A Multi-Mode Adaptive Automatic Shifting Control Method for Bicycles
By combining an inertial measurement unit and a Hall sensor, the automatic transmission system of a bicycle can accurately identify and perform multi-mode adaptive control when stationary on a slope. This solves the problems of false triggering and unstable cadence, adapts to the gear requirements of different riding scenarios, and improves riding safety and smoothness.
Patent Information
- Application Number
- CN202611233248.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-14
- Publication Date
- 2026-11-17
AI Technical Summary
Existing automatic bicycle transmission systems are prone to accidental gear shifting when stationary on a slope, lack adaptability to multiple riding scenarios, have unstable cadence data collection, and struggle to meet the gear requirements of different riding scenarios.
The system uses an inertial measurement unit (IMU) to detect the actual start of cycling and uses an accelerometer and attitude angular velocity sensor. It sets up a multi-mode adaptive control method, including effort-saving, normal, and racing modes. By mapping the equivalent cadence to gears and combining acceleration trends with automatic cadence switching modes, it eliminates abnormal pulses and improves the stability of cadence data.
It accurately identifies the start of riding, suppresses accidental triggering when stationary on slopes, automatically switches gears to adapt to different riding scenarios, improves system integration and real-time performance, reduces accidental gear shifting, and enhances riding safety and smoothness.
Smart Images

Figure CN123104843A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent sensing and automatic transmission control technology for bicycles, and in particular to a multi-mode adaptive automatic transmission control method for bicycles. Background Technology
[0002] Multi-speed bicycles adapt to different road conditions and riding styles by changing the gear ratio between the chainring and the cassette. The rider's cadence and the gradient are two fundamental factors that determine the appropriate gear. Automatic transmissions automatically select gears based on riding parameters collected by sensors during the ride, eliminating the need for manual gear shifting.
[0003] To ensure gear shifting adapts to road conditions, existing automatic transmission solutions generally incorporate parameters such as gradient and cadence for decision-making. Patent CN104590479A uses a gyroscope to obtain gradient and combines it with cycling parameters such as cadence collected by sensors. These parameters are then uploaded to a server to calculate the optimal gear ratio before controlling the shifting. Patent CN110450900A determines uphill or downhill conditions based on crank speed and lean angle, and adjusts the shift threshold accordingly for uphill and downhill situations.
[0004] The above-mentioned schemes determine whether a vehicle has entered riding mode based on whether its speed or acceleration exceeds a set threshold. When a bicycle is stopped on a slope, there is a component of gravity along the slope. This component is collected by the acceleration sensor in the forward direction and mixed with the acceleration generated by actual riding. The steeper the slope, the larger this component becomes. Once it exceeds the start-up threshold, the system misinterprets a stationary state as a start, thus triggering a gear shift. The gear has been changed before the rider has even pedaled.
[0005] Different riding scenarios have different requirements for gears. Effortless cruising, commuting on flat roads, and sprinting each have their own focus. When climbing hills, riders need lower gears to save energy, while when racing, they need higher gears to maintain power. Existing solutions mostly use a single shifting strategy, or only make continuous fine adjustments to the threshold under hill conditions. The same logic cannot take into account these distinct riding intentions, nor can it automatically switch between these intentions according to the riding state.
[0006] The cadence acquisition process is also subject to disturbances. Cadence is obtained by counting the pulses generated by the sensor as magnets on the rotating component pass through them. When adjacent magnets are arranged with the same pole, the magnetic fields overlap and become distorted, making the edges of the sensor output level unclear. Electromagnetic interference and road vibrations can also introduce abnormal high-frequency pulses. Once these pulses are included in the cadence calculation, they will cause significant fluctuations in the cadence value, leading to frequent gear shifts. In addition, existing solutions often rely on external devices such as smart bike computers, mobile phones, or servers to complete parameter acquisition and calculation, which limits the system integration and real-time performance.
[0007] Therefore, how to accurately identify the actual start of riding to suppress false triggering caused by stagnation on slopes, how to adaptively select and switch shifting strategies for different riding scenarios, and how to ensure the stability of cadence data collection are problems that automatic transmissions for bicycles need to solve. Summary of the Invention
[0008] The technical problem to be solved by the present invention is to provide a multi-mode adaptive automatic transmission control method for bicycles, which solves the problems of existing automatic transmissions for bicycles being prone to accidental triggering when stationary on a slope and lacking adaptation to multiple riding scenarios.
[0009] To achieve the above objectives, the present invention adopts the following technical solution: A multi-mode adaptive automatic transmission control method for bicycles includes: acquiring real-time acceleration in the forward direction through an acceleration sensor installed in the inertial measurement unit of the bicycle, and calculating the real-time slope angle based on the attitude angular velocity of the inertial measurement unit. Set up a power-saving mode, a normal mode, and a racing mode. Each operating mode corresponds to a mode coefficient, and the mode coefficients of power-saving mode, normal mode, and racing mode increase in that order. Based on real-time cadence n c Mode coefficients relative to the current operating mode k mode Calculate equivalent cadence n c,eq The formula is: n c,eq = n c × (1 + k mode ); The equivalent cadence is mapped to the target shift gear according to a preset cadence-gear mapping relationship; The current operating mode is automatically updated according to the operating mode switching logic. Specifically, when the real-time slope angle is not less than the slope threshold, the labor-saving mode is switched on, and when the real-time slope angle falls below the slope threshold, the labor-saving mode is exited. When the real-time slope angle is less than the slope threshold, the difference between the number of positive events and the number of negative events of the real-time acceleration is determined as the acceleration trend within the statistical time window. When the acceleration trend is positive and the real-time cadence is greater than the racing entry threshold, the racing mode is entered. When the acceleration trend is negative and the real-time cadence is less than the normal entry threshold, the normal mode is entered. The target gear position is compared with the current gear position. If the two are not the same and the current gear position is a valid gear position, the gear is shifted up or down according to the relationship between the two.
[0010] Optionally, the labor-saving mode is the highest priority interrupt mode; when switching to the labor-saving mode, the current operating mode before switching is recorded as the historical mode; when exiting the labor-saving mode, the current operating mode is restored to the historical mode; the slope threshold is 5 degrees.
[0011] Optionally, the length of the statistical time window is 5 seconds; within the statistical time window, a real-time acceleration greater than a positive judgment threshold is recorded as a positive event, and a real-time acceleration less than a negative judgment threshold is recorded as a negative event, and a sliding window method is used to retain only the positive and negative events within the most recent 5 seconds; the formula for calculating the acceleration trend is: D = N pos - N neg Where D represents the acceleration trend, N pos This represents the number of positive events within the statistical time window. N neg The number of negative events within the statistical time window; when D is greater than zero, the acceleration trend is determined to be positive; when D is less than zero, the acceleration trend is determined to be negative; when D is equal to zero, the current operating mode remains unchanged.
[0012] Optionally, the difference between the maximum and minimum cadence is determined as the shift cadence difference value, and the calculation formulas for the racing cut-in threshold and the normal cut-in threshold are as follows: Δn c = n c,max - n c,min ; n c,racein = n c,max - k sens × ½ × Δn c ; n c,normalin = n c,min + k sens × ½ × Δn c ;in, Δn c This is the shift cadence difference. n c,max For maximum cadence, n c,min Minimum cadence, n c,racein For racing entry threshold, n c,normalin This is the normal cut-in threshold.k sens This is the mode switching sensitivity coefficient, which is set by the user through an application.
