Electric automobile tail door anti-pinch control method based on current and speed of driving motor

Abstract

The invention discloses an electric automobile tail gate anti-pinch control method based on current and speed of a driving motor, and belongs to the technical field of electric automobile tail gate anti-pinch. The method comprises the following steps: acquiring an armature current signal of the driving motor by adopting a phase current sampling mode, and then acquiring a torque signal calculated by utilizing angular speed through a Hall sensor; according to the anti-pinch control method for the tail gate of the electric vehicle based on the current and the speed of the driving motor, a torque signal calculated by utilizing the angular speed is acquired by adopting a phase current sampling mode and a Hall sensor, so that the magnitude of a threshold value can be dynamically adjusted in real time according to different loads of the tail gate of the electric vehicle; the reliability is improved, whether the motor is stable or not can be judged through threshold value change, a current change threshold value I1 and an angular speed change threshold value N1 are set through an integral method, and the phenomenon that due to the fact that a large amount of noise and coupling information are generated in current sampling signals, signal fluctuation influences are likely to be generated, and anti-pinch misjudgment is caused is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of electric vehicle tailgate anti-pinch technology, and in particular to an electric vehicle tailgate anti-pinch control method based on driving motor current and speed. Background Art

[0002] Traditional car tailgates require manual lifting and closing, cannot hover mid-flight, and lack anti-pinch features. Electric tailgates can automatically open and close at the touch of a button, stopping at the appropriate height. Most importantly, electric tailgates have an anti-pinch feature that stops if the tailgate encounters an obstacle, preventing injury.

[0003] At present, in the existing technology, there are two types of anti-pinch technologies for automobile tailgates abroad. One is contact anti-pinch: it is mainly based on sensors installed at the edge or hinge of the tailgate. These sensors monitor whether the tailgate comes into contact with objects or people and produces slight squeezing during its movement. The anti-pinch sensor judges this squeezing signal and transmits the squeezing signal to the controller, which stops or reverses the movement of the tailgate. The second type is non-contact anti-pinch: it is also based on sensors installed around the tailgate, which can detect the space around the tailgate. When an obstacle is detected in the movement path of the tailgate, the system generates an interrupt, thereby triggering the anti-pinch function of the tailgate to avoid collision with objects or squeezing accidents. In contact anti-pinch technology, there are three anti-pinch judgment methods, one is mechanical anti-pinch, the second is the use of electric The electric actuator of the tailgate uses the motor in the tailgate actuator as the signal acquisition object, and identifies and judges through changes in the torque, current, speed, etc. of the driving motor. The third method is to use a pressure sensor. When the second method is used to acquire signal objects, the armature current value and motor torque of the DC motor are reflected in the operating state, and these thresholds are often selected as the basis for judgment. Since a large amount of noise and coupling information are generated in the current sampling signal, it is easy to produce signal fluctuations, resulting in anti-pinch misjudgment, and the stability of the anti-pinch force cannot be guaranteed. At the same time, due to the different vehicle usage environments, the load of the electric vehicle tailgate is uncertain, which leads to the variability of the set threshold. Therefore, it is necessary to invent an electric vehicle tailgate anti-pinch control method based on the driving motor current and speed to solve the above problems. Summary of the Invention

[0004] The purpose of the present invention is to address the defects and shortcomings of the above-mentioned prior art and provide an electric vehicle tailgate anti-pinch control method based on the driving motor current and speed. The method adopts the phase current sampling method and the Hall sensor to collect the torque signal calculated by the angular velocity, and at the same time constructs a learning-based electric vehicle network data model, selects the sliding square average of the torque, and adopts a steady-state Kalman filter algorithm, so that the threshold value can be adjusted in real time according to the different dynamics of the electric vehicle tailgate load, and the reliability can be increased while judging whether the motor is stable by the threshold change to increase the self-checking performance. By setting a current change threshold I1 and an angular velocity change threshold N1 by the integral method in the anti-pinch prejudgment process, and cooperating with the anti-pinch force detection, the motor current and angular velocity changes can be detected in real time during the lifting and lowering of the electric vehicle tailgate, and the difference calculation and judgment are performed based on the change value, and finally combined with the torque estimation value T M A double comparison is performed with the dynamic threshold Tth value of the torque in the steady state to avoid the occurrence of signal fluctuations caused by a large amount of noise and coupling information in the current sampling signal, which may lead to misjudgment of anti-pinch. At the same time, the stability of the anti-pinch force is guaranteed, which increases the user experience and improves the overall efficiency.

