collection process in a simple way and avoids the method from stalling due to indecisiveness.

[0047] In some embodiments, the method corresponds the operating condition of the cycling train to the storage zone based on the different ranges of the cycling train's output current. When the cycling train is adjusted to the operating condition corresponding to the lowest-scoring storage zone, the output current of the cycling train is controlled within the current value range corresponding to that storage zone, thereby achieving the purpose of adjusting the operating condition.

[0048] The resistance output of a cycling trainer is essentially determined by the electromagnetic control current. Different current values ​​directly correspond to different load torques. Therefore, using the current range as the physical medium for dividing and controlling operating conditions can transform abstract data storage requirements into specific control commands. This allows the method to accurately drive the cycling trainer into the target operating condition based on the scoring results, ensuring the feasibility of collecting data from multiple operating conditions.

[0049] In this embodiment, the current is divided into zones based on the maximum current from 0% to 100%, with each 5% increment corresponding to a different current level. For example, when the maximum current is 4A (amperes), the operating conditions corresponding to the storage zone are 0 to 200mA (milliamperes), 200mA to 400mA, 400mA to 600mA, 600mA to 800mA, and so on. Regardless of whether the user's actual riding habits are biased towards high or low resistance areas, the method can gradually supplement the data samples of each zone through zone scheduling, providing data support for subsequent adjustments.

[0050] It is understandable that the method can also classify the operating conditions of the cycling train based on conditions such as wheel speed and resistance value. This embodiment only uses the current condition as an example for illustration.

[0051] In some embodiments, during the storage mode process, when adjusting the operating conditions of the cycling trainer, a neglect period is set, and during the neglect period, the determination of whether the user has stopped cycling is stopped by the calculation model.

[0052] In another embodiment of this application, during the storage mode, when adjusting the operating conditions of the cycling train, a neglect period is set, and during the neglect period, the standard for the calculation model to determine that the user is currently cycling is improved.

[0053] When a method adjusts the cycling trainer's operating conditions to collect data under specific conditions, this adjustment itself affects the wheel speed. This effect is similar to the user's active pedaling or stopping behavior, which can easily trigger misjudgments in the computational model. By setting an ignore duration, the sensitivity to triggering exit from storage mode can be temporarily disabled or increased, preventing the method's own control actions from interfering with the data collection process. This enhances the stability and anti-interference capability of the data collection method, ensuring a smooth transition to a stable data collection phase after condition adjustments, and improving the success rate of data collection.

[0054] Furthermore, the duration to be ignored can be set through user settings, factory settings, or cloud transmission.

[0055] In some embodiments, the scoring function calculates a score for each storage area based on the time freshness, interval completeness, sampling score, and data variance of the data information. Time freshness represents the distance between the current time and the last time information was stored in the storage area, and the score of the storage area is positively correlated with time freshness. Interval completeness ranges from 0 to 1 and represents the proportion of torque data in the storage area to the total torque data in all storage areas. The score of the storage area is positively correlated with interval completeness. Sampling score represents the number of times data information has been stored in the storage area, and the score of the storage area is positively correlated with sampling score. Data variance represents the dispersion of the data information stored in the storage area, and the score of the storage area is negatively correlated with data variance.

[0056] A single-dimensional evaluation metric cannot fully reflect the true usable value of the data stored in a storage area. By scoring through four dimensions, the scoring results can comprehensively depict the status of each storage area from four aspects: timeliness, completeness, reliability, and stability. This makes the method more accurate in judging the priority of storage areas and improves the efficiency and effectiveness of overall data maintenance.

[0057] In some embodiments, the sampling score is calculated using a step function or a logarithmic function. When the number of times data information has been stored in the storage area exceeds a threshold, the sampling score approaches the maximum score. This avoids a situation where the score grows indefinitely due to excessive sampling.

