Control method and system suitable for one-way running motor of two-wheeled electric vehicle

Abstract

The invention relates to the technical field of motor control, and particularly provides a control method and system suitable for a one-way running motor of a two-wheeled electric vehicle. The method aims to solve the problems of unstable motor operation and insufficient reliability caused by torque pulsation in the prior art. The method comprises the following steps: executing off-line identification of motor parameters to obtain initialized parameters; multi-source operation state signals are collected, and a vehicle working condition judgment result is obtained through fusion analysis; performing fault diagnosis and grading evaluation to form a fault-tolerant instruction; dynamic dead zone compensation and self-adaptive overmodulation are implemented by combining the parameters and the working conditions, and an optimization control instruction is output; generating a PWM driving signal through space vector modulation; and responding to the fault-tolerant instruction, and fusing a sensorless algorithm and a health sensor signal to realize degraded operation. According to the invention, the smoothness and efficiency of motor operation and the fault-tolerant capability of the system are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motor control, and specifically provides a control method and system suitable for one-way running motor of two-wheeled electric vehicle. BACKGROUND

[0002] At present, two-wheeled electric vehicles generally use permanent magnet synchronous motors as driving units, and the control performance thereof directly determines the riding quality, energy efficiency and reliability of the whole vehicle. However, the traditional motor control method still has many bottlenecks when dealing with complex operating conditions. First, when climbing at low speed or starting under heavy load, the fixed dead time setting will cause significant current waveform distortion and torque ripple, causing vehicle vibration and running noise, which seriously affects the riding smoothness. Secondly, when the battery power is insufficient, the conventional modulation algorithm cannot maintain sufficient power output, causing the vehicle to stall, which poses a safety hazard. In addition, the fault tolerance capability of the existing control system is generally weak, and once the key sensor fails, the motor is often forced to stop, causing the user to be stranded in the middle of the ride, which greatly affects the use experience and system reliability.

[0003] Although the prior art attempts to optimize the algorithm or increase hardware redundancy to improve the above problems, there are often limitations. For example, the dead zone compensation strategy proposed by some schemes is relatively simple and cannot be adaptively adjusted in the full operating condition range, while the over-modulation strategy may damage the motor performance due to excessive harmonic distortion. In terms of fault-tolerant control, most schemes only realize fault diagnosis and lack a smooth degradation operation mechanism. Therefore, there is an urgent need in the art for a comprehensive solution that can deeply integrate state perception, adaptive control and intelligent fault tolerance to systematically improve the control quality and robustness of two-wheeled electric vehicle motors under complex actual operating conditions. SUMMARY

[0004] The purpose of the present application is to provide a control method and system suitable for one-way running motor of two-wheeled electric vehicle to solve the problems in the background art.

[0005] The purpose of the present application can be achieved by the following technical solutions: A control method suitable for one-way running motor of two-wheeled electric vehicle, comprising the following steps: Step 1: Perform motor parameter offline identification through system power-on self-test to obtain the initialization parameters of motor stator resistance and direct and quadrature axis inductance; Step 2: Collect multi-source operating state signals of handlebar command, phase current, bus voltage and vehicle body inclination, and generate current vehicle operating condition judgment result through vehicle state fusion analysis; Step 3: Complete fault diagnosis and hierarchical evaluation using real-time monitored multi-source operating state signals to form hierarchical fault-tolerant instructions; Step four: combine the initialization parameters with the current vehicle operating conditions to implement dynamic dead zone compensation and adaptive overmodulation algorithm, output optimized control instructions containing dynamic compensation amount; Step five: use optimized control instructions and dynamic compensation amount to complete space vector pulse width modulation processing, generate PWM drive signal to drive power inverter bridge; Step six: in response to the hierarchical fault-tolerant instructions, use sensorless estimation algorithm and residual healthy sensor signals to perform data fusion, and realize fault-tolerant degraded operation of the motor.

[0006] As preferred, the motor parameter offline identification in step one includes: Perform a complete system power-on self-test process, including comprehensive function verification of the controller internal memory, sampling circuit and power drive circuit; After the system power-on self-test process passes, inject a non-rotating test signal accurately controlled into the motor three-phase winding, the frequency of the test signal is set within a suitable proportion of the motor rated frequency; Collect the winding voltage and current transient response under the test signal excitation through a high-precision sampling module, and use high-frequency sampling to ensure data accuracy; Based on the transient response data, use recursive least squares parameter identification algorithm to calculate the motor stator resistance and cross-axis inductance parameters; Store the calculated initialization parameters in the dedicated parameter storage area of the non-volatile memory.

[0007] As preferred, the vehicle state fusion analysis in step two includes: Collect the analog voltage signal output by the handle through the sampling channel and perform multi-stage digital filtering processing; Real-time monitor the motor three-phase current sampling value and DC bus voltage sampling value, and use fixed sampling period to ensure data synchronization; Obtain three-axis acceleration and angular velocity data output by the inertial measurement unit through the communication interface; Based on the multi-source operating state signals, use Kalman filter data fusion algorithm for multi-sensor information fusion processing; Combine the pre-set state decision logic, which includes the judgment conditions of multiple typical working conditions, to output the accurate judgment result of the current vehicle working condition.