[0005] The technical solution adopted by the present invention to solve its technical problems is: a method for controlling the anti-pinch of the tailgate of an electric vehicle based on the current and speed of the driving motor. The method can improve and dynamically adjust the threshold value in real time according to the different loads of the tailgate of the electric vehicle. While increasing reliability, it can judge whether the motor is stable by changing the threshold value to increase self-testing performance, avoid the occurrence of a large amount of noise and coupling information in the current sampling signal, which is easy to cause signal fluctuations and cause anti-pinch misjudgment, while ensuring the stability of the anti-pinch force, increasing user experience and improving overall usage efficiency.

[0006] Method flow:

[0007] Step S1: Signal acquisition

[0008] The armature current signal of the drive motor is obtained by adopting the phase current sampling method, and the torque signal calculated by the angular velocity is collected by the Hall sensor. The motor current signal and Hall signal are updated according to a fixed time, and the data is collected in real time.

[0009] Step S2: Establishing a data calculation model

[0010] By building a learning-based electric vehicle network data model, the relationship between motor torque and armature current is analyzed and obtained, and the motor torque is selected as the state variable for anti-pinch detection.

[0011] Step S3: Data calculation

[0012] The torque state is estimated by using a steady-state Kalman filter algorithm to obtain a torque estimate value, and the sliding square average of the estimated torque is calculated;

[0013] Step S4: Anti-pinch determination

[0014] The present invention compares the sliding square average of the torque with the torque dynamic threshold to determine whether the opening and closing process of the electric vehicle tailgate encounters resistance. Before the anti-pinch judgment is effective, the anti-pinch pre-judgment is performed first, and the anti-pinch force is detected at the same time. If the anti-pinch judgment is effective, the abnormal signal is identified. If it is determined that the anti-pinch pre-judgment is caused by the abnormal signal, the calculated value in the anti-pinch force detection is cleared, and the data signal is obtained again in real time to update the data. If the anti-pinch judgment is effective, the motor is reversed and the protection is rolled back.

[0015] Preferably, in step S1, the electric vehicle tailgate drive motor adopts a permanent magnet DC brushed motor, and the electric vehicle tailgate drive motor adopts PWM pulse width modulation technology to realize DC motor speed regulation. By configuring the timer to the PWM output mode, PWM signals with different duty cycles are generated as inputs of the H-bridge power drive circuit, wherein the H-bridge power drive circuit is composed of a MOS tube driver chip IR2104 and an H-bridge composed of four MOS tubes. When the upper left and lower right MOS tubes in the H-bridge are turned on, the motor current flows from point A to point B, and the motor rotates forward; when the lower left and upper right MOS tubes in the H-bridge are turned on, the motor current flows from point B to point A, and the motor reverses; when the PWM signal duty cycle is set to 0, the MOS tube is cut off and the motor stops, thereby realizing motor control. The DC motor drive system has the following four basic electromagnetic equations:

[0016] Voltage balance equation: LdI d / dT=U d -I d RE a

[0017] Among them, U d The armature voltage of the motor can be controlled by adjusting the PWM duty cycle generated by the timer. d is the motor armature current, obtained by phase current sampling, E a is the back electromotive force of the motor, which hinders the change of the armature current, L is the motor inductance, and R is the motor resistance;

