Self-adaptive stabilizing method and system for scooter head tilting and anti-rolling single-wheel walking

Abstract

This invention discloses an adaptive stabilization method and system for single-wheel scooter wheelies and anti-rollover riding. The method includes: predicting the fundamental vibration frequency using the rear wheel speed, guiding the Goertzel algorithm to extract the dominant frequency and amplitude, generating a reverse harmonic current injected into the motor through adaptive filtering to actively cancel high-frequency mechanical vibration; obtaining smooth pitch and roll angles using complementary filtering; constructing a virtual potential function containing nonlinear exponential decay damping and pitch-roll coupling terms, deriving a model-free equilibrium control law, and generating a basic equilibrium torque; designing a stability coefficient that adaptively changes with pitch, roll, and current torque, and generating an anti-rollover compensation torque by combining the proportional-differential combination of roll angles; estimating the equivalent center of mass distance online to compensate for the static error of center of gravity movement, and canceling steering disturbances in a feedforward manner. This application eliminates the interference of vibration on attitude measurement through active vibration suppression, the virtual potential function is independent of physical parameters and adaptable to different riders, the anti-rollover and pitch control are deeply coupled and do not conflict, the adaptive compensation of the center of mass eliminates steady-state errors, the computational load is low, and it significantly improves the stability and versatility of single-wheel scooter wheelies.

Description

Technical Field

[0001] This invention relates to the field of scooter control technology, specifically to a method and system for adaptive stabilization of scooter wheelies and anti-tilt single-wheel walking. Background Technology

[0002] Existing electric scooters or self-balancing scooters typically employ PID or LQR controllers with conventional two-wheel support, using gyroscopes and accelerometers to measure pitch angles for balance control. However, when the user rides in a "wheel-drive" posture (i.e., front wheel off the ground, rear wheel only supporting the vehicle), the system becomes a highly nonlinear inverted pendulum model. Existing technologies suffer from the following two core drawbacks: First, it lacks the ability to actively suppress high-frequency mechanical vibrations. During rapid acceleration of the motor, the friction between the rear wheels and the ground generates high-frequency vibrations of 100-400Hz, which is close to the characteristic frequency band (<10Hz) of the vehicle's pitch motion. Existing solutions typically use low-pass filters (cutoff frequency 10-20Hz) to filter IMU data, but low-pass filtering also attenuates the effective pitch signal, introducing phase lag (typically 20°-30°), resulting in sluggish controller response and, in severe cases, instability. Furthermore, this vibration frequency continuously changes with vehicle speed and motor speed, belonging to time-varying narrowband vibration, making it difficult for existing fixed-parameter filtering schemes to adaptively track and suppress it.

[0003] Secondly, the controller parameters are fixed and cannot adapt to the real-time movement of the rider's center of gravity. When riding with the head tilted up, the rider's body moves back and forth, causing a change in the position of the center of gravity of ±0.2 meters. Traditional PID or LQR controllers rely on pre-calibrated physical parameters such as mass and moment of inertia. Once these parameters are fixed, they cannot adapt to this time-varying characteristic, leading to increased steady-state error or oscillations. Existing methods lack online adaptive adjustment capabilities.

[0004] Therefore, there is an urgent need for a balance control method for wheelie scooters that can actively suppress mechanical vibrations and has the ability to adapt the center of mass without relying on an accurate model. Summary of the Invention

[0005] In view of the problems mentioned in the background art, the technical problem to be solved by the present invention is to provide a method and system for adaptive stability of single-wheel walking of a scooter to prevent wheelies and tilting.