[0013] Optionally, before calculating the target gear shift, the method further includes: putting the inertial measurement unit into a low-power state and monitoring the real-time acceleration in the forward direction; marking a pre-start state when the real-time acceleration is greater than a pre-start acceleration threshold, and calculating the dynamic effective acceleration using the following formula: a dynamic = a x - g × sinθ; where, a dynamic For dynamic and effective acceleration, a x The real-time acceleration is g, the gravitational acceleration is θ, and the real-time slope angle is θ. When the dynamic effective acceleration is greater than the pre-start acceleration threshold, the pre-start state is confirmed to be valid; otherwise, the pre-start state is cleared. When the pre-start state is valid and valid guide wheel detection data is collected, automatic speed change is activated, and the current operating mode is set to normal mode.
[0014] Optionally, the acquisition of real-time cadence includes: alternately attaching N-pole permanent magnets and S-pole permanent magnets circumferentially to the inner ring surface of the bicycle guide wheel; using a unipolar Hall sensor to output pulse signals to one of the effective magnetic poles of the N-pole and S-pole permanent magnets; statistically counting the number of effective pulses output by the unipolar Hall sensor within the measurement time window; and calculating the real-time pulse frequency, the equivalent number of pulses per chainring revolution, and the real-time cadence using the following formula: f pulse = X / T; N pulse,cr = N eff × Z cr / Z gw ; n c = f pulse / N pulse,cr ;in, f pulse X represents the real-time pulse frequency, X represents the number of effective pulses within the measurement time window, and T represents the length of the measurement time window. N pulse,cr This represents the equivalent number of pulses per revolution of the dental disc. N eff The number of effective magnetic poles per revolution of the guide wheel. Z cr The number of teeth on the dental disc. Zgw The number of guide wheel teeth, n c For real-time cadence; the pulse frequency corresponding to the highest cadence of the system is determined as the maximum permissible pulse frequency, calculated using the following formula: f pulse,max = n c,max × N pulse,cr ;in, f pulse,max The maximum permissible pulse frequency, n c,max The system's highest cadence is used; when the real-time pulse frequency is greater than the maximum permissible pulse frequency, the data within the measurement time window is determined to be abnormal and discarded, and the real-time cadence obtained from the previous measurement time window is used.
[0015] Optionally, the attitude angular velocity output by the inertial measurement unit includes the real-time yaw angular velocity; when the absolute value of the real-time yaw angular velocity is greater than the turning determination threshold, the bicycle is determined to be in a turning state, and upshifting and downshifting are prohibited in the turning state; when the absolute value of the real-time yaw angular velocity falls back below the turning determination threshold, upshifting and downshifting are resumed.
[0016] Optionally, the real-time acceleration is collected within a set turbulence determination window, and the difference between the maximum and minimum values of the real-time acceleration within the turbulence determination window is determined as the acceleration fluctuation amplitude, calculated using the formula: R = a max - a min Where R is the amplitude of acceleration fluctuation, a max The maximum value of the real-time acceleration within the bump determination window. a min The minimum real-time acceleration within the bump determination window; when the acceleration fluctuation amplitude is greater than the bump determination threshold, it is determined to be a bumpy road condition, and the operation mode switching and upshifting / downshifting are suspended; when the acceleration fluctuation amplitude falls back below the bump determination threshold, the operation mode switching and upshifting / downshifting are resumed.
[0017] Optionally, a comfortable cadence estimate is obtained by taking a moving average of the real-time cadence within a set statistical period, and the mode coefficient of the current operating mode is corrected online based on the deviation between the comfortable cadence estimate and the reference cadence. The correction formula is as follows: k mode ′ = k mode + β × ( n comfort - nref ) / n ref ;in, k mode ′ represents the corrected mode coefficient. k mode The mode coefficients are before correction, and β is the adaptive gain. n comfort For the aforementioned comfort cadence estimation, n ref For reference cadence, the modified mode coefficient is limited to a preset range of mode coefficient values, and the modified mode coefficient is used to replace the mode coefficient of the current operating mode in the equivalent cadence calculation.
[0018] Optionally, the length of the measurement time window is dynamically adjusted according to the real-time cadence, so that the length of the measurement time window is negatively correlated with the real-time cadence, that is, the larger the real-time cadence, the smaller the length of the measurement time window, and the smaller the real-time cadence, the larger the length of the measurement time window, and the length of the measurement time window is limited to a preset window length value range.
[0019] The beneficial effects of this invention are as follows: This invention uses the accelerometer and attitude angular velocity of an inertial measurement unit to accurately identify the actual start of riding, suppressing false triggers caused by stationary positions on slopes. When the real-time acceleration exceeds the pre-start acceleration threshold, it is first marked as pre-start. Then, the component of gravity along the forward direction is subtracted from the real-time acceleration by the real-time slope angle to obtain the dynamic effective acceleration. Automatic shifting is only officially activated when the dynamic effective acceleration still exceeds the threshold and valid guide wheel detection data is collected. The component of gravity along the slope changes sinusoidally with the slope angle; the remaining component after subtracting this component is the actual riding acceleration. Existing schemes that determine start solely based on vehicle speed or initial acceleration cannot distinguish between the gravity component and the actual riding acceleration on slopes, leading to misjudgments of starting when stationary on a slope. This invention, by subtracting this component, can distinguish between them, and shifting is no longer triggered when stationary.
[0020] This invention employs three operating modes—effort-saving, normal, and racing—along with mode coefficients to differentiate gear selection for different cycling scenarios, balancing effort-saving for climbing and power for racing. Each operating mode corresponds to an progressively increasing mode coefficient. The equivalent cadence equals the real-time cadence multiplied by the mode coefficient and one; the smaller the mode coefficient, the lower the equivalent cadence, resulting in a lower gear and thus less effort. Gear shifting uses the equivalent cadence instead of the original cadence input mapping, allowing the same real-time cadence to derive different gears in different modes. Existing single strategies can only use one mapping to handle all scenarios; this invention uses a single mode coefficient to apply the same mapping to different scenarios, eliminating the need for a separate gear shifting logic for each scenario.
[0021] This invention automatically switches between three modes based on gradient, acceleration trend, and cadence, eliminating the need for cyclists to manually select modes. When the gradient angle is not less than a threshold, it switches to the energy-saving mode with the highest priority and reverts to the previous mode after the gradient decreases. When the gradient is normal, the acceleration trend is obtained by the difference between the number of positive and negative acceleration events within a statistical time window. If the trend is positive and the cadence is higher than the racing mode threshold, it switches to racing mode; if the trend is negative and the cadence is lower than the normal mode threshold, it switches to normal mode. The switching threshold varies with a user-adjustable sensitivity coefficient. Existing solutions require cyclists to manually switch or switch based solely on gradient. This invention uses both acceleration trend and cadence as criteria, reducing accidental switching at critical points.