[0008] As preferred, the fault diagnosis and hierarchical evaluation in step three includes: Real-time monitor the integrity of the multi-source operating state signals, including signal continuity detection and reasonableness verification; Determine whether the phase current in the multi-source operating state signal exceeds the safety limit value and whether the DC bus voltage is within the normal working range; Detect the temperature state of the power device through the temperature monitoring system, using periodic sampling method; Based on the monitoring and judgment results, divide the system state into multiple levels according to the fault severity, and generate corresponding hierarchical fault-tolerant instructions.

[0009] As preferred, the dynamic dead zone compensation in step four includes: According to the real-time values of the motor speed and the DC bus voltage in the current vehicle operating condition, the optimal compensation amount is dynamically calculated through a pre-calibrated lookup table; based on the compensation amount, a variable compensation time is inserted in the space vector pulse width modulation signal, and the compensation time is set to a reasonable range according to the system characteristics; when a specific operating condition is detected, the strength of the compensation amount is automatically enhanced according to the real-time load condition, and a configurable enhancement coefficient is used; and the calculated dynamic compensation amount is output to the modulation process in real time.

[0010] Preferably, the adaptive over-modulation algorithm in step four includes: The real-time judgment is made on whether the DC bus voltage in the multi-source operating state signal meets the output voltage demand at the current speed; when voltage deficiency is detected, the control algorithm automatically and smoothly enters the over-modulation mode, and the transition process is kept stable; based on the optimal control instruction, a specific modulation algorithm is used to adjust the voltage vector trajectory, and the over-modulation degree is controlled within a reasonable range; and the total harmonic distortion rate of the modulation waveform is kept not to exceed the allowable value through harmonic analysis.

[0011] Preferably, the space vector pulse width modulation process in step five includes: The optimal control instruction is converted into a voltage vector reference value in the stationary coordinate system through coordinate transformation; the sector number in which the voltage vector is located is determined by using geometric analysis method; the action time of the basic vector and the zero vector distribution time are calculated based on the modulation principle, and the calculation process is kept accurate; based on the action time, a symmetric modulation strategy is used to generate multiple complementary drive signals, and the switching frequency is set to an appropriate value.

[0012] Preferably, the sensorless estimation algorithm in step six includes: An observer based on the motor model is constructed, and the observer parameters are set according to system requirements; based on the phase current and voltage sampling values in the multi-source operating state signal, the rotor position is estimated in real time by using numerical calculation method, and the estimation process is kept stable; the improved observer is used for rotor position tracking, and the observer gain is adaptively adjusted according to the operating state; and the fused rotor position information estimated is output to the control system through the communication interface.

[0013] Preferably, the data fusion in step six includes: The output of the sensorless algorithm and the remaining healthy sensor signal are adaptively weighted and fused, and the weight coefficient is dynamically adjusted according to the signal quality; an advanced filtering algorithm is used to improve the position estimation accuracy, and the filter parameters are optimized; based on the hierarchical fault-tolerant instruction, the fusion weight is dynamically adjusted, and the fusion strategy is automatically adjusted when the sensor performance decreases; when the sensor completely fails, it is smoothly switched to a pure sensorless operating mode, and the motor output capability is appropriately limited.

[0014] As preferred, a control system suitable for one-way operation motor of two-wheel electric vehicle, comprising parameter self-learning and initialization module, multi-source information fusion and state decision module, fault diagnosis and fault-tolerant management module, dynamic core control algorithm module, space vector pulse width modulation module and fault-tolerant operation and signal fusion module; The parameter self-learning and initialization module performs motor parameter offline identification through system power-on self-test to obtain initialization parameters of motor stator resistance and direct and quadrature axis inductance; The multi-source information fusion and state decision module collects multi-source operating state signals of handlebar instructions, phase currents, bus voltages and vehicle body inclination, and generates current vehicle working condition judgment results through vehicle state fusion analysis; The fault diagnosis and fault-tolerant management module completes fault diagnosis and hierarchical evaluation using real-time monitored multi-source operating state signals to form hierarchical fault-tolerant instructions; The dynamic core control algorithm module implements dynamic dead-time compensation and adaptive over-modulation algorithm in combination with initialization parameters and current vehicle working conditions to output optimized control instructions containing dynamic compensation; The space vector pulse width modulation module uses optimized control instructions and dynamic compensation to complete space vector pulse width modulation processing to generate PWM drive signals of the drive power inverter bridge; The fault-tolerant operation and signal fusion module responds to hierarchical fault-tolerant instructions, uses sensorless estimation algorithm and residual healthy sensor signals to perform data fusion to realize fault-tolerant degraded operation of the motor.

[0015] The beneficial effects of the present application are: The present application introduces a dynamic dead-time compensation mechanism, effectively solving the problem of insufficient control accuracy caused by traditional fixed dead-time under complex working conditions. This technology can automatically adjust the compensation according to real-time speed and load conditions, significantly improve the current waveform quality under low-speed high-torque working conditions, fundamentally suppress the torque ripple phenomenon, make the motor run more smoothly under starting, climbing and other working conditions, effectively reduce the running noise, and greatly improve the comfort and smoothness of riding.