[0006] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows: A dynamic balance control method for a hoverboard scooter based on active vibration injection and virtual potential function control includes the following steps: Step S1: Acquire sensor data, including the pitch angle output by the inertial measurement unit. Pitch angular velocity Roll angle Roll rate The rear wheel speed output by the motor encoder The motor phase current output by the current sensor and the yaw rate output by the steering angle sensor. ; Step S2: Generate active vibration suppression current. Extract mechanical vibration characteristics based on the motor phase current, and use the acceleration signal output by the inertial measurement unit to generate a harmonic current command that is opposite to the phase of the mechanical vibration using an adaptive filtering algorithm. This command is then superimposed on the given value of the motor current loop. Step S3: Calculate the state estimate and calculate the vehicle speed based on the rear wheel speed. Complementary filtering is applied to the pitch and roll angles output by the inertial measurement unit to obtain a smoothed pitch angle. Roll angle and its angular velocity , ; Step S4: Calculate the virtual potential function's fundamental equilibrium torque and construct a smooth pitch angle. and pitch angular velocity The virtual energy-type potential function is used as the independent variable. Based on this potential function, the equilibrium control law is derived, and the roll-pitch coupling term is added to obtain the basic equilibrium torque. Its characteristic is that the balance control law is: ,in It is an integral term; Step S5: Calculate the anti-roll compensation moment based on the smooth roll angle. Smooth roll rate Smooth pitch angle and the current basic equilibrium torque According to the formula Calculate the overall stability coefficient, then multiply it by the proportional-differential combination of the roll angle and its angular velocity. Generate additional longitudinal acceleration command Then, through the scaling factor of the vehicle longitudinal dynamics model Convert the acceleration command into anti-roll compensation torque Finally, the compensation torque is limited. The design of the comprehensive stability coefficient follows the principle of 'pitch priority': when the basic balance torque is close to the saturation limit, the anti-roll intervention intensity is automatically reduced to ensure that the pitch stability of single-wheel travel is not compromised. Step S6: Synthesize and output the final total target torque. Perform algebraic summation of the above torques on the motor torque components to obtain the final total target torque. And output to the motor driver.

[0007] Preferably, the mechanical vibration characteristics described in step S2 are extracted using Goertzel single-frequency transform.

[0008] Preferably, the adaptive filtering algorithm in step S2 is an orthogonal demodulation minimum mean square algorithm, which updates the injected phase in real time by generating two orthogonal reference signals.

[0009] Preferably, the virtual energy potential function in step S4 includes a pitch rate squared term, a pitch angle squared term, and a nonlinear exponentially decaying cross term resulting from the product of the pitch angle and pitch rate; the roll-pitch coupling term is proportional to the product of the roll angle and roll rate, and decays exponentially as the absolute value of the pitch angle increases.

[0010] Preferably, the method also includes a virtual center-of-mass compensation torque calculation step: based on the current smooth pitch angle, pitch velocity, and basic equilibrium torque, the estimated value of the equivalent center-of-mass distance is updated online using an adaptive law based on torque error; then, the virtual compensation torque is calculated based on the total mass, gravitational acceleration, and the difference between the estimated center-of-mass distance and the expected value, as one of the additional optional torques in step S6; the adaptive law is: in For the equivalent centroid distance, For adaptive gain, This represents the desired distance.

[0011] Preferably, the method also includes a step for calculating the steering feedforward compensation torque: calculating the total disturbance torque based on the centrifugal force effect and the gyroscopic effect. The negative value of the total disturbance torque is taken as the feedforward compensation torque and is used as one of the additional optional torques in step S6. The feedforward compensation torque is used to suppress the main pitch disturbance component caused by steering and can work with the rider's active control to maintain the balance of a single wheel.

[0012] A dynamic balance control system for a hoverboard scooter based on active vibration injection and virtual potential function control, used to implement the above-mentioned method, includes: a main control microcontroller, a six-axis inertial measurement unit, a rear wheel brushless motor and its field orientation controller, an optical encoder, a steering angle sensor, a current sampling circuit, and an active vibration injection module, a complementary filter state estimation module, a virtual potential function control module, and an anti-tilt acceleration compensation module running in the main control microcontroller.

[0013] Preferably, the active vibration injection module periodically performs Goertzel transformation to extract vibration frequencies and uses historical data during non-calculation periods.