[0058] Step functions and logarithmic functions are mathematically two types of functions with saturation properties. A step function's output jumps after the input reaches a certain threshold (e.g., from 0 to 1), while a logarithmic function's growth rate decreases as the input increases, ultimately limiting the output to a finite range. Using these functions for calculation avoids the unlimited growth of scores caused by repeated sampling of a particular work condition. This allows the scoring function to balance considerations such as time freshness, interval completeness, and data variance, prioritizing the collection of data from truly outdated or missing work conditions, rather than repeatedly collecting data from conditions with sufficient existing data.

[0059] Understandably, the number of attempts threshold can be set according to the actual situation.

[0060] In some embodiments, time freshness is calculated using a time function, which is inversely proportional to the time elapsed since the last stored information. Over time, physical factors such as wear and tear on the internal mechanical structure and lubrication conditions gradually alter the response characteristics of the cycling train under different operating conditions. Therefore, older data has lower reference value for guiding current calibration. A data aging evaluation mechanism is established using a monotonically decreasing time function as the transformation rule. This allows the data collection method to automatically prioritize operating conditions that have not been updated for a long time, effectively combating the negative impact of equipment time-varying characteristics on data accuracy and ensuring the overall timeliness of the stored data.

[0061] Furthermore, the time function is a normalization function that decays as the storage area increases from the last stored information. The normalization function is a mathematical transformation rule that maps the original value to a unified standard interval. Through the normalization function, the storage area from the last stored information can be mapped to a specified interval (e.g., between 0 and 1). The normalization function can convert the duration data into a time freshness that is easy to compare, which is convenient for the scoring function to perform scoring calculations.

[0062] In this embodiment, the time function calculates the time freshness using an exponential decay function. The shorter the time since the last stored information in the storage area, the closer the time freshness is to 1. The longer the time since the last stored information in the storage area, the closer the time freshness is to 0.

[0063] The time function formula in this embodiment is: f(T diff )=e (-λt) Where, f(T) diff ) represents the time function, e is the natural constant, λ is the decay coefficient, and t is the time elapsed since the last stored information in the storage area.

[0064] Understandably, the attenuation coefficient can be set according to the actual situation.

[0065] In some embodiments, the scoring function calculates scores by weighting time freshness, interval integrity, sampling score, and data variance. By reasonably configuring the weights of the conditions, the scoring function can accurately identify the operating condition partitions that most need updating, thereby collecting data and providing data support for subsequent power calibration.

[0066] In this embodiment, the formula for the scoring function is: S=W1×T diff +W2×C+W3×N-W4×V Where S represents the scoring function, T diff The time freshness is calculated using the time function, C represents the interval completeness, N represents the sampling score calculated using the step function or logarithmic function, V represents the data variance, and W1, W2, W3, and W4 are preset weighting coefficients.

[0067] That is, S (scoring function) = W1 (first weighting coefficient) × T diff (Time freshness) + W2 (second weight coefficient) × C (interval completeness) + W3 (third weight coefficient) × N (sampling score) - W4 (fourth weight coefficient) × V (data variance).

[0068] Understandably, W1, W2, W3, and W4 can be configured according to the actual situation.

[0069] Preferably, W1 (first weight coefficient) + W2 (second weight coefficient) + W3 (third weight coefficient) = 1, which limits the values ​​of the three weights to a fixed total amount. This makes the output of the scoring function more stable and controllable, and has good portability and comparability between different devices or different stages of use. It is also convenient to make targeted adjustments to W1, W2, and W3 according to the actual effect without worrying about the overall deviation.

[0070] In some embodiments, during the detection mode, the acquired operating information also includes the angular acceleration of the cycling train. When the calculation model makes a judgment, if at least one of the following conditions is met: the angular acceleration is not less than zero, the difference between the angular acceleration at this moment and the angular acceleration at the previous moment exceeds the preset angular acceleration amplitude, or the change in angular acceleration per unit time exceeds the preset angular acceleration, the calculation model determines that the user is currently cycling.