[0022] This invention adjusts the mode coefficient online based on the cyclist's historical cadence, allowing shifting timing to align with individual riding habits. A comfortable cadence estimate is obtained by taking a sliding average of the real-time cadence, and the current mode coefficient is adjusted proportionally to its deviation from a reference cadence, remaining within a defined range. Existing mode coefficients or thresholds are mostly factory-fixed values, treating all cyclists with different cadence habits the same. This invention adjusts according to the cyclist's own cadence, providing different shifting timings for different people within the same mode.
[0023] This invention pauses gear shifting during turns and on bumpy roads, improving riding safety and smoothness. A turn is identified when the absolute value of the yaw rate exceeds the turn threshold, and upshifting or downshifting is prohibited during a turn. Similarly, a bumpy road condition is identified when the difference between the maximum and minimum real-time acceleration values within the bump detection window exceeds the bump detection threshold, pausing mode switching and upshifting / downshifting. Shifting gears when the chain is under stress during turns or on bumpy roads can easily lead to skipped teeth and chain slippage. Existing solutions that only consider cadence or incline do not distinguish between these two conditions, potentially leading to shifting at the least suitable time. This invention uses the same inertial signal to identify turns and bumps, pausing gear shifting in both situations.
[0024] This invention uses a magnetic ring with alternating N and S poles in conjunction with a unipolar Hall sensor to collect pedal frequency, and eliminates abnormal pulses and adaptively adjusts the measurement window according to the pedal frequency to ensure stable pedal frequency data. The permanent magnets in the inner ring of the guide wheel teeth have alternating N and S poles to avoid distortion caused by the superposition of magnetic fields of the same pole, resulting in regular pulse edges; the unipolar Hall sensor is triggered only by one effective magnetic pole; when the real-time pulse frequency exceeds the maximum allowable pulse frequency calculated from the system's highest pedal frequency, it is judged as abnormal and the pedal frequency of the previous window is used; the measurement window length is negatively adjusted according to the real-time pedal frequency, striking a trade-off between response speed and anti-interference. Existing fixed-window acquisition methods without abnormality elimination will cause pedal frequency jumps and incorrect gear shifts when encountering interference pulses or magnetic field distortions. This invention eliminates disturbances in three ways: magnet arrangement, abnormality elimination, and adaptive window adjustment.
[0025] This invention uses an onboard inertial measurement unit and Hall effect sensors to complete the riding status perception and gear shifting decision on the bicycle itself, without relying on external devices such as smart bike computers, mobile phones or servers. The system integration and real-time performance are higher than solutions that rely on external devices. Attached Figure Description
[0026] Figure 1 This is a schematic diagram of the structure of the bicycle automatic transmission control system provided in an embodiment of the present invention; Figure 2 A schematic diagram illustrating the mechanism of slope gravity component decomposition and start-up determination provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the mapping between equivalent cadence and target shift gear under three operating modes provided in the embodiments of the present invention; Figure 4 This is a schematic diagram illustrating the automatic switching of operating modes provided in an embodiment of the present invention. Figure 5 This is a schematic diagram of the acceleration trend statistics within a statistical time window provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of the guide wheel alternating magnetic ring and unipolar Hall tread frequency acquisition provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the Hall pulse signal and the abnormal high-frequency filtering signal provided in an embodiment of the present invention; Figure 8 A schematic diagram showing the comparison between turning suppression, bump suppression, and online adaptive mode coefficients provided in an embodiment of the present invention; Figure 9 This is a flowchart illustrating the multi-mode adaptive automatic transmission control method for bicycles provided in an embodiment of the present invention.
[0027] Explanation of reference numerals in the attached figures: 10-Inertial Measurement Unit; 11-Acceleration Sensor; 20-Guide Wheel; 21-Crankset; 22-Chain; 23-N-pole permanent magnet; 24-S-pole permanent magnet; 25-Unipolar Hall Sensor; 30-Bicycle; 40-Real-time Acceleration Curve; 50-Pulse Signal. Detailed Implementation
[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions in the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0029] Figure 1This is a schematic diagram of the structure of a bicycle automatic gear shifting control system provided in an embodiment of the present invention. (Refer to...) Figure 1 The bicycle 30 is equipped with a chainring 21, guide wheels 20, and a chain 22. The chain 22 passes around the chainring 21 and is tensioned by the guide wheels 20 before engaging with the rear sprocket. The guide wheels 20 are tension wheels on the rear derailleur mechanism and rotate synchronously with the chain 22 and chainring 21. When the rider pedals the chainring 21, the chain 22 drives the rear sprocket to rotate. The rotational speed of the guide wheels 20 and the rotational speed of the chainring 21 maintain a fixed proportional relationship determined by mechanical transmission parameters. Therefore, the rotational information of the guide wheels 20 can be used to infer the rider's cadence. An inertial measurement unit 10 is mounted on the frame of the bicycle 30 and integrates an accelerometer 11. The accelerometer 11 collects the real-time acceleration in the forward direction of the bicycle 30, and the inertial measurement unit 10 also outputs the attitude angular velocity of the bicycle body. A unipolar Hall sensor 25 is located near the guide wheels 20, maintaining a sensing gap with the inner ring surface of the guide wheels 20, and is used to output pulse signals when the guide wheels 20 rotate.
[0030] In one embodiment, the inertial measurement unit 10 employs a six-axis accelerometer-gyroscope sensor, which simultaneously provides three-axis acceleration and three-axis angular velocity outputs within the same chip. The accelerometer sensor 11 is the portion providing the forward acceleration component. The attitude angular velocity output by the inertial measurement unit 10 is integrated and corrected for gravity direction to calculate the real-time slope angle of the bicycle 30. This calculation is a conventional process in attitude measurement and can be implemented using complementary filtering or Kalman filtering. By directly mounting the inertial measurement unit 10 and the unipolar Hall sensor 25 onto the bicycle 30 body, the perception of riding status and gear shifting decisions are completed within the bicycle 30 itself, without relying on external devices such as smart bike computers, mobile phones, or servers, thus ensuring the real-time performance of perception and calculation.
[0031] Figure 9 This is a flowchart illustrating the multi-mode adaptive automatic transmission control method for bicycles provided in an embodiment of the present invention. (Refer to...) Figure 9 The method includes steps S101 to S107. In step S101, real-time acceleration in the forward direction is collected and the real-time slope angle is calculated. In step S102, automatic shifting is activated when the pre-start state is valid and valid guide wheel detection data is collected. In step S103, the operating mode is automatically switched according to slope angle, acceleration trend, and real-time cadence. In step S104, the real-time cadence is converted into an equivalent cadence using the mode coefficient of the current operating mode. In step S105, the target shift gear is determined according to the mapping relationship between cadence and gear position. In step S106, it is determined whether the target shift gear is inconsistent with the current gear. In step S107, if they are inconsistent, upshifting or downshifting is performed according to their relative magnitudes. The processing involved in each of the above steps will be explained in detail below with reference to the accompanying drawings.