[0014] Preferably, the exponential function and hyperbolic tangent function involved in the virtual potential function control module are implemented using linear interpolation with lookup tables.

[0015] Preferably, the anti-roll acceleration compensation module includes a roll angle complementary filtering submodule, a comprehensive stability coefficient calculation submodule, an additional acceleration command generation submodule, and a torque limiting submodule, and each submodule is executed sequentially in each control cycle.

[0016] Compared with the prior art, the present invention has the following advantages: The present application obtains attitude angle and angular velocity through inertial measurement unit, extracts vibration frequency by Goertzel transform and injects reverse harmonic current to actively suppress vibration, adopts complementary filtering to smooth state estimation, derives model-free equilibrium control law by virtual potential function, and simultaneously achieves anti-roll based on roll state and basic equilibrium torque adaptively adjusting longitudinal acceleration. Attached Figure Description

[0017] The present invention will be further described in detail below with reference to the accompanying drawings and preferred embodiments. However, those skilled in the art will understand that these drawings are drawn only for the purpose of explaining the preferred embodiments and therefore should not be regarded as a limitation on the scope of the present invention. In addition, unless specifically indicated, the drawings are only schematic representations of the composition or structure of the described objects and may contain exaggerated displays, and the drawings are not necessarily drawn to scale.

[0018] Figure 1 A flowchart illustrating the steps of an adaptive stabilization method for single-wheel walking to prevent scooter tipping and tilting. Detailed Implementation

[0019] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Those skilled in the art will appreciate that these descriptions are merely descriptive and exemplary and should not be construed as limiting the scope of protection of the present invention.

[0020] It should be noted that similar labels in the following figures indicate similar items; therefore, once an item is defined in one figure, it may not be further defined and explained in subsequent figures. Example

[0021] This embodiment mainly describes the title of the adaptive stabilization method and system for scooter wheelies and anti-tilt single-wheel walking, as follows: The adaptive stabilization method for scooter wheelies and anti-tilt single-wheel walking includes the following steps: Step S1: Acquire sensor data In this step, the following data is collected in real time using a six-axis inertial measurement unit (IMU) mounted at the center of gravity of the vehicle frame, an optical encoder at the rear wheel axle, a three-channel shunt resistor current sampling circuit, and a Hall effect angle sensor mounted on the steering handlebar central shaft: pitch angle. and pitch angular velocity Roll angle and roll rate Rear wheel speed Three-phase current of motor , , yaw rate The above data provides the basic input for subsequent pitch balancing, roll suppression, vibration suppression, and steering compensation.

[0022] Step S2: Generate active vibration suppression current Vibration can severely contaminate IMU measurement signals. Since the mechanical vibration of the scooter's rear wheel drive motor is mainly caused by cogging torque pulsations and current commutation harmonics, its fundamental vibration frequency has a definite integer multiple relationship with the motor speed. Therefore, to suppress the high-frequency mechanical vibration generated during rapid motor acceleration, the rear wheel speed obtained in step S1 is first used... By combining the known number of pole pairs and the vibration characteristic order of the motor—which typically corresponds to the number of stator slots or the current harmonic order—the fundamental vibration frequency under the current operating condition is predicted and used as the detection target for the Goertzel algorithm. Then, a single-frequency transformation is performed at this predicted frequency point, thereby quickly and accurately extracting the true dominant frequency of the mechanical vibration without performing a full-band scan. and its amplitude Traditional FFT computation is computationally intensive and unsuitable for embedded real-time systems. In contrast, the Goertzel algorithm requires only two multiply-accumulate operations per sampling point for single-frequency detection, completing 256-point spectrum analysis within 20ms and accurately extracting the current dominant frequency of mechanical vibration. and amplitude The Goertzel transformation is performed every 20 control cycles. In non-execution cycles, the frequency and amplitude obtained from the most recent calculation are used to significantly reduce CPU load.