[0071] In another embodiment of this application, during the detection mode, the acquired operating information also includes the angular acceleration of the cycling platform. When the calculation model makes a judgment, if the following conditions are met: the angular acceleration is not less than zero, the difference between the angular acceleration at this moment and the angular acceleration at the previous moment exceeds the preset angular acceleration amplitude, and the change in angular acceleration per unit time exceeds all of the preset angular acceleration conditions, the calculation model determines that the user is cycling at this moment.

[0072] When a user actively pedals on the trainer, energy is continuously input into the wheels, at which point the wheel's angular acceleration is necessarily greater than or equal to zero. Furthermore, the force exerted by the user's pedaling action has a periodic discontinuity, resulting in a significantly higher instantaneous rate of change of angular acceleration than the natural decay during pure inertial gliding. Therefore, by monitoring the absolute value and time-varying rate of angular acceleration and comparing the monitored values ​​with preset angular acceleration amplitudes or preset angular acceleration values, it is possible to distinguish between actively driven and undriven free-gliding states. This improves the sensitivity and accuracy of identifying the user's true riding intentions under complex riding postures such as weak pedaling and intermittent pedaling, preventing invalid data from being incorrectly included in the collection process due to missed user pedaling behavior, thereby ensuring the reliability of data collection trigger conditions.

[0073] It is understandable that angular acceleration can be the angular acceleration of the damping wheel of the trainer, or the angular acceleration of other components of the trainer that move through the user's control.

[0074] Furthermore, the preset angular acceleration amplitude can be set through user settings, factory settings, or cloud transmission.

[0075] In some embodiments, during the detection mode, the acquired operating information also includes the resistance value of the cycling train, and the preset angular acceleration is calculated based on the wheel speed and resistance value using an angular acceleration function.

[0076] The load torque on the wheels of a cycling trainer varies depending on the resistance setting. At higher resistance, even without pedaling, angular acceleration decreases rapidly due to the resistance torque; conversely, at lower resistance, the decay of angular acceleration is much more gradual. Using a fixed angular acceleration threshold cannot adapt to environments with significant resistance variations, easily misinterpreting gliding as pedaling at high resistance or pedaling as gliding at low resistance. Dynamically generating a judgment threshold that matches the current operating conditions using an angular acceleration function ensures consistent recognition accuracy across the entire operating range from zero to high resistance, increasing the adaptability of the data collection method to complex cycling environments.

[0077] In this embodiment, the preset angular acceleration is equal to 0.1 plus (3 multiplied by the current resistance value plus 300) divided by (the current wheel speed divided by 2 plus 100).

[0078] The formula for the angular acceleration function is: limit=0.1+(3×resistance+300) / (rpm / 2+100) Where limit is the calculated preset angular acceleration, resistance is the current resistance value, and rpm is the current wheel speed.

[0079] That is, limit (preset angular acceleration) = 0.1 + [3 × resistance (current resistance value) + 300] / [rpm (current wheel speed) / 2 + 100] By comparing the value calculated by the above formula with the current angular acceleration, it can be determined whether the user is riding under the current resistance value. Combined with a fixed preset angular acceleration amplitude, a comprehensive judgment is made, thereby increasing the accuracy of the judgment.

[0080] In some embodiments, during the neglect period, the preset angular acceleration is set to a fixed threshold, which is greater than the value calculated by the angular acceleration function. By setting the preset angular acceleration to a large fixed value, the standard for the calculation model to make a judgment on the user pausing riding is improved, thereby reducing the impact of the riding train's adjustment of operating conditions on the calculation model's judgment, preventing the method from jittering between the two states of the user riding and stopping riding, and improving the stability of the method.

[0081] In some embodiments, when the calculation model makes a judgment, if all the conditions are met, such as the actual decrease in wheel speed within a preset time period exceeding a preset decrease value and the actual wheel speed being lower than a preset speed, the calculation model determines that the user should stop riding at this time.