[0018] Back EMF equation: E a =K e *ω

[0019] Among them, K e is the motor back electromotive force constant, ω is the motor angular velocity,

[0020] Torque balance equation: Jdω / dT=TC -T L -T μ +μ V

[0021] Among them, J is the moment of inertia, Tc is the electromagnetic torque of the DC motor, T L is the total external load torque of the DC motor, T μ is the friction torque generated by the friction between the motor components, μ V is the vibration torque generated by the vibration of the motor body and its components. Due to the uncertainty of the vibration torque, it is assumed to be a white noise input with a variance mean of 0.

[0022] Electromagnetic torque equation: T c =K t *Id,T L =T load +T p ,T μ =B μ *ω

[0023] Among them, K t is the torque constant, B μ is the viscous friction coefficient, T load is the load torque generated by the tailgate of the electric vehicle, T p is the clamping torque generated when the tailgate encounters an obstacle when it is raised or lowered. Using the above equation, the relationship between the motor's total load torque and armature current is: T L =T load +T p =K t *Id-J*ω-B μ *ω+μ V .

[0024] Preferably, in step S2, the network data model is to first configure a Linux+GPU server environment, create a deep learning sandbox environment based on the Keras framework, adopt a ResNet network learning method of feature extraction and residual learning, and complete the configuration of the network environment and the construction of the model based on the GitHub code framework.

[0025] Preferably, in step S2, the specific process of using the steady-state Kalman filter algorithm to estimate the torque state is to establish the state space equation of the torque according to the electromagnetic equation of the DC motor:

[0026]

[0027]

[0028] X=[T d ωT L ]T , u=U d , Y=I d ,W=[0μ V μ T ] T

[0029] Among them, X is the state variable matrix of the system, Y is the output variable matrix of the system, u is the input variable of the system, F is the system parameter matrix, G is the input matrix, H is the output matrix, μ T is the white noise input with zero mean assumed for the uncertainty of the vibration torque, W is the process noise during the operation of the electric vehicle tailgate, and it has a normal distribution with covariance Q and mean 0, and V is the measurement noise during signal acquisition, and it has a normal distribution with covariance R and mean 0.

[0030] Preferably, in step S3, the steady-state Kalman torque prediction iterative equation is:

[0031]

[0032] In the formula, the angle is α, the angular velocity is ω,

[0033] Assuming that the state quantity is s(n)=[α(n)ω(n)], the state equation can be obtained from the kinematic equation

[0034]

[0035] The measured value can be regarded as the state value plus a measurement noise to obtain the measurement equation:

[0036] x(n)=Cs(n)+v(n)=[1 0][α(n)ω(n)]+v(n)

[0037] x(n) is the posterior estimation matrix of the nth state variable, that is, the best estimated value of the torque. The specific method for calculating the sliding square average of the estimated torque is: The sliding square average formula is as follows:

[0038] T_rms=√(1 / T_total_time)∫(T(t)dt)

[0039] It is equivalent to the rated torque, so it is an important parameter for us to choose the motor. We choose the rated torque of the motor according to the effective value of the torque.

[0040]

[0041] Where T-total time represents the total time of torque measurement, and T(t) represents the torque value at time t.

[0042] Preferably, in the step S4, during the anti-pinch prejudgment process, a current change threshold I1 and an angular velocity change threshold N1 are set. When the current increases and the change exceeds the threshold I1, and the angular velocity decreases and the change exceeds the threshold N1, the anti-pinch flag F=1 is set, and the current value before the signal change is selected as the current reference value I ref , the angular velocity value before the signal changes is the velocity reference value N ref , and switch to anti-pinch force detection at the same time. Otherwise, clear the calculated value in the anti-pinch force detection, and then re-acquire the data signal in real time to update the data. During the anti-pinch force detection process, the current reference value I ref Based on the current integration, I S , with the angular velocity reference value being N ref Basic difference calculation N S When the result difference exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2, the final anti-pinch judgment is made in combination with the torque threshold.