[0016] The adaptive over-modulation technology and efficiency optimization strategy used in the present application can achieve the best balance between power and economy in the full working condition range. The system can intelligently sense the DC bus voltage state, automatically enable the over-modulation mode when the voltage drops to maintain the continuity of power output, and avoid the risk of power interruption during vehicle driving. At the same time, by intelligently distributing the direct-axis and quadrature-axis current components, the running efficiency of the motor under different loads is optimized, the power performance is guaranteed, the energy utilization rate is significantly improved, and the vehicle range is effectively extended.

[0017] The perfect fault diagnosis and fault-tolerant operation mechanism established by the application significantly improves the reliability and safety of the system. Through multi-level fault monitoring and intelligent grading strategy, the system can quickly identify abnormal state of the sensor, and based on the deep fusion of sensorless algorithm and healthy sensor data, it realizes smooth mode switching under fault condition. This intelligent fault-tolerant architecture ensures that the system can still maintain degraded operation when the key sensor fails, providing reliable limp-home function for users, greatly improving the practical value and safety protection ability of the system. BRIEF DESCRIPTION OF DRAWINGS

[0018] Fig. 1 A flowchart of a control method suitable for a one-way running motor of a two-wheeled electric vehicle according to the application.

[0019] Fig. 2 A structural diagram of a control system suitable for a one-way running motor of a two-wheeled electric vehicle according to the application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the application will be described in detail below with reference to the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0021] Embodiment: Please refer to Figs. 1-2 A control method suitable for a one-way running motor of a two-wheeled electric vehicle, as shown in the figure, comprises the following steps: Step 1: Perform motor parameter offline identification through system power-on self-test to obtain the initialization parameters of motor stator resistance and direct and quadrature axis inductance; Step 2: Collect multi-source running state signals of handlebar command, phase current, bus voltage and vehicle body inclination, and generate current vehicle working condition judgment result through vehicle state fusion analysis; Step 3: Use real-time monitored multi-source running state signals to complete fault diagnosis and grading evaluation, and form grading fault-tolerant instructions; Step 4: Combine the initialization parameters and the current vehicle working condition to implement dynamic dead zone compensation and adaptive overmodulation algorithm, and output optimized control instructions containing dynamic compensation amount; Step 5: Use the optimized control instructions and dynamic compensation amount to complete space vector pulse width modulation processing to generate PWM drive signals for the drive power inverter bridge; Step 6: Respond to the grading fault-tolerant instructions, use sensorless estimation algorithm and remaining healthy sensor signals to perform data fusion, and realize fault-tolerant degraded operation of the motor.

[0022] In this embodiment, the motor parameter offline identification is performed by system power-on self-test to obtain the initialization parameters of motor stator resistance and cross-axis inductance, and the specific implementation includes: First, the system power-on self-test is performed to ensure that each hardware module works normally; then a specific test signal is injected to excite the motor; then the dynamic response data of the motor is collected; finally, the accurate motor parameters are calculated through the parameter identification algorithm. The whole process is completed in the controller initialization stage to provide accurate motor model parameters for the subsequent control algorithm.

[0023] After the system is powered on, the controller first performs a complete self-test process. This process includes reading and writing verification of the internal FLASH memory to ensure the integrity of the parameter storage area; calibration of the reference voltage and zero point of the ADC sampling circuit to ensure sampling accuracy; functional test of each phase bridge arm of the power drive circuit to verify the switching characteristics of the MOSFET by injecting a weak test current. During the self-test process, the controller monitors the response time, signal quality and other indicators of each functional module to establish a system health status report. Only when all self-test items pass, the system will enter the parameter identification stage. This design ensures that the identification process is carried out under the condition that the hardware system is completely normal, thereby ensuring the reliability of the parameter identification from the source.

[0024] After completing the system self-test, the controller injects a precisely designed non-rotating voltage test signal into the motor three-phase winding. The test signal is composed of multiple sinusoidal signal components with different frequencies, and its mathematical expression is: where, and represent the fundamental frequency and auxiliary excitation frequency, respectively, is the signal amplitude, controlled within 5%-10% of the motor rated voltage. The design of this multi-frequency composite signal can simultaneously excite different dynamic characteristics of the motor, and since the vector sum of the three-phase signals is zero, it will not produce net torque on the motor, achieving true "non-rotating" excitation and ensuring the safety of the identification process.

[0025] While injecting the test signal, the controller synchronously collects the voltage and current transient responses of the three-phase winding through the high-precision ADC module. The sampling frequency is set to 100kHz to ensure accurate capture of the motor's high-frequency dynamic characteristics. To improve the signal-to-noise ratio, multiple sampling and averaging are used at each sampling point. The raw data collected is first processed by a digital filter to eliminate high-frequency noise interference during sampling. The sampled voltage and current data are stored in the controller's buffer area and organized in the following format: Where N is the total number of sampling points, usually 1000-2000 points are taken to ensure the complete dynamic response process. After data acquisition is completed, the system will check the integrity of the data, eliminate abnormal data points, and ensure the accuracy of subsequent parameter calculation.

[0026] Based on the collected transient response data, recursive least squares method is used for parameter identification. First, the voltage equation of permanent magnet synchronous motor in the stationary coordinate system is established: .