[0082] In another embodiment, when the calculation model makes a judgment, if at least one of the following conditions is met: the actual decrease in wheel speed within a preset time exceeds a preset decrease value, or the actual wheel speed is lower than a preset speed, the calculation model determines that the user should stop riding.

[0083] When the user stops inputting energy into the trainer, the wheel system is only subjected to resistance load, and its rotational speed will inevitably experience a clear and measurable cumulative decrease within a preset time. Furthermore, regardless of the initial speed, when the wheel system decelerates to an extremely low speed range, its remaining kinetic energy is insufficient to maintain an effective riding posture, which is objectively equivalent to the cessation of riding. By comparing the physical parameters corresponding to the above two phenomena, a criterion for determining whether the user has stopped riding is provided. This method allows the user to confirm that they have stopped riding based on rotational speed even when the angular acceleration signal is interfered with, ensuring the accuracy of subsequent storage mode triggering.

[0084] Furthermore, the preset descent value and preset speed can be set through user settings, factory settings, or cloud transmission.

[0085] In some embodiments, in the program of storage mode, the wheel speed and the time interval between adjacent samples under the corresponding working conditions are collected, the torque is calculated by the wheel speed and the time interval, and the wheel speed, time interval and torque are stored in the corresponding storage area. The stored wheel speed is stored with an accuracy of 0.1 rpm and the stored time interval is stored with an accuracy of 0.1 ms (milliseconds).

[0086] Furthermore, in this embodiment, the torque is calculated using the physical formula of multiplying the moment of inertia of the gear train by the change in angular velocity and dividing by the time interval. The change in angular velocity is equal to the change in rotational speed multiplied by 2, multiplied by π and divided by 60. The torque can be calculated based on the collected wheel rotational speed and time interval using the aforementioned known physical formula.

[0087] The calculation formula is: T total (Rk, rpm) = J × Δω / Δt Δω = Δrpm × 2π / 60 Among them, T total (Rk, rpm) represents the calculated torque, Rk represents the working condition zone corresponding to the currently calculated torque, J is the moment of inertia of the gear train, Δω is the change in angular velocity, Δt is the corresponding time interval, and Δrpm is the change in wheel speed.

[0088] That is, Ttotal(Rk, rpm) (torque in a certain working condition zone) = J (moment of inertia of the gear train) × Δω (change in angular velocity) / Δt (time interval) Δω (change in angular velocity) = Δrpm (change in wheel speed) × 2π / 60 It is understandable that the moment of inertia of the gear train can be adjusted according to the material and size of the wheels. In this embodiment, the moment of inertia of the gear train is set to 0.023735 for illustrative purposes.

[0089] The torque of electromagnetic resistance cycling equipment is essentially controlled by current. Physical wear of its internal transmission components (such as belts, chains, and rollers), resistance fluctuations caused by temperature changes, and installation gap misalignment caused by long-term vibration can all cause deviations between the actual resistance and the value displayed on the instrument. If not calibrated or physically adjusted regularly, users may feel that the resistance is lighter or uneven at the same gear, which not only affects the accuracy of training intensity but may also cause the motor controller to overheat and trigger protection.

[0090] Therefore, this application also proposes a method for adjusting the parameters of a cycling train, which adjusts the parameters of the cycling train based on the data information collected by the above-mentioned method for collecting cycling train data, and adjusts the torque parameters of the cycling train under the corresponding working conditions by means of a torque function based on the torque stored in the storage area.

[0091] The torque data stored in each storage area during the data collection phase directly reflects the combined effect of all resistance sources, including mechanical friction, electromagnetic resistance, and transmission losses. When parameter adjustments are needed, the torque data from the corresponding storage area is directly used as the input variable. The data is then processed through a preset torque function, enabling the riding train to be precisely calibrated based on data collected during actual use. This effectively eliminates deviations caused by individual differences, assembly tolerances, and other factors, improving the consistency of parameter adjustments under different operating conditions.