[0043] Preferably, in step S4, during the anti-pinch determination process, the dynamic threshold of the estimated torque is set to:

[0044] Tth=μT+3*σT

[0045] Wherein, μT is the estimated torque average value within the monitoring period T, σT is the estimated torque standard deviation within the monitoring period T, and the confidence interval of the estimated torque is determined to be 95% of the estimated torque value. The estimated torque data within the monitoring period T are compared with the confidence interval of the estimated torque. If all the data fall within the confidence interval, the motor reaches a stable state. Otherwise, it is not stable. When the motor is stable, the dynamic threshold value Tth of the torque in the steady state is obtained. The torque estimation value T when the difference between the above results exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2 is calculated. M After smoothing, it is compared with the dynamic threshold value Tth of the torque in the steady state. If the torque estimate T M If the value exceeds the dynamic threshold value Tth of the torque in steady state, the anti-pinch judgment will take effect. If the opening and closing process of the electric vehicle tailgate is blocked, the motor will reverse and fall back to protection. If the torque estimation value T M If the value is lower than the dynamic threshold value Tth of the torque in the steady state, the anti-pinch judgment is not effective, and the data signal is acquired again in real time to update the data.

[0046] Beneficial effects:

[0047] 1. The present invention is an electric vehicle tailgate anti-pinch control method based on the driving motor current and speed. The method adopts a phase current sampling method and a Hall sensor to collect the torque signal calculated by the angular velocity. At the same time, a learning-based electric vehicle network data model is constructed, the sliding square average of the torque is selected, and a steady-state Kalman filter algorithm is adopted. Therefore, the threshold value can be dynamically adjusted in real time according to the different loads of the electric vehicle tailgate. While increasing reliability, the threshold change can be used to judge whether the motor is stable and increase self-testing performance.

[0048] 2. The present invention is based on the anti-pinch control method of the electric vehicle tailgate based on the driving motor current and speed. In the process of anti-pinch prediction, the current change threshold I1 and the angular velocity change threshold N1 are set by the integration method. In combination with the anti-pinch force detection, the motor current and angular velocity changes can be detected in real time during the lifting and lowering of the electric vehicle tailgate. The difference calculation and judgment are performed based on the change values, and finally the torque estimation value T is combined with the anti-pinch force detection. M A double comparison is performed with the dynamic threshold Tth value of the torque in the steady state to avoid the occurrence of signal fluctuations caused by a large amount of noise and coupling information in the current sampling signal, which may lead to misjudgment of anti-pinch. At the same time, the stability of the anti-pinch force is guaranteed, which increases the user experience and improves the overall efficiency. DETAILED DESCRIPTION

[0049] The following is a clear and complete description of the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of the present invention.

[0050] Example 1

[0051] The present invention provides a technical solution: an anti-pinch control method for the tailgate of an electric vehicle based on the current and speed of a driving motor, comprising the following steps:

[0052] Step S1: Signal acquisition

[0053] The present invention obtains the armature current signal of the drive motor by adopting a phase current sampling method, and then uses a Hall sensor to collect the torque signal calculated using the angular velocity, and updates the motor current signal and the Hall signal according to a fixed time to collect data in real time. The electric vehicle tailgate drive motor adopts a permanent magnet DC brushed motor, and the electric vehicle tailgate drive motor adopts PWM pulse width modulation technology to realize DC motor speed regulation. By configuring the timer to the PWM output mode, PWM signals with different duty cycles are generated as inputs of the H-bridge power drive circuit, wherein the H-bridge power drive circuit is composed of a MOS tube driver chip IR2104 and an H-bridge composed of four MOS tubes. When the upper left and lower right MOS tubes in the H-bridge are turned on, the motor current flows from point A to point B, and the motor rotates forward; when the lower left and upper right MOS tubes in the H-bridge are turned on, the motor current flows from point B to point A, and the motor rotates reversely; when the PWM signal duty cycle is set to 0, the MOS tube is cut off and the motor stops, thereby realizing motor control. The DC motor drive system has the following four basic electromagnetic equations:

[0054] Voltage balance equation: LdI d / dT=U d -I d RE a

[0055] Among them, U d The armature voltage of the motor can be controlled by adjusting the PWM duty cycle generated by the timer. d is the motor armature current, obtained by phase current sampling, E a is the back electromotive force of the motor, which hinders the change of the armature current, L is the motor inductance, and R is the motor resistance;

[0056] Back EMF equation: E a =K e *ω

[0057] Among them, K e is the motor back electromotive force constant, ω is the motor angular velocity,

[0058] Torque balance equation: Jdω / dT=T C -T L -T μ +μ V

[0059] Among them, J is the moment of inertia, Tc is the electromagnetic torque of the DC motor, T L is the total external load torque of the DC motor, T μ is the friction torque generated by the friction between the motor components, μ Vis the vibration torque generated by the vibration of the motor body and its components. Due to the uncertainty of the vibration torque, it is assumed to be a white noise input with a variance mean of 0.

[0060] Electromagnetic torque equation: T c =K t *Id,T L =T load +T p ,T μ =B μ *ω

[0061] Among them, K t is the torque constant, B μ is the viscous friction coefficient, T load is the load torque generated by the tailgate of the electric vehicle, T p is the clamping torque generated when the tailgate encounters an obstacle when it is raised or lowered. Using the above equation, the relationship between the motor's total load torque and armature current is: T L =T load +T p =K t *Id-J*ω-B μ *ω+μ V ;

[0062] Step S2: Establishing a data calculation model

[0063] By building a learning-based electric vehicle network data model, the relationship between motor torque and armature current is analyzed and obtained. At the same time, motor torque is selected as the state variable for anti-pinch detection. The network data model is to first configure the Linux+GPU server environment, create a deep learning sandbox environment based on the Keras framework, and adopt the ResNet network learning method of feature extraction and residual learning. The network environment configuration and model construction are completed based on the GitHub code framework. The specific process of using the steady-state Kalman filter algorithm to estimate the torque state is as follows: According to the electromagnetic equation of the DC motor, the state space equation of the torque is established:

[0064]

[0065] X=[T d ωT L ] T , u=U d , Y=I d ,W=[0μ V μ T ] T

[0066] Among them, X is the state variable matrix of the system, Y is the output variable matrix of the system, u is the input variable of the system, F is the system parameter matrix, G is the input matrix, H is the output matrix, μ T is the white noise input with zero mean assumed for the uncertainty of the vibration torque, W is the process noise during the operation of the electric vehicle tailgate, and it has a normal distribution with covariance Q and mean 0, and v is the measurement noise during signal acquisition, and it has a normal distribution with covariance R and mean 0;

[0067] Step S3: Data calculation

[0068] The torque state is estimated by using the steady-state Kalman filter algorithm to obtain the torque estimate value, and the sliding square average of the estimated torque is calculated. The steady-state Kalman torque prediction iterative equation is:

[0069]

[0070] In the formula, the angle is α, the angular velocity is ω,

[0071] Assuming that the state quantity is s(n)=[α(n)ω(n)], the state equation can be obtained from the kinematic equation

[0072]

[0073] The measured value can be regarded as the state value plus a measurement noise to obtain the measurement equation:

[0074] x(n)=Cs(n)+v(n)=[1 0][α(n)ω(n)]+v(n)

[0075] x(n) is the posterior estimation matrix of the nth state variable, that is, the best estimated value of the torque. The specific method for calculating the sliding square average of the estimated torque is: The sliding square average formula is as follows:

[0076] T_rms=√(1 / T_total_time)∫(T(t)dt)

[0077] Where T-total time represents the total time of torque measurement, and T(t) represents the torque value at time t;