[0023] Subsequently, the Z-axis acceleration signal output by the IMU is transmitted through a... The residual vibration signal was obtained by using a second-order IIR bandpass filter with a center frequency and a bandwidth of 10Hz. The injected phase is updated in real time using the least mean square (LMS) algorithm with orthogonal demodulation: generating two orthogonal reference signals. and ,according to , Update coefficients, step size factor Set to 0.05; Inject phase Ultimately, harmonic current commands are generated. Injection gain To further avoid overcompensation under low-speed, low-vibration conditions and insufficient vibration suppression under high-speed, high-vibration conditions, the injection gain can be adaptively and dynamically adjusted according to the real-time extracted vibration amplitude. Specifically, a minimum injection gain (e.g., 0.2), a maximum injection gain (e.g., 1.0), and a reference vibration amplitude (e.g., 0.5 A) are preset. When the extracted vibration amplitude is much smaller than the reference amplitude, it indicates that the current vibration is slight, and the injection gain automatically approaches the minimum value, applying only a weak reverse harmonic current to avoid generating unnecessary high-frequency torque during normal driving and affecting the riding feel. When the vibration amplitude gradually increases and approaches or exceeds the reference amplitude, the injection gain smoothly transitions to the maximum value to fully counteract the mechanical vibration. The change in gain with the vibration amplitude can be smoothly transitioned using a hyperbolic tangent function or piecewise linear interpolation. The calculation of the hyperbolic tangent function can reuse the existing lookup table resources of this controller without adding extra computational burden. This instruction is superimposed on the q-axis current setpoint of the field-oriented control (FOC), causing the motor to actively generate a harmonic torque that is out of phase and proportional in amplitude to the mechanical vibration, thereby canceling out the mechanical vibration. Active vibration suppression reduces the peak-to-peak noise of the IMU accelerometer, significantly reduces the pitch angle measurement variance, and avoids controller malfunctions caused by vibration; the increase in CPU peak load during this process is relatively small.

[0024] Step S3: Calculate the state estimate IMU gyroscope integration results in long-term drift, while accelerometers are statically accurate but have high dynamic noise. This step employs a complementary filter with extremely low computational cost to combine the advantages of both. Compared to the extended Kalman filter (EKF), the complementary filter requires only 4 multiply-accumulate operations and 2 addition operations per cycle, reducing the computational cost by more than an order of magnitude, making it more suitable for low-cost microcontrollers.

[0025] The specific calculation process is as follows: First, based on the rear wheel speed output by the motor encoder... and rear wheel radius Calculate vehicle speed in meters This is used for subsequent steering feedforward and anti-roll modules. Complementary filtering is applied to the pitch angle. Filter coefficients The corresponding cutoff frequency is approximately 3.2Hz, and the control period is... seconds, initialization Pitch angular velocity is directly measured using gyroscope values. Complementary filtering is also applied to the roll angle: Roll angular velocity The smoothed angle eliminates high-frequency noise and drift, preventing the controller from jittering due to noise.

[0026] Step S3: Calculate the state estimate IMU gyroscope integration results in long-term drift, while accelerometers are statically accurate but have high dynamic noise. This step employs a complementary filter with extremely low computational cost to combine the advantages of both. Compared to the extended Kalman filter (EKF), the complementary filter requires only 4 multiply-accumulate operations and 2 addition operations per cycle, reducing the computational cost by more than an order of magnitude, making it more suitable for low-cost microcontrollers.

[0027] The specific calculation process is as follows: First, based on the rear wheel speed output by the motor encoder... and rear wheel radius Calculate vehicle speed in meters This is used for subsequent steering feedforward and anti-roll modules. Complementary filtering is applied to the pitch angle. Filter coefficients The corresponding cutoff frequency is approximately 3.2Hz, and the control period is... seconds, initialization Pitch angular velocity is directly measured using gyroscope values. Complementary filtering is also applied to the roll angle: Roll angular velocity The smoothed angle eliminates high-frequency noise and drift, preventing the controller from jittering due to noise.