[0092] Furthermore, the torque function is: the adjusted torque equals (1 minus the learning rate) multiplied by the unadjusted torque plus the learning rate multiplied by the torque stored in the storage area.

[0093] The formula for the torque function is: T new (R k ,rpm)=(1-a)×T old (R k ,rpm)+a×T total (R k ,rpm) Among them, T new (R k (rpm) represents the adjusted torque, a represents the learning rate, and T represents the torque. old (R k (rpm) represents the unadjusted torque, T total (R k , rpm) is the torque stored in the storage area.

[0094] That is, T new(R k The adjusted torque (rpm) = [1 - a (learning rate)] × T old (R k (rpm) (unadjusted torque) + a (learning rate) × T total (R k (rpm) (torque stored in the storage area) Preferably, the learning rate is between 0.1 and 0.3 to achieve a balance between historical accumulation and current measurement, avoiding parameter mutations caused by single sampling fluctuations, thereby ensuring system stability. When the score of a storage area is lower than the score threshold, the value of 'a' is set to 1 in this torque adjustment. When the comprehensive score of a storage area is lower than the preset threshold, it indicates that the reliability of historical data under this working condition is seriously insufficient. Continuing to use a low learning rate will delay the introduction of effective information. Therefore, 'a' is temporarily set to 1 at this time, that is, the latest sample collected this time is fully trusted, and the working condition data that urgently needs to be updated can be incorporated into the calibration system as quickly as possible, thereby improving data repair efficiency and calibration response speed.

[0095] The scoring threshold can be set through user settings, factory settings, or cloud transmission.

[0096] In some embodiments, when no torque is stored in the storage area corresponding to a certain working condition, the simulated torque under the working condition corresponding to the storage area where no torque is stored is calculated by using the torque already stored in other storage areas and the corresponding working conditions, and the torque parameters under the corresponding working condition are adjusted by the torque function based on the simulated torque.

[0097] Due to the uneven distribution of users' riding habits, some operating conditions may lack sufficient data samples for extended periods, resulting in gaps in the corresponding data storage. If the torque parameters under these conditions cannot be adjusted, the overall calibration system will have blind spots, affecting the consistency of the equipment's parameters across the entire operating range. By utilizing the mapping relationship between operating conditions and torque from stored data, a calculation model is constructed. This model provides a reasonable estimate of the torque under unsampled conditions, thereby supplementing missing data without relying on additional user intervention. This allows the cycling trainer to effectively adjust for various operating conditions even with incomplete data, ensuring relatively accurate parameter configurations for the equipment under different usage modes.

[0098] Because the method can only store data under limited operating conditions in a single storage mode, it cannot collect data under all operating conditions. For example, in the storage mode, the score is lowest when the wheel speed is between 800 rpm and 1500 rpm. If the method adjusts the wheel speed of the trainer to 1500 rpm, it can only collect data when the wheel speed does not exceed 1500 rpm, and cannot collect data when the wheel speed exceeds 1500 rpm. This results in low efficiency in collecting data and is not conducive to subsequent parameter adjustments.

[0099] Therefore, in some embodiments, multiple storage areas correspond to different wheel speed ranges of the cycling platform under different resistances.

[0100] When the trainer is operating under a specific resistance condition, and the storage areas corresponding to the speeds of two adjacent wheels have stored torque, but the storage areas corresponding to the speeds between the two wheels do not have stored torque (for example, the storage area corresponding to 400rpm-800rpm has stored torque, the storage area corresponding to 800rpm-1200rpm does not have stored torque, and the storage area corresponding to 1200rpm-1600rpm has stored torque), then the simulated torque under this condition is calculated using cubic spline interpolation or linear interpolation, and the calculated simulated torque is stored in the storage area corresponding to this condition for subsequent parameter adjustments.