[0078] Step S4: Anti-pinch determination

[0079] By comparing the square average of the torque sliding with the torque dynamic threshold, it is determined whether the opening and closing process of the electric vehicle tailgate encounters resistance. Before the anti-pinch judgment is effective, the anti-pinch prediction is performed first, and the anti-pinch force is detected at the same time. If the anti-pinch judgment is effective, the abnormal signal is identified. If it is determined that the anti-pinch prediction is caused by the abnormal signal, the calculated value in the anti-pinch force detection is cleared, and the data signal is obtained again in real time for data update. If the anti-pinch judgment is effective, the motor is reversed and the protection is retracted. During the anti-pinch prediction process, a current change threshold I1 and an angular velocity change threshold N1 are set. When the current increases and the change exceeds the threshold I1, and the angular velocity decreases and the change exceeds the threshold N1, the anti-pinch flag F=1 is set, and the current value before the signal change is selected as the current reference value I ref , the angular velocity value before the signal changes is the velocity reference value N ref , and switch to anti-pinch force detection at the same time. Otherwise, clear the calculated value in the anti-pinch force detection, and then re-acquire the data signal in real time to update the data. During the anti-pinch force detection process, the current reference value I ref Based on the current integration, I S , with the angular velocity reference value being N ref Basic difference calculation N S When the result difference exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2, the final anti-pinch judgment is made in combination with the torque threshold. During the anti-pinch judgment process, the dynamic threshold of the estimated torque is set as:

[0080] Tth=μT+3*σT

[0081] Wherein, μT is the estimated torque average value within the monitoring period T, σT is the estimated torque standard deviation within the monitoring period T, and the confidence interval of the estimated torque is determined to be 95% of the estimated torque value. The estimated torque data within the monitoring period T are compared with the confidence interval of the estimated torque. If all the data fall within the confidence interval, the motor reaches a stable state. Otherwise, it is not stable. When the motor is stable, the dynamic threshold value Tth of the torque in the steady state is obtained. The torque estimation value T when the difference between the above results exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2 is calculated. M After smoothing, it is compared with the dynamic threshold value Tth of the torque in the steady state. If the torque estimate T M If the value exceeds the dynamic threshold value Tth of the torque in steady state, the anti-pinch judgment will take effect. If the opening and closing process of the electric vehicle tailgate is blocked, the motor will reverse and fall back to protection. If the torque estimation value T M If the value is lower than the dynamic threshold value Tth of the torque in the steady state, the anti-pinch judgment is not effective, and the data signal is acquired again in real time to update the data.

[0082] The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed of the present invention is to collect the torque signal calculated by angular velocity by adopting the phase current sampling method and the Hall sensor, and at the same time construct a learning-based electric vehicle network data model, select the sliding square average of the torque, and adopt a steady-state Kalman filter algorithm, so that the threshold value can be adjusted dynamically in real time according to the different loads of the electric vehicle tailgate, and the reliability can be increased while judging whether the motor is stable and increasing the self-checking performance by using the threshold change. In the anti-pinch prejudgment process, a current change threshold I1 and an angular velocity change threshold N1 are set by the integral method, and in combination with the anti-pinch force detection, the motor current and angular velocity changes can be detected in real time during the lifting and lowering of the electric vehicle tailgate, and the difference calculation and judgment are performed based on the change value, and finally combined with the torque estimation value T M A double comparison is performed with the dynamic threshold Tth value of the torque in the steady state to avoid the occurrence of signal fluctuations caused by a large amount of noise and coupling information in the current sampling signal, which may lead to misjudgment of anti-pinch. At the same time, the stability of the anti-pinch force is guaranteed, which increases the user experience and improves the overall efficiency.