[0032] The core of automatic transmission lies in selecting the appropriate gear based on the riding state, and the rider's cadence and the slope are two fundamental factors determining the appropriate gear. In step S101, the accelerometer 11 collects real-time acceleration in the forward direction at a fixed period. This collection period can be 1 to 20 milliseconds in some embodiments, for example, 10 milliseconds. If the collection period is too long, instantaneous changes in acceleration will be missed; if it is too short, it will increase the computational burden on the main controller. The attitude angular velocity output by the inertial measurement unit 10 is used to calculate the real-time slope angle. The real-time slope angle represents the degree of inclination of the road surface where the bicycle 30 is located relative to the horizontal plane, positive when going uphill and negative when going downhill. The real-time slope angle is the common input for subsequent start-up determination and operation mode switching, so the collection and calculation are completed at the beginning of the process.
[0033] In step S104, the real-time cadence is converted into an equivalent cadence using the mode coefficient of the current operating mode. This is the core processing step for the multi-mode adaptive function of this invention. This invention sets three operating modes: energy-saving mode, normal mode, and racing mode. Each operating mode corresponds to a mode coefficient, and the mode coefficients of the energy-saving mode, normal mode, and racing mode increase sequentially. Figure 3 This is a schematic diagram illustrating the mapping between equivalent cadence and target shift gear under three operating modes provided in embodiments of the present invention. (Refer to...) Figure 3 The same real-time cadence, after being converted by its respective mode coefficient in three operating modes, yields an equivalent cadence that increases sequentially, which is then mapped to a target gear that increases sequentially. The equivalent cadence is calculated using the following formula: n c,eq = n c × (1 + k mode (Equation 1) in, n c,eq For equivalent cadence, n c For real-time cadence, k modeThis represents the mode coefficient for the current operating mode. As shown in Equation 1, a smaller mode coefficient results in a lower equivalent cadence, a lower target gear, and less effort required for riding; conversely, a larger mode coefficient results in a higher equivalent cadence, a higher target gear, and stronger riding power. By using the equivalent cadence instead of the real-time cadence as the input for gear shift mapping, the same real-time cadence can achieve different target gears in different operating modes. Therefore, only one mode coefficient needs to be adjusted to transfer the same mapping relationship to different riding scenarios without needing to design a separate gear shift logic for each scenario. In one embodiment, the mode coefficient for the effort-saving mode can be 0.1 to 0.3, typically 0.2; the mode coefficient for the normal mode can be 0.4 to 0.6, typically 0.5; and the mode coefficient for the racing mode can be 0.7 to 0.9, typically 0.8. When the mode factor is small, the increase in equivalent cadence relative to real-time cadence is limited, and the cadence required to switch to a higher level is relatively high, making the ride more effortless. When the mode factor is large, a lower real-time cadence is sufficient to switch to a higher level, making the ride more power-oriented. Increasing the three mode factors in sequence can cover the three riding intentions of effortless, smooth, and racing.
[0034] In step S105, the target shift gear is determined according to the mapping relationship between cadence and gear position. This mapping relationship uses equivalent cadence as input, first converting the equivalent cadence into the guide wheel detection frequency, and then determining the target shift gear based on the range in which the guide wheel detection frequency falls. The guide wheel detection frequency is obtained from the equivalent cadence using the following formula: f gw = ( n c,eq / 60) × ( Z cr / Z gw ) × k gw (Equation 2) in, f gw The frequency of guide wheel detection per unit time is expressed in times per second. n c,eq Equivalent cadence, measured in revolutions per minute; Z cr This refers to the number of teeth on the dental disc. Z gw This refers to the number of teeth on the guide wheel. k gwThe guide wheel detection frequency coefficient is defined in Equation 2. This involves first converting the equivalent pedal frequency to the number of chainring rotations per second, then multiplying this by the number of chainring teeth to obtain the total number of chain teeth moving per second. Finally, the guide wheel detection frequency is calculated based on the guide wheel teeth and the detection frequency coefficient. In one embodiment, the chainring teeth are 36, the guide wheel teeth are 12, and the guide wheel detection frequency coefficient is 12. The guide wheel detection frequency coefficient represents the number of times the guide wheel is detected per revolution, and it represents the same detection density as the effective magnetic pole number per revolution of the guide wheel described later. Since the guide wheel detection frequency coefficient is canceled out as a common factor in the calculation of the gear frequency interval and the target shift gear, its value does not affect the final determined target shift gear. Substituting the system's preset maximum cadence into Equation 2 yields the maximum guide wheel detection frequency, and substituting the minimum cadence into Equation 2 yields the minimum guide wheel detection frequency. In some embodiments, the maximum cadence can be 120 to 180 revolutions per minute, typically 150 revolutions per minute; the minimum cadence can be 40 to 70 revolutions per minute, typically 60 revolutions per minute. The gear frequency interval is obtained using the following formula: Δf = ( f gw,max - f gw,min ) / ( L max - 1) (Equation 3) Where Δf is the frequency interval of the gear. f gw,max This is the maximum guide wheel detection frequency. f gw,min Minimum guide wheel detection frequency, L max The maximum shift gear is defined in Equation 3. The span between the maximum and minimum guide wheel detection frequencies is evenly divided into the maximum gear minus one interval, yielding the guide wheel detection frequency width corresponding to a single gear. In one embodiment, the maximum shift gear is set to 12, and the minimum shift gear to 1. The target shift gear is determined in three scenarios: when the guide wheel detection frequency calculated from the equivalent pedal frequency is not lower than the maximum guide wheel detection frequency, the target shift gear is the maximum shift gear; when the guide wheel detection frequency is not higher than the minimum guide wheel detection frequency, the target shift gear is the minimum shift gear; when the guide wheel detection frequency is between the two, the difference between the guide wheel detection frequency and the minimum guide wheel detection frequency is divided by the gear frequency interval and rounded down, then added to the minimum shift gear to obtain the target shift gear. The reason for discretizing the continuous guide wheel detection frequencies into a finite number of gears according to the gear frequency interval is that the bicycle 30's gearshift mechanism only has a finite number of discrete gears. Rounding down ensures that the shifted gear will not be ahead of the gear supported by the current pedal frequency.
[0035] In one embodiment, the calculation of the target gear is triggered once every gear shift detection cycle, which can be from 1 second to 5 seconds, typically 3 seconds. The system increments a timer counter once per second. When the timer counter reaches the gear shift detection cycle, a gear shift detection is triggered and the timer counter is reset. If the gear shift detection cycle is too short, the gear will change frequently with the instantaneous fluctuations in cadence, which can easily cause jerking; if the gear shift detection cycle is too long, the gear will respond slowly to changes in cadence. By using timed triggering instead of point-by-point triggering, the gear shifting action is made sparse in time, avoiding short-term cadence fluctuations from directly causing gear shifts.