[0027] After discretizing the continuous-time model, the standard form of the identifiable model is obtained: , where y(k) is the system output, is the observation matrix, is the parameter vector to be identified, is the observation noise.

[0028] The calculation process of recursive least squares method is as follows: , where K(k) is the gain matrix, P(k) is the covariance matrix, is the forgetting factor, usually 0.95-0.99. The algorithm starts from a set of initial estimates, and through continuous iteration and update, it finally converges to the true parameter value. The entire iteration process lasts about 100ms, ensuring the rapidity and accuracy of parameter identification.

[0029] The identified motor parameters are stored in a dedicated area of non-volatile memory. The data structure stored includes parameter values, identification time stamp, checksum, etc. To prevent data loss, the system uses a dual backup storage strategy, saving parameter copies in two different storage areas. At each power-on, the system reads these parameters and verifies their reasonableness, and if it finds abnormal parameters, it will automatically re-execute the identification process. This design ensures that the system can obtain reliable motor parameters in any situation, laying a solid foundation for high-performance control. At the same time, the system also retains multiple sets of historical parameter data for analyzing the variation of motor parameters with temperature, aging, etc., providing data support for predictive maintenance.

[0030] In this embodiment, the handle command, phase current, bus voltage and vehicle body inclination multi-source operating state signals are collected, and the current vehicle working condition judgment result is generated through vehicle state fusion analysis. The specific implementation includes: In the signal acquisition and preprocessing stage, the system acquires the original sensor data through multi-channel synchronous sampling technology. The handle command signal is collected through a high-precision ADC channel, with a sampling frequency of 1kHz. The original voltage signal collected is first subjected to moving average filtering, and its mathematical expression is: where the filter window size N is dynamically adjusted according to the system real-time requirement, achieving the best balance between response speed and noise suppression. The three-phase current signals are collected by an isolated current sensor and anti-aliasing filtered by a second-order Butterworth filter, whose transfer function in the Z domain is expressed as: .

[0031] The filter coefficients are optimized according to the motor rated frequency and switching frequency. The DC bus voltage is sampled by a differential amplification circuit, and in addition to the conventional filtering, an outlier detection mechanism based on statistics is added to effectively eliminate voltage spike interference caused by power device switching. The IMU module outputs three-axis acceleration and angular velocity data through a high-speed serial interface, and uses a gradient descent algorithm based on quaternions for attitude solving, with the quaternion update formula as: .

[0032] Through a series of preprocessing measures, the accuracy and reliability of the original sensor data are ensured.

[0033] After completing the signal preprocessing, the system performs coordinate system conversion and feature parameter extraction. The three-phase current signals are converted to the synchronous rotating coordinate system through Clarke-Park transformation, with the Clarke transformation formula as: .

[0034] The Park transformation formula is: , where is the real-time electrical angle. At the same time, the system calculates multiple time and frequency domain feature parameters, including the current ripple coefficient: .

[0035] power factor angle, instantaneous output power, and other key indicators. The IMU data is converted from the carrier coordinate system to the inertial coordinate system through coordinate rotation transformation, accurately separating the vehicle dynamic inclination, with the coordinate transformation formula as: ; The inclination calculation uses the acceleration component ratio method: .

[0036] These feature parameters describe the vehicle's running state from different dimensions, providing rich input information for subsequent intelligent recognition.

[0037] In the multi-source information fusion stage, the system uses an improved extended Kalman filter algorithm to deeply fuse various sensor data. A nonlinear state space model of the system is established, with the state vector containing key state variables such as vehicle speed, inclination, inclination rate of change, direct-axis current, and quadrature-axis current. The state equation is expressed as: , and the observation equation is expressed as: , where and are the process noise and observation noise, respectively. The prediction step of the extended Kalman filter algorithm includes state prediction: , and covariance prediction: , the update step includes Kalman gain calculation: , state update: , and covariance update: .

[0038] In view of the characteristics of IMU data being susceptible to vibration interference, an adaptive noise adjustment mechanism based on a vehicle kinematic model is designed, and by estimating the noise statistical characteristics in real time, the process noise covariance matrix and the observation noise covariance matrix are dynamically adjusted, which significantly improves the accuracy and robustness of state estimation.

[0039] In the intelligent state decision stage, the system uses a combination of fuzzy logic and probability reasoning to identify the working condition based on the fused feature parameters. First, the membership functions of the input variables are defined, including the fuzzy set of vehicle speed: , .

[0040] A rule base containing multiple inference rules is established, and the Mamdani fuzzy inference method is used, and the defuzzification uses the barycenter method: .

[0041] At the same time, the system establishes a state transition model based on a hidden Markov model, and defines the transition probabilities between different working condition states. State recognition is completed by solving the maximum a posteriori probability: , where s represents the working condition state and O represents the observation sequence. The posterior probability is calculated by the forward-backward algorithm. The system can accurately identify six typical working conditions: starting, constant speed cruising, accelerating and overtaking, driving on a slope, decelerating and sliding, and emergency braking, and assign appropriate confidence levels to each working condition. The final output of the working condition judgment result not only contains discrete working condition type identifiers, but also contains probability distribution vectors of each type, providing sufficient decision basis for the selection and parameter setting of subsequent adaptive control strategies. The entire processing process is completed within a fixed control period, and algorithm optimization and parallel computing ensure the real-time performance and reliability of the control.