[0083] Example 2

[0084] A method for controlling the tailgate anti-pinch function of an electric vehicle based on the current and speed of the drive motor can improve and dynamically adjust the threshold value in real time according to the load of the electric vehicle's tailgate. This method increases reliability and can also be used to determine whether the motor is stable by changing the threshold value, thereby increasing self-testing capabilities. This method avoids the effects of signal fluctuations caused by a large amount of noise and coupling information in the current sampling signal, which can lead to misjudgment of anti-pinch function. It also ensures the stability of the anti-pinch force, improves user experience, and enhances overall efficiency. Specifically, the method includes the following steps:

[0085] Step S1: Signal acquisition

[0086] The armature current signal of the drive motor is obtained by adopting the phase current sampling method, and the torque signal calculated by the angular velocity is collected by the Hall sensor. The motor current signal and Hall signal are updated according to a fixed time, and the data is collected in real time.

[0087] Step S2: Establishing a data calculation model

[0088] By building a learning-based electric vehicle network data model, the relationship between motor torque and armature current is analyzed and obtained, and the motor torque is selected as the state variable for anti-pinch detection.

[0089] Step S3: Data calculation

[0090] The torque state is estimated by using a steady-state Kalman filter algorithm to obtain a torque estimate value, and the sliding square average of the estimated torque is calculated;

[0091] Step S4: Anti-pinch determination

[0092] The present invention compares the sliding square average of the torque with the torque dynamic threshold to determine whether the opening and closing process of the electric vehicle tailgate encounters resistance. Before the anti-pinch judgment is effective, the anti-pinch pre-judgment is performed first, and the anti-pinch force is detected at the same time. If the anti-pinch judgment is effective, the abnormal signal is identified. If it is determined that the anti-pinch pre-judgment is caused by the abnormal signal, the calculated value in the anti-pinch force detection is cleared, and the data signal is obtained again in real time to update the data. If the anti-pinch judgment is effective, the motor is reversed and the protection is rolled back.

[0093] Preferably, in step S1, the electric vehicle tailgate drive motor adopts a permanent magnet DC brushed motor, and the electric vehicle tailgate drive motor adopts PWM pulse width modulation technology to realize DC motor speed regulation. By configuring the timer to the PWM output mode, PWM signals with different duty cycles are generated as inputs of the H-bridge power drive circuit, wherein the H-bridge power drive circuit is composed of a MOS tube driver chip IR2104 and an H-bridge composed of four MOS tubes. When the upper left and lower right MOS tubes in the H-bridge are turned on, the motor current flows from point A to point B, and the motor rotates forward; when the lower left and upper right MOS tubes in the H-bridge are turned on, the motor current flows from point B to point A, and the motor reverses; when the PWM signal duty cycle is set to 0, the MOS tube is cut off and the motor stops, thereby realizing motor control. The DC motor drive system has the following four basic electromagnetic equations:

[0094] Voltage balance equation: LdI d / dT=U d -I d RE a

[0095] Among them, U d The armature voltage of the motor can be controlled by adjusting the PWM duty cycle generated by the timer. d is the motor armature current, obtained by phase current sampling, E a is the back electromotive force of the motor, which hinders the change of the armature current, L is the motor inductance, and R is the motor resistance;

[0096] Back EMF equation: E a =K e *ω

[0097] Among them, K e is the motor back electromotive force constant, ω is the motor angular velocity,

[0098] Torque balance equation: Jdω / dT=T C -T L -T μ +μ V

[0099] Among them, J is the moment of inertia, Tc is the electromagnetic torque of the DC motor, T L is the total external load torque of the DC motor, T μ is the friction torque generated by the friction between the motor components, μ V is the vibration torque generated by the vibration of the motor body and its components. Due to the uncertainty of the vibration torque, it is assumed to be a white noise input with a variance mean of 0.