[0036] Figure 4 This is a schematic diagram illustrating the automatic switching of operating modes provided in an embodiment of the present invention. (Refer to...) Figure 4 In step S103, the operating mode is automatically switched based on slope angle, acceleration trend, and real-time cadence. The automatic switching of operating modes has two levels: interruption of the effort-saving mode triggered by slope, and a dual-condition switch between normal and racing modes. After the system activates automatic gear shifting, it defaults to normal mode. When the real-time slope angle is not less than the slope threshold, it switches to effort-saving mode; when the real-time slope angle falls below the slope threshold, it exits effort-saving mode. When the real-time slope angle is less than the slope threshold, it switches between normal and racing modes based on a dual-condition switch between acceleration trend and real-time cadence: when the acceleration trend is positive and the real-time cadence is greater than the racing entry threshold, it switches to racing mode; when the acceleration trend is negative and the real-time cadence is less than the normal entry threshold, it switches to normal mode. The reason for prioritizing the slope-triggered effort-saving mode switch before the dual-condition switching is that riders prioritize effort saving when climbing, thus giving the effort-saving mode the highest priority. In non-climbing conditions, the switching between normal and racing modes is based on the rider's power exertion and cadence. In one embodiment, the slope threshold can range from 3 to 10 degrees, typically 5 degrees. A lower slope threshold triggers the effort-saving mode even on gentle slopes, and slight undulations on flat roads may cause a mode switch. A higher slope threshold requires a steeper slope to activate the effort-saving mode, and the effort-saving benefits on moderate slopes cannot be obtained promptly.
[0037] In step S102, automatic transmission is activated when the pre-start state is valid and valid guide wheel detection data is collected. This step determines the actual start of riding before automatic transmission is activated, which is used to suppress false triggering caused by factors such as standing still on a slope. Figure 2 This is a schematic diagram illustrating the mechanism of ramp gravity component decomposition and start-up determination provided in an embodiment of the present invention. (Refer to...) Figure 2When bicycle 30 is parked or riding on a slope, the real-time acceleration in the forward direction collected by acceleration sensor 11 includes both the actual acceleration generated by riding and the component of gravity along the forward direction. The component of gravity along the forward direction is equal to the product of gravitational acceleration and the sine of the real-time slope angle; the steeper the slope, the larger this component. If the determination of whether riding has started is based solely on whether the real-time acceleration in the forward direction exceeds a threshold, then when bicycle 30 is parked on a slope, the component of gravity along the forward direction may be misinterpreted as actual riding acceleration, thus mistakenly activating automatic transmission and triggering gear shifting before the rider has even pedaled.
[0038] To eliminate the aforementioned misjudgments, a triple-condition activation mechanism is employed. Before automatic transmission is activated, the inertial measurement unit 10 operates in a low-power state and continuously monitors the real-time acceleration in the forward direction. When the real-time acceleration exceeds the pre-activation acceleration threshold, a pre-activation state is established. Subsequently, the dynamic effective acceleration is calculated using the following formula: a dynamic = a x - g × sinθ (Equation 4) in, a dynamic For dynamic effective acceleration; a x θ represents the real-time acceleration; g represents the gravitational acceleration; and θ represents the real-time slope angle. Equation 4 subtracts the component of gravity along the forward direction from the real-time acceleration in the forward direction; the remaining value is the actual riding acceleration. When the dynamic effective acceleration is still greater than the pre-start acceleration threshold, the pre-start state is confirmed to be valid; otherwise, the pre-start state is cleared and the system returns to low-power monitoring. After the pre-start state is valid, valid guide wheel detection data must be collected, i.e., the guide wheel 20 has rotated and outputs a recognizable pulse signal, before the automatic transmission is finally activated and the current operating mode is set to normal mode. Figure 2 The dynamic effective acceleration vector is not plotted separately. Dynamic effective acceleration is the difference between the real-time acceleration vector in the forward direction and the component vector of gravity along the forward direction. In one embodiment, the pre-start acceleration threshold can be between 0.05g and 0.2g, typically 0.1g, where g is the gravitational acceleration. If the pre-start acceleration threshold is too small, even slight vibrations or pushing can trigger pre-start, reducing its anti-interference capability; if the pre-start acceleration threshold is too large, a slow start may not trigger pre-start, resulting in delayed activation. By using the component of gravity along the forward direction as a subtraction term, the bicycle 30 stopped on the ramp will not be misjudged as starting due to the gravity component; only the acceleration component generated by actual pedaling can cause the dynamic effective acceleration to exceed the threshold.
[0039] Figure 9Steps S106 and S107 complete the gear shifting process. In step S106, the target gear is compared with the current gear to determine if they are inconsistent. In step S107, if they are inconsistent and the current gear is a valid gear, if the target gear is higher than the current gear, an upshift is performed; if the target gear is lower than the current gear, a downshift is performed. If the target gear is the same as the current gear, no gear shift is performed. Here, a valid gear means that the current gear is within the valid range from the minimum to the maximum gear. If the current gear is not yet established and is in an invalid gear during the initial startup phase, no gear shift is performed either. The transmission mechanism is only driven when a gear change is required, avoiding idle operation when no gear change is needed.
[0040] The following describes several supplementary embodiments of the present invention based on the core solution.
[0041] In some embodiments, the power-saving mode is the highest-priority interrupt mode. (See also...) Figure 4 The energy-saving mode is located at the top of the status diagram, while the normal and racing modes are located at the bottom. The energy-saving mode is bidirectionally connected to both the normal and racing modes. When switching to energy-saving mode, the current operating mode before switching is recorded as the historical mode; when exiting energy-saving mode, the current operating mode is restored to this historical mode. Since cyclists may climb hills in normal mode or racing mode, energy-saving mode needs to be able to enter from either normal or racing mode and return to the previous mode after the gradient decreases. The historical mode recording process ensures this: for example, if a cyclist encounters an uphill in racing mode, and the gradient angle is not less than a gradient threshold, they switch to energy-saving mode and the racing mode is recorded as the historical mode; when the gradient decreases to below the gradient threshold, they revert to racing mode, rather than always returning to normal mode. In one embodiment, the aforementioned gradient threshold is 5 degrees, which also serves as the criterion for switching to and exiting energy-saving mode. Optionally, the entry slope threshold and exit slope threshold of the labor-saving mode can be different values. The exit slope threshold is slightly smaller than the entry slope threshold, and the two form a hysteresis interval to avoid the labor-saving mode repeatedly entering and exiting when the slope fluctuates near the threshold.
[0042] Figure 5 This is a schematic diagram of an acceleration trend statistical signal within a statistical time window provided in an embodiment of the present invention. In some embodiments, the acceleration trend is obtained by statistically analyzing positive and negative events within the statistical time window. (Refer to...) Figure 5 , Figure 5The real-time acceleration curve 40 in the figure is the solid line, representing the change of real-time acceleration in the forward direction over time. Within the statistical time window, a real-time acceleration greater than the positive judgment threshold is recorded as a positive event, and a real-time acceleration less than the negative judgment threshold is recorded as a negative event. A sliding window method is used to retain only positive and negative events within the most recent statistical time window. The acceleration trend is calculated using the following formula: D = N pos - N neg (Equation 5) Where D represents the acceleration trend; N pos To count the number of positive events within a time window; N neg The statistical time window is used to count the number of negative events. When D is greater than zero, the acceleration trend is considered positive, indicating that the rider is primarily accelerating within the time window, supporting a switch to racing mode. When D is less than zero, the acceleration trend is considered negative, indicating that the rider is primarily decelerating, supporting a switch to normal mode. When D equals zero, the current operating mode remains unchanged. The difference between the number of positive and negative events within the statistical time window is used as the acceleration trend, rather than the sign of a single acceleration, because a single acceleration is greatly affected by road bumps and instantaneous pedaling rhythm. Counting the number of events within a time window can filter out instantaneous fluctuations and reflect a more stable force application trend of the rider. In one embodiment, the statistical time window can be 3 to 8 seconds, typically 5 seconds, and the length of the statistical time window is greater than the aforementioned shift detection cycle. When the statistical time window is too short, the acceleration trend is sensitive to instantaneous fluctuations, and the mode is prone to frequent switching; when the statistical time window is too long, the acceleration trend responds more slowly to changes in the rider's force application. In some embodiments, the positive and negative judgment thresholds are set to 0.03g to 0.1g and their opposites, for example, 0.05g and -0.05g, respectively. When the judgment threshold is too large, some real acceleration and deceleration events are missed because they do not exceed the threshold, resulting in a weak acceleration trend. When the judgment threshold is too small, tiny fluctuations close to zero during constant-speed riding are counted as positive and negative events, making the acceleration trend susceptible to noise interference. The statistical work on acceleration trends is carried out on near-flat road conditions where the slope angle is less than the slope threshold. At this time, the component of gravity along the forward direction corresponding to the residual slope is small. In some embodiments, the real-time acceleration used to count positive and negative events can also first deduct the component of gravity along the forward direction before comparing it with the judgment threshold to eliminate the influence of the residual slope component on the event statistics.