[0028] right Find the time derivative, let As the desired dissipative form, and by adding a roll-pitch coupling term to utilize the gyroscopic effect to aid in roll suppression, the balance control law is obtained: Among them, the integral term Used to eliminate static error, defined as For a tilting equilibrium, the desired angular velocity is... The discretized form is simplified to And set a limit To prevent integral saturation.

[0029] Preferably, the values ​​of each parameter are: proportionality coefficients N·m / rad determines the system stiffness; nonlinear damping coefficient N·m·s / rad; exponential decay coefficient rad⁻¹; linear damping coefficient N·m·s / rad; integral coefficient N·m / rad; Coupling coefficient N·m·s / rad²; Coupling attenuation coefficient . The fundamental balancing torque directly drives the motor to maintain pitch stability and is the core instruction of the control system.

[0030] Furthermore, to eliminate minute angular jitter that may be caused by sensor quantization errors or dead zones in the transmission system friction, this step may also employ at least one of the following two optimization measures: First, a torque dead zone is set. Before outputting the basic balance torque, a dead zone threshold (e.g., 0.5 N·m) is set. When the absolute value of the calculated torque is less than this threshold, zero torque is forced to be output to avoid the controller over-responding to extremely small errors and to prevent frequent output of micro-torque near the ideal equilibrium point, which would cause jitter. When the absolute value of the torque exceeds the threshold, the torque value is reduced by the threshold before output to ensure the continuity of the control action.

[0031] Secondly, enhanced damping at small angles. The original fixed nonlinear damping coefficient is replaced with a variable damping coefficient that varies with the pitch angle amplitude. When the pitch angle is extremely small (e.g., less than about 1.7 degrees), the damping coefficient is temporarily increased to 2 to 3 times the normal value (e.g., from 10 N·m·s / rad to 25 N·m·s / rad) to enhance the ability to suppress small jitters; when the pitch angle exceeds this small angle range, the damping coefficient smoothly returns to the normal value, ensuring that the dynamic response at large angles is not affected.

[0032] It should be noted that in the virtual potential function control law proposed in this invention, when the pitch angle approaches zero, the nonlinear exponential decay cross-damping term in the control law automatically degenerates into a standard linear damping term. The function is a continuous and smooth function near the zero point, and there are no switching or discontinuous links. Theoretically, under the first-order approximation, it does not have the conditions to generate self-excited oscillations.

[0033] Step S5: Calculate the anti-roll compensation moment coupled with pitch control depth. The basic principle of anti-roll is to generate a precession torque that resists roll by changing the rear wheel speed and utilizing the gyroscopic effect. However, if the pitch angle is large or the foundation balancing torque is already high, excessive anti-roll intervention may lead to pitch instability. Therefore, this step designs a comprehensive stability coefficient that adaptively varies with the pitch angle, roll angle, and foundation balancing torque. : in, , , , This coefficient is 1 under undisturbed conditions and can reach approximately 2.2 under moderate disturbance conditions, allowing the anti-roll strength to automatically increase as needed. Under extreme conditions where the base balance torque is close to saturation, the anti-roll command is automatically reduced through the negative adjustment of κ3, ensuring priority is given to pitch stability. Preferably, the coefficient... The basic balancing torque is positive when it is not saturated to enhance its disturbance rejection capability; when the basic balancing torque approaches the saturation threshold, a nonlinear saturation function is introduced into the system to make... The term exhibits negative adjustment characteristics in the saturation region, thereby automatically reducing anti-roll commands.

[0034] Then, a proportional-derivative acceleration command is used: roll angle. Reflects the roll rate and the degree of roll. Reflecting the roll trend, the linear combination of the two can provide an early response, with the additional longitudinal acceleration command being... ,in , The dimensions of the combination are correct.

[0035] Based on vehicle longitudinal dynamics Rear wheel moment of inertia Rear wheel radius Total mass The acceleration-torque conversion coefficient is obtained. Anti-roll compensation moment And limit the width: This torque is generated by the motor driving the rear wheels to accelerate and decelerate, changing their rotational speed and utilizing the gyroscopic effect to counteract body roll.