[0042] In this embodiment, fault diagnosis and hierarchical evaluation are completed using real-time monitored multi-source operating state signals to form hierarchical fault-tolerant instructions, and the specific implementation manner includes: In the sensor signal validity verification link, the system performs real-time reliability evaluation on multi-source sensor data. For current sensor signals, a verification method based on signal feature consistency is used to detect sensor deviation by comparing the vectors of three-phase currents: .

[0043] When the deviation value exceeds the set threshold, the corresponding sensor is marked as suspicious state. For position sensors, a model-based prediction verification method is adopted to identify sensor failures by comparing the differences between observed values and measured values. The verification of voltage signals is carried out through the power balance principle, and a comparison model of input and output power is established. When the power deviation continuously exceeds the set range, the voltage sensor diagnostic program is triggered.

[0044] In the actuator state monitoring link, the system focuses on real-time monitoring of power inverters and motor bodies. The health status of power devices is evaluated through online parameter identification, and a conduction resistance model including temperature influence is established. The health status of motor windings is evaluated through insulation resistance monitoring and temperature monitoring, and a winding temperature rise calculation model is established. At the same time, by analyzing the current harmonic components in real time, the motor abnormalities are detected, and when the total harmonic distortion exceeds the safety threshold, it is suggested that there may be winding partial short circuit or permanent magnet demagnetization failure.

[0045] In the system-level fault diagnosis link, an intelligent diagnosis method based on multi-source information fusion is adopted. By designing a residual generator to detect system abnormalities: , the residual signal is analyzed statistically, and a fault detection index is constructed based on Mahalanobis distance. To further locate the fault source, a fault-symptom correlation model is established based on Bayesian network inference mechanism, and the most possible fault cause is determined by calculating the posterior probability: .

[0046] In the hierarchical evaluation decision link, the system establishes a four-level classification system according to the severity of the fault and the degree of influence on the system performance. The evaluation process adopts fuzzy comprehensive evaluation method, considering multiple factors such as fault recoverability, propagation and timeliness. For different fault levels, the system generates corresponding fault-tolerant instructions, including sensor signal reconstruction, control strategy switching and system performance degradation measures. All fault-tolerant instructions are encapsulated and transmitted through a special safety protocol to ensure the integrity and timeliness of the instructions. The system also establishes a fault-tolerant effect evaluation mechanism to verify the effectiveness of the fault-tolerant strategy by monitoring the changes of key performance indicators, forming a complete fault handling closed loop to ensure real-time monitoring and rapid response to system status.

[0047] In this embodiment, dynamic dead zone compensation and adaptive overmodulation algorithm are implemented in combination with initialization parameters and current vehicle operating conditions to output optimized control instructions containing dynamic compensation amount. The specific implementation includes: This section effectively solves the torque ripple problem caused by fixed dead time in traditional control, as well as the power output stability problem when the voltage is insufficient, through intelligent algorithm design. The specific implementation process includes four key stages: calculation of dynamic dead time compensation, decision-making of adaptive overmodulation strategy, coordinated execution of two algorithms, and generation of final optimized control commands.

[0048] During the dynamic dead-zone compensation calculation phase, the system dynamically determines the optimal compensation parameters based on real-time acquired motor speed, DC bus voltage, and load current using a pre-calibrated three-dimensional lookup table. The compensation voltage is calculated based on the following mathematical model: ,in, and For compensation coefficient, This is the real-time DC bus voltage. As the reference voltage, Electric angular velocity, For quadrature axis current, This is the rated current. The model fully considers the effects of voltage variations, speed, and load conditions on the dead-zone effect. During actual operation, the system monitors the switching characteristics of the power devices in real time and performs temperature compensation corrections to ensure optimal compensation performance at different operating temperatures. Specifically, under low-speed, high-load conditions, the system automatically enhances the compensation intensity using a nonlinear enhancement factor: ,in, To enhance the coefficient, Using the current threshold, this adaptive enhancement mechanism effectively suppresses torque ripple at low speeds.

[0049] During the adaptive overmodulation decision-making phase, the system continuously assesses the matching degree between the DC bus voltage state and output demand. The determination of overmodulation demand is based on the voltage utilization rate metric. ,when When this occurs, it indicates that the current voltage cannot meet the output demand, and the system begins a smooth transition to the overmodulation region. The transition process employs a gradual strategy, and the modulation depth adjustment follows these rules: in, For the rated modulation depth, For modulation depth increment, The time constant is used. Within the overmodulation region, the system employs an algorithm based on voltage vector trajectory optimization to maximize voltage output by adjusting the duration of the fundamental vector. Simultaneously, real-time harmonic monitoring ensures the total harmonic distortion (THD) remains within acceptable limits. Harmonic analysis utilizes the Fast Fourier Transform (FFT) method to perform spectral analysis on the output voltage waveform, ensuring the safe operation of the motor.