[0043] In some embodiments, the racing cut-in threshold and the normal cut-in threshold are determined by the shift cadence difference and the mode switching sensitivity coefficient. The difference between the maximum and minimum cadence is defined as the shift cadence difference, and the racing cut-in threshold and the normal cut-in threshold are calculated using the following formula: Δn c = n c,max - n c,min (Equation 6) n c,racein = n c,max - k sens × (1 / 2) × Δn c (Equation 7) n c,normalin = n c,min + k sens × (1 / 2) × Δn c (Equation 8) in, Δn c This is the shift cadence difference. n c,max Maximum cadence; n c,min Minimum cadence; n c,racein The threshold for entering a racing mode; n c,normalin This is the normal cut-in threshold; k sens This refers to the mode switching sensitivity coefficient. As shown in Equations 7 and 8, the racing cut-in threshold shifts downwards from the maximum cadence, while the normal cut-in threshold shifts upwards from the minimum cadence. The amount of shift is directly proportional to the mode switching sensitivity coefficient. The mode switching sensitivity coefficient is set by the user through the application and can range from 0 to 1. When the mode switching sensitivity coefficient is large, the racing cut-in threshold shifts downwards and the normal cut-in threshold shifts upwards, making it easier to trigger the transition between normal and racing modes, resulting in more sensitive mode switching. Conversely, when the mode switching sensitivity coefficient is small, the cut-in threshold is closer to the extremes of the cadence, resulting in slower mode switching. By allowing the user to adjust the ease of switching through the mode switching sensitivity coefficient, the same switching logic can adapt to different cyclists' preferences for mode switching sensitivity.
[0044] Figure 6 This is a schematic diagram of the alternating magnetic rings on the guide wheel and the unipolar Hall sensor for ergonomic frequency acquisition provided in an embodiment of the present invention. In some embodiments, the real-time ergonomic frequency is acquired by a magnetic ring disposed on the guide wheel 20 and a unipolar Hall sensor 25. (Refer to...) Figure 6N-pole permanent magnets 23 and S-pole permanent magnets 24 are alternately pasted circumferentially on the inner ring surface of the guide wheel 20, meaning that the magnetic poles of adjacent permanent magnets are opposite. A unipolar Hall sensor 25 is disposed outside the ring of the guide wheel 20 and only generates a trigger response to one of the effective magnetic poles of the N-pole permanent magnets 23 and S-pole permanent magnets 24. The unipolar Hall sensor 25 outputs a pulse signal each time the guide wheel 20 rotates past an effective magnetic pole. The permanent magnets are arranged alternately with N-pole and S-pole because if adjacent permanent magnets are arranged with the same pole, the magnetic fields of adjacent magnets will overlap and be distorted, making the magnetic field boundaries detected by the unipolar Hall sensor 25 blurry and the transition edges irregular. The alternating arrangement isolates the magnetic fields of adjacent magnets, making the edges of the pulse signal regular and reducing the probability of false triggering from a hardware perspective. In one embodiment, the number of effective magnetic poles per revolution of the guide wheel 20 can be 6 to 12.
[0045] Within the measurement time window, the number of effective pulses output by the unipolar Hall sensor 25 is counted, and the real-time cadence is calculated based on the pulse equivalent. The number of teeth that the chain 22 moves when the chain 21 rotates once is equal to the number of teeth on the chain 21. The guide wheel 20 rotates synchronously by the corresponding number of teeth. Therefore, the equivalent effective pulse count corresponding to one rotation of the chain 21 is equal to the product of the number of effective magnetic poles of the guide wheel 20 per rotation and the number of teeth on the chain 21, divided by the number of teeth on the guide wheel. The real-time cadence is equal to the number of effective pulses within the measurement time window divided by the equivalent effective pulse count corresponding to one rotation of the chain 21, then divided by the length of the measurement time window and converted into revolutions per minute. The above conversion reverses the total number of pulses within the measurement time window to the number of rotations of the chain 21, and then converts it into rotational speed per unit time, which is the average cadence within the measurement time window.
[0046] Figure 7 This is a schematic diagram of a Hall pulse signal and an abnormal high-frequency filtered signal provided in an embodiment of the present invention. (Refer to...) Figure 7 , Figure 7 The pulse signal 50 in the figure is represented by the solid line. In some embodiments, abnormal high-frequency filtering is performed on the acquired pulse signal. The pulse frequency corresponding to the highest system cadence is determined as the maximum permissible pulse frequency. The maximum permissible pulse frequency is equal to the product of the highest system cadence and the equivalent effective pulse number corresponding to one revolution of the chainring 21. The highest system cadence and the real-time pulse frequency are compared using the same time base. When the real-time pulse frequency within the measurement time window is greater than the maximum permissible pulse frequency, the data within that measurement time window is determined to be abnormal and discarded, and the real-time cadence obtained from the previous measurement time window is used. (Refer to...) Figure 7During normal riding, the pulse signal 50 is relatively sparse, while abnormal high-frequency pulses introduced by electromagnetic interference, hardware malfunctions, or road vibrations are relatively dense. The corresponding real-time pulse frequency exceeds the maximum permissible pulse frequency calculated from the system's highest cadence. The reason for using the maximum permissible pulse frequency calculated from the system's highest cadence as the criterion is that there is a physiological upper limit to a rider's cadence. Signals with pulse frequencies exceeding this upper limit cannot be generated by actual pedaling and can only come from interference. Discarding the data in this window and using the previous value can prevent abnormal pulses from being included in the cadence calculation and causing incorrect gear shifting.