[0036] Furthermore, to ensure priority is given to pitch stability, when the foundation equilibrium torque... When the threshold value is exceeded (e.g., 25 N·m), the system will dynamically reduce the overall stability coefficient. The gain is increased until the anti-tilt compensation torque returns to zero, ensuring that all motor torque is used to maintain pitch balance.

[0037] Step S6: Calculate and output the final total target torque. The basic equilibrium torque Anti-roll compensation moment And optional virtual centroid compensation torque and steering feedforward compensation torque By summing them up, we obtain the final total target torque. This instruction is converted into a q-axis current via FOC, and simultaneously converts the active vibration suppression current generated in step S2. The current is superimposed on the q-axis current setpoint and driven by SVPWM modulation. The torques do not conflict with each other at the summation node, thus achieving multi-objective coordinated control.

[0038] Furthermore, this method can selectively enable at least one of the following two modules to further improve performance: (a) Virtual centroid compensation Based on the current smooth pitch angle, pitch velocity, and foundation equilibrium torque, an adaptive law based on torque error is used to update the estimated equivalent centroid distance online. : Adaptive gain , , Expected virtual centroid distance Virtual compensation torque This torque can counteract the static error caused by the rider's center of gravity shift, and can be added to the total target torque as an optional additional torque.

[0039] (ii) Steering feedforward compensation Calculate the total disturbance torque based on the centrifugal force effect and the gyroscopic effect: ,in For vehicle speed, Yaw angular velocity, This is the distance from the center of gravity to the rear wheel axle. Rear wheel speed (unit: rad / s), correction factor , Feedforward compensation torque and apply saturation limiting. It can suppress pitch fluctuations caused by steering and can be added to the total target torque as an optional additional torque.

[0040] To reduce the computational burden on the controller, this embodiment also employs the following optimization measures: exponential function. Linear interpolation using lookup tables, lookup table range radians, step size Hyperbolic Tangent Lookup range Step length Sine and cosine share the same sheet A lookup table is used, reducing the execution time of the transcendental function to less than 1 microsecond. The Goertzel algorithm executes every 20ms, using historical data in non-execution cycles; complementary filtering replaces the EKF; the integral term is limited; and the STM32F407's built-in FPU is used for acceleration. The anti-roll acceleration compensation module is internally divided into a roll angle complementary filtering submodule, a comprehensive stability coefficient calculation submodule, an additional acceleration command generation submodule, and a torque limiting submodule, each of which executes sequentially in each control cycle.

[0041] Correspondingly, this embodiment also provides a single-wheel adaptive stabilization system for anti-tilt walking of a scooter that implements the above method, including: a main control microcontroller, a six-axis inertial measurement unit, a rear wheel brushless motor and its magnetic field orientation controller, a photoelectric encoder, a steering angle sensor, a current sampling circuit, and an active vibration injection module, a complementary filter state estimation module, a virtual potential function control module, and an anti-tilt acceleration compensation module running in the main control microcontroller.

[0042] The main control microcontroller is a microcontroller with a built-in floating-point unit (FPU), whose operating frequency meets the real-time computing requirements. The six-axis inertial measurement unit is installed in the middle section of the frame, corresponding to the estimated system center of gravity position determined based on a rider of standard weight in a standard wheelie balance posture, and is connected to the main control microcontroller via a serial communication bus. The rear wheel brushless motor is installed in the rear wheel hub and is driven by the field-oriented controller (FOC). The photoelectric encoder is installed at the end of the motor shaft and is used to output orthogonal differential signals to obtain the motor speed and position. The steering angle sensor is a non-contact angle sensor installed on the central shaft of the steering handlebar. The current sampling circuit includes a shunt resistor and a differential amplifier, used to collect the three-phase current of the motor and perform signal conditioning to meet the accuracy requirements of high-frequency current analysis.