[0050] In the algorithm cooperative execution stage, the system establishes a cooperative working mechanism of dynamic dead-time compensation and adaptive overmodulation. The outputs of the two algorithms are fused to form a unified voltage command: wherein, is the basic voltage command, is the dead-time compensation amount, is the overmodulation additional voltage. The system ensures the reasonable cooperation of the two algorithms under different working conditions through a priority arbitration mechanism. In the normal voltage condition, dynamic dead-time compensation is mainly used to improve the quality of the output waveform; when the voltage is insufficient, the weight of adaptive overmodulation is automatically increased to prioritize power output. The mutual influence between the two algorithms is also considered in the cooperative control, and the interference between them is eliminated through feedforward compensation to ensure stable operation of the system.

[0051] In the optimization control command generation stage, the system converts the processed voltage command into the final optimization control signal. The control command contains amplitude, phase and frequency information, and is output in the form of space vector modulation. In order to ensure the smoothness of the control, slope limitation and filtering are added during the generation of the command: , wherein, is a saturation function, is the maximum allowed voltage rate of change. The system also outputs dynamic compensation amount information to provide data support for subsequent monitoring and protection. The entire processing process is completed in each control period to ensure the real-time and accuracy of the control. Through the fine processing of the four stages, the system can provide optimal control performance under various working conditions, significantly improving the efficiency and stability of the motor.

[0052] In this embodiment, the space vector pulse width modulation processing is completed by the optimized control command and the dynamic compensation amount to generate the PWM drive signal for driving the power inverter bridge, and the specific implementation includes: The specific implementation of space vector pulse width modulation processing is the key link of converting the optimized control command into the actual drive signal. This process realizes accurate control of the power inverter bridge through advanced modulation algorithms to ensure that the motor obtains ideal three-phase voltage supply. The specific implementation includes four core stages: reference voltage vector synthesis, space sector judgment, basic vector action time calculation, and PWM waveform generation, each of which uses optimized algorithms to achieve efficient and accurate control.

[0053] In the reference voltage vector synthesis stage, the system receives the optimized control command from the previous stage processing, and first converts it from the rotating coordinate system to the stationary coordinate system. This process is realized through inverse Park transformation, and its mathematical expression is: wherein, is the current electrical angle, and are the direct-axis and quadrature-axis voltage commands respectively. The converted voltage vector in the stationary coordinate system is the reference voltage vector that needs to be synthesized. The system incorporates dynamic compensation into the reference vector, and corrects the voltage deviation caused by the dead-time effect through vector superposition, ensuring the accuracy of the reference voltage. The calculation accuracy of this process directly affects the subsequent modulation effect, so 32-bit floating-point operations are used to ensure calculation accuracy.

[0054] In the space sector judgment stage, the system needs to determine the specific sector position of the reference voltage vector. The plane is divided into six sectors, each covering a range of 60 degrees. Sector judgment is achieved through the following logical relationship: The sector number N is calculated by the formula This judgment method only requires simple arithmetic operations and comparison operations, and has high calculation efficiency, making it suitable for fast execution in digital signal processors. The system will re-determine the sector position at each control period to ensure timely tracking of changes in the reference voltage vector.

[0055] In the basic vector action time calculation stage, the system calculates the action time of the adjacent two basic voltage vectors and the zero vector according to the sector where the reference voltage vector is located. Taking the case where the reference voltage vector is located in the first sector, the action time calculation is as follows: where is the PWM period, is the DC bus voltage, and correspond to the action time of the basic vectors and respectively, is the zero vector action time. The system dynamically adjusts the action time calculation by real-time monitoring of the DC bus voltage, ensuring accurate synthesis of the required output voltage vector under different input voltage conditions. When the calculation result appears saturated, the system will automatically perform time redistribution to ensure modulation linearity.

[0056] In the PWM waveform generation stage, the system generates the actual drive signal according to the calculated action time. The seven-segment SVPWM modulation strategy is adopted, with each PWM period containing seven segments, effectively reducing switching loss and current harmonics. Taking the first sector as an example, the switching time points of each phase bridge arm are calculated as follows: . The system generates six complementary PWM drive signals by generating corresponding comparison matching values through a timer. To ensure the safe operation of the power tube, a dead-time is inserted in the complementary signals: where is the power tube gate charge, is the drive voltage swing, is the gate drive current, is the circuit delay compensation time. The dead-time is dynamically adjusted according to the characteristics and operating temperature of the power device, while preventing shoot-through and minimizing the impact on the output waveform. The final generated PWM signal is amplified by the drive circuit and directly controls the switching state of each phase arm of the power inverter bridge, achieving precise driving of the motor.

[0057] The entire SVPWM processing is completed within a single control cycle, ensuring real-time performance through optimization algorithms and parallel computing. The system also includes a comprehensive protection mechanism that monitors abnormal conditions in real time during modulation to ensure the safe operation of power devices. Through this precise modulation technique, the system can effectively improve the utilization rate of the DC bus voltage, reduce output harmonic distortion, and improve the operating performance of the motor.

[0058] In this embodiment, in response to the hierarchical fault-tolerant instruction, a sensorless estimation algorithm and residual healthy sensor signals are used to perform data fusion to achieve fault-tolerant degraded operation of the motor. The specific implementation includes: This process intelligently fuses sensorless algorithms and healthy sensor signals to build a reliable motor state observation system, enabling seamless switching from normal operation to degraded mode. The specific implementation process mainly includes four key steps: sensorless algorithm activation, multi-source signal fusion processing, fault-tolerant control strategy execution, and system performance management.