[0101] Cubic spline interpolation is a mathematical method that smoothly connects known data points by constructing a piecewise cubic polynomial curve. It ensures that the entire curve has continuous first and second derivatives at the nodes, resulting in a smooth and natural interpolation result. Linear interpolation, on the other hand, directly connects two adjacent known data points with a straight line to estimate the value at the intermediate point. Through cubic spline interpolation or linear interpolation, the simulated torque at intermediate speeds can be calculated. This allows for the refinement of the total torque calibration table across the entire speed range without additional user intervention, providing complete data support for subsequent parameter adjustments under various operating conditions.

[0102] Furthermore, when the wheel speed corresponding to the unstored torque storage area is not between the wheel speeds corresponding to the stored torque storage area under a specific resistance condition (for example, the storage area corresponding to 700rpm-1000rpm stores torque, while the storage area corresponding to 1700rpm-2000rpm does not store torque), the least squares method is used to perform polynomial curve fitting to construct an equation model between wheel speed and torque. The simulated torque under this condition is calculated through the equation model, and the calculated simulated torque is stored in the storage area corresponding to this condition for subsequent parameter adjustments.

[0103] Least squares polynomial curve fitting is a mathematical optimization technique that finds a polynomial curve that best reflects the trend of data changes by minimizing the sum of squares of the errors between the actual observed data points and the fitted curve. Under a specific resistance condition on a cycling trainer, if a single natural gliding motion can only collect torque samples within a limited speed range, then using these measured sample points, a second-order or higher-order polynomial model of torque variation with speed can be fitted using the least squares method. This allows for the calculation of torque values ​​in the unsampled low-speed or high-speed blind zones under that condition, thus completing the total torque calibration data across the entire speed range and providing a data foundation for accurate correction of subsequent power calculations.

[0104] Preferably, the formula used in the equation model is: T = a × rpm 2 +b×rpm+c Where T is the simulated torque, rpm is the wheel speed, and a, b, and c are unknown parameters. The measured torque and its corresponding wheel speed are substituted into the above formula to solve for the values ​​of a, b, and c. Then, the wheel speed under the condition corresponding to the storage area where the torque was not measured is substituted into the formula to calculate the simulated torque. The calculated simulated torque is stored in the storage area corresponding to the condition for subsequent parameter adjustments.

[0105] That is, T (simulated torque) = a (unknown parameter) × rpm (wheel speed) 2 +b (unknown parameter) × rpm (wheel speed) +c (unknown parameter) Furthermore, when no torque is stored in the storage area corresponding to multiple wheel speeds under a certain resistance or multiple resistances, the simulated torque at the same wheel speed under the resistance condition is calculated using bilinear interpolation or surface smoothing algorithms (such as gradient-based surface fitting) with torque data at the same wheel speed under the resistance condition that is close to the resistance condition. The calculated simulated torque is then stored in the storage area corresponding to the resistance condition for subsequent parameter adjustment.

[0106] Bilinear interpolation is a method for estimating intermediate values ​​on a two-dimensional grid by linearly weighting four adjacent known data points. Surface smoothing algorithms (such as gradient-based surface fitting) construct continuously differentiable mathematical surfaces (such as polynomial or spline surfaces) to smoothly approximate and interpolate discrete, sparse, or noisy data globally. Using the collected torque at adjacent resistance levels and adjacent speeds, simulated torque under missing operating conditions is derived through bilinear interpolation or surface fitting. This allows for the acquisition of complete torque data even without data collection, ensuring that reasonable torque parameters are available for power correction and resistance control across the entire operating range.

[0107] In some embodiments, when the method calculates the simulated torque, a monotonicity constraint check is performed, meaning that at the same wheel speed, the total torque must increase monotonically with the increase of resistance; and under the same resistance condition, the total torque curve cannot show abnormal "dip" or "bulge". If the simulated torque violates the monotonicity constraint, the simulated torque is discarded, and only the directly sampled torque is retained.