[0043] The software modules of this system have clearly defined signal input / output and hierarchical coupling relationships. The complete signal flow logic is as follows: Raw data from all hardware sensors, including pitch angle, pitch rate, roll angle, roll rate, rear wheel speed, three-phase motor current, and yaw rate, are synchronously input to the main microcontroller. Pitch and roll attitude data, along with rear wheel speed, are fed into a complementary filtering state estimation module as basic input. This module calculates vehicle speed and smooths the attitude data, then outputs smoothed pitch angle, smoothed pitch rate, smoothed roll angle, smoothed roll rate, and vehicle speed signals. These signals are output in two paths: one path is transmitted to the virtual potential function control module for calculating the basic balance torque. Another path transmits the data to the anti-roll acceleration compensation module to participate in the calculation of the comprehensive stability coefficient and the additional acceleration command.

[0044] The fundamental equilibrium torque output by the virtual potential function control module There is bidirectional signal interaction: on the one hand, the total torque superposition node is directly fed as the core control component, and on the other hand, it is input in reverse to the anti-roll acceleration compensation module, which is used to adaptively adjust the anti-roll intervention intensity in real time in combination with pitch and roll states, so as to realize the coupled control logic that prioritizes pitch stability.

[0045] The three-phase current of the motor and the rear wheel speed are fed separately into a parallel active vibration injection module. This module extracts mechanical vibration characteristics and generates vibration-suppressing harmonic currents using an adaptive filtering algorithm. It is independently output to the current setpoint superposition terminal of the FOC current loop and acts in parallel with the torque control branch.

[0046] The system also includes two sets of optional extended compensation computation units: the virtual centroid compensation unit takes the smooth pitch attitude output by the complementary filter and the basic equilibrium torque output by the virtual potential function module as input, and outputs the virtual centroid compensation torque. The steering feedforward compensation unit takes the vehicle speed output from the complementary filter, the yaw rate collected by the steering sensor, and the rear wheel speed fed back by the photoelectric encoder as inputs, and outputs a steering feedforward compensation torque. .

[0047] Basic equilibrium torque Anti-roll compensation moment Virtual centroid compensation torque Steering feedforward compensation torque The total target torque is obtained by performing algebraic accumulation at the front end of the FOC current loop. After the total target torque is converted into the given q-axis base current of the motor, the vibration suppression harmonic current output by the active vibration injection module is then superimposed. Finally, the SVPWM modulation output drives the rear wheel brushless motor FOC driver; the motor is equipped with a photoelectric encoder to collect the rear wheel speed in real time and transmit it back to the front end of the system, completing the closed-loop signal feedback of the whole process.

Claims

1. A method for adaptive stabilization of a scooter's wheelie and anti-tilt single-wheel movement, characterized in that, Includes the following steps: Step S1: Acquire sensor data, including the pitch angle output by the inertial measurement unit. Pitch angular velocity Roll angle Roll rate The rear wheel speed output by the motor encoder The motor phase current output by the current sensor and the yaw rate output by the steering angle sensor. ; Step S2: Generate active vibration suppression current. Extract mechanical vibration characteristics based on the motor phase current, and use the acceleration signal output by the inertial measurement unit to generate a harmonic current command that is opposite in phase to the mechanical vibration using an adaptive filtering algorithm. This command is then superimposed on the given value of the motor current loop. Step S3: Calculate the state estimate, calculate the vehicle speed based on the rear wheel speed, and perform complementary filtering on the pitch and roll angles output by the inertial measurement unit to obtain the smoothed pitch angle. Roll angle and its angular velocity 、; Step S4: Calculate the virtual potential function's fundamental equilibrium torque and construct a smooth pitch angle. and pitch angular velocity The virtual energy-type potential function is used as the independent variable. Based on this potential function, the equilibrium control law is derived, and a roll-pitch coupling term is added to obtain the basic equilibrium torque. ; Its characteristic is that the balance control law is: ,in It is an integral term; Step S5: Calculate the anti-roll compensation moment based on the smooth roll angle. Smooth roll rate Smooth pitch angle and the current basic equilibrium torque According to the formula Calculate the overall stability coefficient, then multiply it by the proportional-differential combination of the roll angle and its angular velocity. An additional longitudinal acceleration command is generated, and then the acceleration command is converted into an anti-roll compensation torque through the scaling factor of the vehicle's longitudinal dynamics model. Finally, the compensation torque is limited. Step S6: Synthesize and output the final total target torque. The basic balancing torque, anti-tilt compensation torque, and additional optional torque are algebraically added to the motor torque components to obtain the final total target torque, which is then output to the motor driver.