[0059] In the sensorless algorithm activation phase, the system automatically starts the sensorless operation mode based on the received hierarchical fault-tolerant instruction. The system uses a sliding mode observer as the core estimation algorithm to estimate the rotor position and speed by establishing an observation model of the motor's back electromotive force. The observer design is based on the voltage equation of the motor in the stationary coordinate system, which continuously corrects the observer state by comparing the estimated current with the actual measured current in real time. The introduction of the sliding mode control term ensures the robustness of the system to parameter changes and external disturbances. The back electromotive force signal output by the observer is processed by a phase-locked loop to effectively extract the rotor position information, and the estimation process has good dynamic performance and stability.

[0060] In the multi-source signal fusion processing phase, the system uses an adaptive Kalman filter algorithm to optimally fuse the sensorless estimation results and the remaining healthy sensor signals. This algorithm establishes a state space model of the system to estimate the statistical characteristics of the process noise and observation noise in real time. The adjustment of the fusion weight is based on the reliability index of each signal source. When an abnormal sensor signal is detected, the system automatically reduces its weight to achieve a smooth transition to sensorless estimation. This adaptive fusion mechanism not only ensures the continuity of rotor position estimation, but also effectively avoids estimation jumps during mode switching, ensuring stable observation of the system state.

[0061] In the fault-tolerant control strategy execution phase, the system reconstructs the control law according to the fused state estimation results. The speed controller adopts an anti-windup PID structure, preventing integral windup phenomenon through conditional integration strategy. The current regulator is based on feedforward decoupling control, automatically adjusting the controller parameters to enhance system robustness in fault-tolerant operation mode. At the same time, the system also introduces model predictive control as a backup strategy, achieving better dynamic response by optimizing performance indicators. The output amplitude of the control system is dynamically adjusted according to the fault level to ensure safety in degraded operation.

[0062] In the system performance management phase, the system implements a hierarchical performance management strategy according to the severity of the fault. The system dynamically adjusts the system's operating boundaries by monitoring key performance indicators, including maximum output torque, maximum operating speed, and other parameters in real time. When a single sensor fault is detected, the system maintains a high level of performance; in the case of multiple sensor faults, the system appropriately reduces performance indicators to ensure safe operation. In addition, the system also establishes a comprehensive fault recovery mechanism that automatically executes a smooth switching process when a sensor is detected to be normal, gradually restoring the full functionality of the system. The entire fault-tolerant operation process is monitored through a closed loop to ensure the stability and reliability of the system, providing basic driving support for vehicles.

[0063] It should be noted that in this document, relational terms such as first and second and the like can be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0064] While embodiments of the application have been shown and described, it is to be understood that various modifications, substitutions, combinations, and variations can be made in the embodiments without departing from the spirit and scope of the present application, which is defined by the appended claims and their equivalents.

Claims

1. A control method for a unidirectional motor in a two-wheeled electric vehicle, characterized in that, Includes the following steps: Step 1: Perform offline identification of motor parameters through system power-on self-test to obtain the initialization parameters of motor stator resistance and AC / DC inductance; Step 2: Collect multi-source operating status signals such as throttle command, phase current, bus voltage and vehicle tilt angle, and generate the current vehicle operating condition judgment result through vehicle status fusion analysis; Step 3: Utilize real-time monitored multi-source operating status signals to complete fault diagnosis and hierarchical assessment, and generate hierarchical fault-tolerant instructions; Step 4: Combine the initialization parameters with the current vehicle operating conditions to implement dynamic dead zone compensation and adaptive overmodulation algorithms, and output optimized control commands containing dynamic compensation amounts; Step 5: Use optimized control commands and dynamic compensation to complete space vector pulse width modulation processing, and generate PWM drive signals to drive the power inverter bridge; Step Six: In response to the graded fault-tolerant command, the sensorless estimation algorithm and the remaining health sensor signals are used to perform data fusion to achieve fault-tolerant degraded operation of the motor.

2. The control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 1, characterized in that, The offline identification of motor parameters in step one includes: A complete system power-on self-test (POST) procedure is executed, including comprehensive functional verification of the controller's internal memory, sampling circuit, and power drive circuit. After the POST procedure is passed, a precisely controlled non-rotating test signal is injected into the three-phase windings of the motor. The frequency of the test signal is set within an appropriate range of the motor's rated frequency. The transient response of the winding voltage and current under the excitation of the test signal is acquired through a high-precision sampling module, and a high-frequency sampling method is used to ensure data accuracy. Based on the transient response data, the stator resistance and quadrature-axis inductance parameters of the motor are calculated using a recursive least squares parameter identification algorithm. The calculated initialization parameters are stored in a dedicated parameter storage area in the non-volatile memory.