[0047] Figure 8 This is a schematic diagram comparing turn suppression, bump suppression, and online adaptive mode coefficients provided in an embodiment of the present invention. In some embodiments, gear shifting is suppressed when the bicycle makes a 30° turn. (Refer to...) Figure 8 In the left sidebar, the attitude angular velocity output by the inertial measurement unit 10 includes the real-time yaw angular velocity. When the absolute value of the real-time yaw angular velocity is greater than the turning determination threshold, the bicycle 30 is determined to be in a turning state, and upshifting and downshifting are prohibited in the turning state. When the absolute value of the real-time yaw angular velocity falls below the turning determination threshold, upshifting and downshifting are resumed. The reason for prohibiting gear shifting in the turning state is that shifting gears under the stress of turning can easily cause the chain 22 to skip teeth or even fall off. Pausing gear shifting during turning can improve riding safety. In some embodiments, the turning determination threshold can be set from 20 degrees per second to 60 degrees per second.
[0048] In some embodiments, shifting is suppressed under bumpy road conditions. (See reference...) Figure 8 In the middle column, real-time acceleration is collected within the set bump judgment window. The difference between the maximum and minimum real-time acceleration values within the bump judgment window is determined as the acceleration fluctuation amplitude, calculated using the following formula: R = a max - a min (Equation 9) Where R is the acceleration fluctuation amplitude; a max This represents the maximum real-time acceleration within the turbulence detection window. a minThe minimum real-time acceleration value within the bump detection window is used. When the acceleration fluctuation exceeds the bump detection threshold, the road condition is determined to be bumpy, and the operation mode switching and upshifting / downshifting are suspended. When the acceleration fluctuation falls below the bump detection threshold, the operation mode switching and upshifting / downshifting are resumed. The difference between the maximum and minimum acceleration values within the bump detection window is used to characterize the road bumpiness because bumpy roads cause acceleration to fluctuate drastically in a short period of time, increasing the difference between the maximum and minimum values. This allows for the distinction between bumpy and smooth road conditions, preventing the drastic acceleration fluctuations caused by bumps from being misinterpreted as changes in the rider's force and triggering gear shifting. In some embodiments, the bump detection threshold can be between 0.3g and 0.8g; the length of the bump detection window can be between 0.5 seconds and 2 seconds. Both cornering suppression and bump suppression utilize signals already acquired by the inertial measurement unit 10, eliminating the need for additional sensors.
[0049] In some embodiments, the mode coefficients are adaptively corrected online. (See reference...) Figure 8 The right-hand column calculates a moving average of the real-time cadence over a set statistical period to obtain a comfort cadence estimate, which represents the cadence level the cyclist is accustomed to. Based on the deviation between the comfort cadence estimate and the reference cadence, the mode coefficient of the current operating mode is corrected online using the following formula: k mode ' = k mode + β × ( n comfort - n ref ) / n ref (Equation 10) in, k mode ' represents the corrected mode coefficients; k mode The mode coefficients are before correction; β is the adaptive gain. n comfort For comfort cadence estimation; n refThe reference cadence is used. As shown in Equation 10, when the cyclist's accustomed cadence is higher than the reference cadence, the estimated comfortable cadence is also higher than the reference cadence, the mode coefficient is increased, and a lower real-time cadence is sufficient to shift to a higher gear, thus advancing the shift point. Conversely, when the cyclist's accustomed cadence is lower than the reference cadence, the mode coefficient is decreased, and the shift point is delayed. The corrected mode coefficient is limited to a preset range, and then used to replace the current operating mode's mode coefficient in the calculation of the equivalent cadence. The reason for using the cyclist's own historical cadence to correct the mode coefficient online is that the factory-set mode coefficient is designed for standard cyclists, while different cyclists have different cadence habits. By adjusting the mode coefficient online based on the deviation between the estimated comfortable cadence and the reference cadence, the same operating mode can provide different cyclists with shift points that match their cadence habits. In one embodiment, the reference cadence can be 60 to 120 revolutions per minute, typically 90 revolutions per minute; the adaptive gain can be 0.05 to 0.3; and the statistical duration for obtaining the comfort cadence estimate can be 30 to 300 seconds. When the statistical duration is too short, the comfort cadence estimate fluctuates with the rider's instantaneous cadence, and the mode coefficient correction is unstable; when the statistical duration is too long, the mode coefficient tracks changes in the rider's cadence habits more slowly.
[0050] In some embodiments, the length of the measurement time window is adaptively adjusted. The length of the measurement time window is negatively correlated with the real-time cadence; that is, the higher the real-time cadence, the shorter the measurement time window; and the lower the real-time cadence, the longer the measurement time window. The length of the measurement time window is limited to a preset range. In some embodiments, the length of the measurement time window can be between 0.2 seconds and 1 second. The measurement time window is the statistical period for pulse counting, and its length determines the response speed and anti-interference capability of cadence acquisition: a longer measurement time window enhances the averaging effect within the window and improves the ability to resist random interference, but it masks short-term changes in cadence, weakening the timeliness of the response; a shorter measurement time window improves the response speed, but it is more susceptible to interference from single pulses and magnet installation deviations, increasing the fluctuation in cadence calculation. When the cyclist's cadence is high, the pulses are denser per unit time, and a shorter measurement time window can accumulate enough pulses; shortening the window is beneficial for timely reflection of cadence changes. When the cyclist's cadence is low, the pulses are sparser, and extending the window can accumulate enough pulses to suppress fluctuations. The length of the measurement time window is negatively correlated with the real-time cadence, that is, an adaptive trade-off is made between response speed and anti-interference ability with cadence.
[0051] This invention also provides an automatic transmission control system for bicycles to implement the aforementioned method. This system can be implemented as a controller mounted on a bicycle 30. The controller is connected to an inertial measurement unit 10, a unipolar Hall sensor 25, and the transmission mechanism of the bicycle 30. It receives real-time acceleration and attitude angular velocity output from the inertial measurement unit 10, and receives pulse signals output from the unipolar Hall sensor 25. Following the aforementioned method, it acquires real-time cadence and real-time slope angle, automatically switches operating modes, calculates equivalent cadence and target gear, and executes gear shifts. Functionally, the controller can be divided into a data acquisition unit, a mode management unit, and a gear calculation unit, and also includes a gear shift execution unit. The data acquisition unit acquires and preprocesses the aforementioned real-time acceleration, attitude angular velocity, and pulse signals; the mode management unit automatically switches operating modes and performs online adaptive correction of mode coefficients; the gear calculation unit calculates equivalent cadence and target gear; and the gear shift execution unit drives upshifts and downshifts. Each functional unit corresponds to one of the steps in the aforementioned method, and will not be repeated here.
[0052] In one embodiment, the initial settings of the aforementioned thresholds, parameters, and operating modes can be configured via an application communicatively connected to the bicycle 30, and the mode switching sensitivity coefficient is set by the user through the application. Optionally, the cyclist can also manually specify the current operating mode through the application. In this case, the manually specified operating mode takes precedence over automatic switching based on acceleration trends and real-time cadence, while the interruption of the effort-saving mode triggered by the slope still maintains the highest priority.