2. The adaptive stabilization method for scooter wheelies and anti-tilt single-wheel walking as described in claim 1, characterized in that, The mechanical vibration features described in step S2 are extracted using Goertzel single-frequency transform.

3. The adaptive stabilization method for scooter wheelies and anti-tilt single-wheel walking as described in claim 1, characterized in that, The adaptive filtering algorithm described in step S2 is an orthogonal demodulation minimum mean square algorithm, which updates the injected phase in real time by generating two orthogonal reference signals.

4. The adaptive stabilization method for scooter wheelies and anti-tilt single-wheel walking as described in claim 1, characterized in that, The virtual energy potential function in step S4 includes a pitch rate squared term, a pitch angle squared term, and a nonlinear exponentially decaying cross term resulting from the product of the pitch angle and pitch rate; the roll-pitch coupling term is proportional to the product of the roll angle and roll rate, and decays exponentially as the absolute value of the pitch angle increases.

5. The adaptive stabilization method for scooter wheelies and anti-tilt single-wheel walking as described in claim 1, characterized in that, It also includes a virtual center-of-mass compensation torque calculation step: based on the current smooth pitch angle, pitch velocity, and basic equilibrium torque, the estimated value of the equivalent center-of-mass distance is updated online using an adaptive law based on torque error; then, the virtual compensation torque is calculated based on the total mass, gravitational acceleration, and the difference between the estimated center-of-mass distance and the expected value, as one of the additional optional torques in step S6; the adaptive law is: ,in For the equivalent centroid distance, For adaptive gain, This represents the desired distance.

6. The adaptive stabilization method for scooter wheelies and anti-tilt single-wheel walking according to claim 1, characterized in that, It also includes a step for calculating the steering feedforward compensation torque: the total disturbance torque is calculated based on the centrifugal force effect and the gyro effect. The total disturbance torque is equal to the product of the first correction coefficient multiplied by the vehicle speed, yaw rate and center of gravity distance, plus the product of the second correction coefficient multiplied by the motor speed and yaw rate. The negative value of the total disturbance torque is taken as the feedforward compensation torque and is used as one of the additional optional torques in step S6.

7. A scooter anti-tilt single-wheel adaptive stabilization system, used to implement the scooter anti-tilt single-wheel adaptive stabilization method according to any one of claims 1 to 6, characterized in that, include: The system includes a main control microcontroller, a six-axis inertial measurement unit, a rear wheel brushless motor and its field orientation controller, an optical encoder, a steering angle sensor, a current sampling circuit, and an active vibration injection module, a complementary filter state estimation module, a virtual potential function control module, and an anti-roll acceleration compensation module running within the main control microcontroller.

8. The system according to claim 7, characterized in that, The active vibration injection module periodically performs Goertzel transformation to extract vibration frequencies and uses historical data during non-calculation periods.

9. The system according to claim 7, characterized in that, The exponential and hyperbolic tangent functions involved in the virtual potential function control module are implemented using linear interpolation with lookup tables.

10. The system according to claim 7, characterized in that, The anti-roll acceleration compensation module includes a roll angle complementary filtering submodule, a comprehensive stability coefficient calculation submodule, an additional acceleration command generation submodule, and a torque limiting submodule. Each submodule is executed sequentially in each control cycle.

Images