3. The control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 1, characterized in that, The vehicle state fusion analysis in step two includes: The analog voltage signal output from the throttle is acquired through the sampling channel and subjected to multi-level digital filtering; the three-phase current sampling value of the motor and the DC bus voltage sampling value are monitored in real time, and a fixed sampling period is used to ensure data synchronization; the three-axis acceleration and angular velocity data output by the inertial measurement unit are acquired through the communication interface; based on the multi-source operating status signal, a Kalman filter data fusion algorithm is used to perform multi-sensor information fusion processing; combined with the preset state decision logic, which includes judgment conditions for various typical operating conditions, the accurate judgment result of the current vehicle operating condition is output.

4. The control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 1, characterized in that, The fault diagnosis and grading assessment in step three includes: The system monitors the integrity of the multi-source operating status signals in real time, including signal continuity detection and rationality verification; it determines whether the phase current in the multi-source operating status signals exceeds the safety limit and whether the DC bus voltage is within the normal operating range; it detects the temperature status of power devices through a temperature monitoring system, using a periodic sampling method; based on the monitoring and judgment results, it divides the system status into multiple levels according to the severity of the fault and generates corresponding graded fault-tolerant instructions.

5. The control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 1, characterized in that, The dynamic dead-time compensation in step four includes: Based on the real-time values ​​of motor speed and DC bus voltage under the current vehicle operating conditions, the optimal compensation amount is dynamically calculated using a pre-calibrated lookup table; based on the compensation amount, a variable compensation time is inserted into the space vector pulse width modulation signal, and the compensation time is set within a reasonable range according to the system characteristics; when a specific operating condition is detected, the intensity of the compensation amount is automatically increased according to the real-time load condition, using a configurable enhancement coefficient; the calculated dynamic compensation amount is output to the modulation process in real time.

6. The control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 5, characterized in that, The adaptive overmodulation algorithm in step four includes: The system continuously judges whether the DC bus voltage in the multi-source operating status signal meets the output voltage requirement at the current speed. When insufficient voltage is detected, the control algorithm automatically and smoothly enters the overmodulation mode, and the transition process remains stable. Based on the optimized control command, a specific modulation algorithm is used to adjust the voltage vector trajectory, and the degree of overmodulation is controlled within a reasonable range. Harmonic analysis is used to keep the total harmonic distortion rate of the modulated waveform from exceeding the allowable value.

7. The control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 1, characterized in that, The space vector pulse width modulation process in step five includes: The optimized control command is converted into a voltage vector reference value in a stationary coordinate system through coordinate transformation; the sector number of the voltage vector is determined by geometric analysis; the action time of the basic vector and the zero vector allocation time are calculated based on the modulation principle, and the calculation process is kept accurate; based on the action time, a symmetrical modulation strategy is used to generate multiple complementary drive signals, and the switching frequency is set to an appropriate value.

8. A control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 1, characterized in that, The sensorless estimation algorithm in step six includes: An observer based on a motor model is constructed, and the observer parameters are set according to system requirements. Based on the phase current and voltage sampling values ​​in the multi-source operating state signal, the rotor position is estimated in real time using a numerical calculation method, and the estimation process remains stable. An improved observer is used for rotor position tracking, and the observer gain is adaptively adjusted according to the operating state. The estimated fused rotor position information is output to the control system through a communication interface.

9. A control method for a unidirectional motor of a two-wheeled electric vehicle according to claim 8, characterized in that, The data fusion in step six includes: The sensorless algorithm output and the remaining health sensor signals are adaptively weighted and fused, with the weight coefficients dynamically adjusted according to signal quality. An advanced filtering algorithm is used to improve position estimation accuracy, and the filter parameters are optimized. The fusion weights are dynamically adjusted based on the hierarchical fault-tolerant instructions, and the fusion strategy is automatically adjusted when sensor performance deteriorates. When the sensors completely fail, the system smoothly switches to a pure sensorless operation mode while appropriately limiting the motor output capacity.

10. A control system for a unidirectional motor of a two-wheeled electric vehicle, used to implement the control method for a unidirectional motor of a two-wheeled electric vehicle as described in any one of claims 1-9, characterized in that, It includes a parameter self-learning and initialization module, a multi-source information fusion and state decision module, a fault diagnosis and fault-tolerant management module, a dynamic core control algorithm module, a space vector pulse width modulation module, and a fault-tolerant operation and signal fusion module; The parameter self-learning and initialization module performs offline identification of motor parameters through system power-on self-test to obtain the initialization parameters of motor stator resistance and quadrature-direct axis inductance; The multi-source information fusion and state decision module collects multi-source operating status signals such as throttle command, phase current, bus voltage and vehicle tilt angle, and generates the current vehicle operating condition judgment result through vehicle state fusion analysis. The fault diagnosis and fault tolerance management module uses real-time monitored multi-source operating status signals to complete fault diagnosis and graded assessment, and forms graded fault tolerance instructions. The dynamic core control algorithm module combines the initialization parameters with the current vehicle operating conditions to implement dynamic dead zone compensation and adaptive overmodulation algorithms, and outputs optimized control commands containing dynamic compensation amounts. The space vector pulse width modulation module uses optimized control commands and dynamic compensation to complete the space vector pulse width modulation process, generating a PWM drive signal to drive the power inverter bridge. The fault-tolerant operation and signal fusion module responds to the graded fault-tolerant command, uses a sensorless estimation algorithm and the remaining health sensor signals to perform data fusion, and realizes the fault-tolerant degraded operation of the motor.

Images