[0108] When extrapolating simulated torque under unsampled conditions, since torque in a real physical system inevitably increases with increasing resistance, and torque under the same resistance should exhibit a smooth and monotonically decreasing or increasing trend with wheel speed, interpolation and fitting algorithms may produce distorted curves that violate physical principles due to uneven sample distribution. By performing monotonicity constraint checks, unacceptable extrapolation results can be filtered out at the physical mechanism level, thereby preventing abnormal simulated torque from contaminating the collected data and improving the stability of the calibration method.

[0109] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for collecting data from a cycling trainer, characterized in that, The steps include: In detection mode, when a user is using the cycling trainer, the system acquires the trainer's operating information, including wheel speed. Based on this information, a calculation model is used to determine whether the user should stop cycling and whether the user's cycling time has exceeded a preset time. And / or, whether the user's riding power exceeds the preset power; And / or, whether the wheel speed is within the preset range; If not, repeat the above steps; if yes, enter storage mode. In storage mode, multiple storage areas are set up so that each storage area corresponds to a specific operating condition of the cycling train, and the data information of the cycling train under that operating condition is stored in the corresponding storage area.

2. The cycling trainer data collection method according to claim 1, characterized in that, During the storage mode, the system continuously uses operational information and a computational model to determine whether the user has stopped riding. If the system determines that the user is riding, it re-enters the detection mode.

3. The cycling trainer data collection method according to claim 2, characterized in that, During the storage process, it is determined whether the storage area contains data information. If so, the scoring function scores each storage area based on the data information, adjusts the operating condition of the cycling train to the operating condition corresponding to the storage area with the lowest score, and stores the data information of the cycling train under that operating condition into the corresponding storage area.

4. The cycling trainer data collection method according to claim 3, characterized in that, During the storage mode, when adjusting the operating conditions of the cycling trainer, a ignore period is set. During the ignore period, the calculation model is stopped from determining whether the user has stopped cycling. Alternatively, improve the criteria for the computational model to determine if a user is riding a bicycle.

5. The cycling trainer data collection method according to claim 3, characterized in that, The scoring function calculates the score of each storage area based on the time freshness, interval integrity, sampling score and data variance in the data information. The time freshness is calculated by a time function, and the time freshness is inversely proportional to the time since the last storage of information in the storage area.

6. The cycling trainer data collection method according to claim 1, characterized in that, During the detection process, the acquired operational information also includes the angular acceleration of the cycling train. When the calculation model makes a judgment, the angular acceleration is not less than zero. And / or, when the difference between the angular acceleration at this moment and the angular acceleration at the previous moment exceeds the preset angular acceleration amplitude; And / or, when the change in angular acceleration per unit time exceeds the preset angular acceleration, the calculation model determines that the user is riding a bicycle.

7. The cycling trainer data collection method according to claim 6, characterized in that, During the detection mode, the acquired operating information also includes the resistance value of the cycling platform. The preset angular acceleration is calculated based on the wheel speed and resistance value using an angular acceleration function.

8. The cycling trainer data collection method according to claim 1, characterized in that, When the calculation model makes a judgment, if the actual decrease in wheel speed within a preset time exceeds the preset decrease value; And / or, if the actual wheel speed is lower than the preset speed, the calculation model will determine that the user should stop riding.

9. A method for adjusting the parameters of a cycling trainer, characterized in that, When adjusting parameters, the parameters of the cycling train are adjusted using the data collected by the cycling train data collection method according to claim 1. The torque parameters of the cycling train under the corresponding working conditions are adjusted by the torque function based on the torque stored in the storage area.

10. The method for adjusting the parameters of a cycling train according to claim 9, characterized in that, When no torque is stored in the storage area corresponding to a certain working condition, the simulated torque under the working condition corresponding to the storage area without stored torque is calculated by using the torque already stored in other storage areas and their corresponding working conditions. The torque parameters under the corresponding working condition are then adjusted by the torque function based on the simulated torque.

Images