[0053] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit them. Those skilled in the art should understand that modifications can be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features, and such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A multi-mode adaptive automatic transmission control method for bicycles, characterized in that, include: The real-time acceleration in the forward direction is collected by the accelerometer sensor installed in the inertial measurement unit of the bicycle, and the real-time slope angle is calculated based on the attitude angular velocity of the inertial measurement unit. Set up a power-saving mode, a normal mode, and a racing mode. Each operating mode corresponds to a mode coefficient, and the mode coefficients of power-saving mode, normal mode, and racing mode increase in that order. Based on real-time cadence n c Mode coefficients relative to the current operating mode k mode Calculate equivalent cadence n c,eq The formula is: n c,eq = n c × (1 + k mode ); The equivalent cadence is mapped to the target shift gear according to a preset cadence-gear mapping relationship; The current operating mode is automatically updated according to the operating mode switching logic. Specifically, when the real-time slope angle is not less than the slope threshold, the labor-saving mode is switched on, and when the real-time slope angle falls below the slope threshold, the labor-saving mode is exited. When the real-time slope angle is less than the slope threshold, the difference between the number of positive events and the number of negative events of the real-time acceleration is determined as the acceleration trend within the statistical time window. When the acceleration trend is positive and the real-time cadence is greater than the racing entry threshold, the racing mode is entered. When the acceleration trend is negative and the real-time cadence is less than the normal entry threshold, the normal mode is entered. The target gear position is compared with the current gear position. If the two are not the same and the current gear position is a valid gear position, the gear is shifted up or down according to the relationship between the two.
2. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, The labor-saving mode is the highest priority interrupt mode; When switching to the labor-saving mode, the current operating mode before switching is recorded as the historical mode; When exiting the effort-saving mode, the current operating mode will be restored to the historical mode; The slope threshold is 5 degrees.
3. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, The length of the statistical time window is 5 seconds; Within the statistical time window, a real-time acceleration greater than the positive judgment threshold is recorded as a positive event, and a real-time acceleration less than the negative judgment threshold is recorded as a negative event. Only positive and negative events within the most recent 5 seconds are retained using a sliding window method. The formula for calculating the acceleration trend is: D = N pos - N neg ; Where D represents the acceleration trend. N pos This represents the number of positive events within the statistical time window. N neg The number of negative events within the statistical time window; When D is greater than zero, the acceleration trend is considered positive; when D is less than zero, the acceleration trend is considered negative; when D equals zero, the current operating mode remains unchanged.
4. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, The difference between the maximum and minimum cadence is defined as the shift cadence difference value. The calculation formulas for the racing entry threshold and the normal entry threshold are as follows: Δn c = n c,max - n c,min ; n c,racein = n c,max - k sens × ½ × Δn c ; n c,normalin = n c,min + k sens × ½ × Δn c ; in, Δn c This is the shift cadence difference. n c,max For maximum cadence, n c,min Minimum cadence, n c,racein For racing entry threshold, n c,normalin This is the normal cut-in threshold. k sens This is the mode switching sensitivity coefficient, which is set by the user through an application.
5. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, Before calculating the target gear shift, the method further includes: putting the inertial measurement unit into a low-power state and monitoring the real-time acceleration in the forward direction; When the real-time acceleration is greater than the pre-start acceleration threshold, a pre-start state is marked, and the dynamic effective acceleration is calculated using the following formula: a dynamic = a x - g × sinθ; in, a dynamic For dynamic and effective acceleration, a x The real-time acceleration is g, the gravitational acceleration is θ, and the real-time slope angle is θ. When the dynamic effective acceleration is greater than the pre-start acceleration threshold, the pre-start state is confirmed to be valid; otherwise, the pre-start state is cleared. When the pre-start state is valid and valid guide wheel detection data is collected, automatic speed change is activated and the current operating mode is set to normal mode.
6. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, The acquisition of real-time pedal frequency includes: alternately attaching N-pole permanent magnets and S-pole permanent magnets along the circumferential direction on the inner ring surface of the bicycle guide wheel, and using a unipolar Hall sensor to output a pulse signal to one of the effective magnetic poles of the N-pole permanent magnets and S-pole permanent magnets. The number of effective pulses output by the unipolar Hall sensor within the statistical measurement time window is calculated, and the real-time pulse frequency, the equivalent number of pulses per crank rotation, and the real-time cadence are calculated using the following formula: f pulse = X / T; N pulse,cr = N eff × Z cr / Z gw ; n c = f pulse / N pulse,cr ; in, f pulse X represents the real-time pulse frequency, X represents the number of effective pulses within the measurement time window, and T represents the length of the measurement time window. N pulse,cr This represents the equivalent number of pulses per revolution of the dental disc. N eff The number of effective magnetic poles per revolution of the guide wheel. Z cr The number of teeth on the dental disc. Z gw The number of guide wheel teeth, n c For real-time cadence; The pulse frequency corresponding to the highest cadence of the system is determined as the maximum permissible pulse frequency, and calculated using the following formula: f pulse,max = n c,max × N pulse,cr ; in, f pulse,max The maximum permissible pulse frequency, n c,max This is the system's highest cadence; When the real-time pulse frequency is greater than the maximum allowable pulse frequency, the data within the measurement time window is determined to be abnormal and discarded, and the real-time cadence obtained from the previous measurement time window is used.
7. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, The attitude angular velocity output by the inertial measurement unit includes the real-time yaw angular velocity; When the absolute value of the real-time yaw rate is greater than the turning determination threshold, the bicycle is determined to be in a turning state, and upshifting and downshifting are prohibited in the turning state. When the absolute value of the real-time yaw rate falls below the turning determination threshold, upshifting and downshifting resume.
8. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, The real-time acceleration is collected within a set turbulence determination window. The difference between the maximum and minimum values of the real-time acceleration within the turbulence determination window is determined as the acceleration fluctuation amplitude, calculated using the formula: R = a max - a min ; Where R is the acceleration fluctuation amplitude, a max The maximum value of the real-time acceleration within the bump determination window. a min The minimum value of the real-time acceleration within the bump determination window; When the acceleration fluctuation amplitude is greater than the bump detection threshold, it is determined to be a bumpy road condition, and the operation mode switching and upshifting / downshifting are suspended. When the acceleration fluctuation amplitude falls below the bump determination threshold, the operation mode switching and upshifting / downshifting are resumed.
9. The multi-mode adaptive automatic transmission control method for bicycles according to claim 1, characterized in that, Within a set statistical period, a moving average is taken from the real-time cadence to obtain a comfort cadence estimate. The mode coefficient of the current operating mode is then corrected online based on the deviation between the comfort cadence estimate and the reference cadence. The correction formula is as follows: k mode ′ = k mode + β ×( n comfort - n ref ) / n ref ; in, k mode ′ represents the corrected mode coefficient. k mode The mode coefficients are before correction, and β is the adaptive gain. n comfort For the aforementioned comfort cadence estimation, n ref For reference cadence; The modified mode coefficients are limited to a preset range of mode coefficient values, and the modified mode coefficients are used to replace the mode coefficients of the current operating mode in the equivalent cadence calculation.
10. The multi-mode adaptive automatic transmission control method for bicycles according to claim 6, characterized in that, The length of the measurement time window is dynamically adjusted based on the real-time cadence, so that the length of the measurement time window is negatively correlated with the real-time cadence. That is, the larger the real-time cadence, the smaller the length of the measurement time window, and the smaller the real-time cadence, the larger the length of the measurement time window. The length of the measurement time window is limited to a preset window length value range.
Citation Information
Patent Citations
Bicycle speed variator, bicycle, and method and device for speed change of bicycle
CN104590479A
Bicycle derailleur control apparatus
CN110450900A