[0100] Electromagnetic torque equation: T c =K t *Id,T L =T load +T p ,T μ =B μ *ω

[0101] Among them, K t is the torque constant, B μ is the viscous friction coefficient, T load is the load torque generated by the tailgate of the electric vehicle, T p is the clamping torque generated when the tailgate encounters an obstacle when it is raised or lowered. Using the above equation, the relationship between the motor's total load torque and armature current is: T L =T load +T p =K t *Id-J*ω-B μ *ω+μ V .

[0102] Preferably, in step S2, the network data model is to first configure a Linux+GPU server environment, create a deep learning sandbox environment based on the Keras framework, adopt a ResNet network learning method of feature extraction and residual learning, and complete the configuration of the network environment and the construction of the model based on the GitHub code framework.

[0103] Preferably, in step S2, the specific process of using the steady-state Kalman filter algorithm to estimate the torque state is to establish the state space equation of the torque according to the electromagnetic equation of the DC motor:

[0104]

[0105] X=[T d ωT L ] T , u=U d , Y=I d ,W=[0μ V μ T ] T

[0106] Among them, X is the state variable matrix of the system, Y is the output variable matrix of the system, u is the input variable of the system, F is the system parameter matrix, G is the input matrix, H is the output matrix, μ T is the white noise input with zero mean assumed for the uncertainty of the vibration torque, W is the process noise during the operation of the electric vehicle tailgate, and it has a normal distribution with covariance Q and mean 0, and V is the measurement noise during signal acquisition, and it has a normal distribution with covariance R and mean 0.

[0107] Preferably, in step S3, the steady-state Kalman torque prediction iterative equation is:

[0108]

[0109] In the formula, the angle is α, the angular velocity is ω,

[0110] Assuming that the state quantity is s(n)=[α(n)ω(n)], the state equation can be obtained from the kinematic equation

[0111]

[0112] The measured value can be regarded as the state value plus a measurement noise to obtain the measurement equation:

[0113] x(n)=Cs(n)+v(n)=[1 0][α(n)ω(n)]+v(n)

[0114] x(n) is the posterior estimation matrix of the nth state variable, that is, the best estimated value of the torque. The specific method for calculating the sliding square average of the estimated torque is: The sliding square average formula is as follows:

[0115] T_rms=√(1 / T_total_time)∫(T(t)dt)

[0116] It is equivalent to the rated torque, so it is an important parameter for us to choose the motor. We choose the rated torque of the motor according to the effective value of the torque.

[0117]

[0118] Where T-total time represents the total time of torque measurement, and T(t) represents the torque value at time t.

[0119] Preferably, in the step S4, during the anti-pinch prejudgment process, a current change threshold I1 and an angular velocity change threshold N1 are set. When the current increases and the change exceeds the threshold I1, and the angular velocity decreases and the change exceeds the threshold N1, the anti-pinch flag F=1 is set, and the current value before the signal change is selected as the current reference value Iref , the angular velocity value before the signal changes is the velocity reference value N ref , and switch to anti-pinch force detection at the same time. Otherwise, clear the calculated value in the anti-pinch force detection, and then re-acquire the data signal in real time to update the data. During the anti-pinch force detection process, the current reference value I ref Based on the current integration, I S , with the angular velocity reference value being N ref Basic difference calculation N S When the result difference exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2, the final anti-pinch judgment is made in combination with the torque threshold.

[0120] Preferably, in step S4, during the anti-pinch determination process, the dynamic threshold of the estimated torque is set to:

[0121] Tth=μT+3*σT

[0122] Wherein, μT is the estimated torque average value within the monitoring period T, σT is the estimated torque standard deviation within the monitoring period T, and the confidence interval of the estimated torque is determined to be 95% of the estimated torque value. The estimated torque data within the monitoring period T are compared with the confidence interval of the estimated torque. If all the data fall within the confidence interval, the motor reaches a stable state. Otherwise, it is not stable. When the motor is stable, the dynamic threshold value Tth of the torque in the steady state is obtained. The torque estimation value T when the difference between the above results exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2 is calculated. M After smoothing, it is compared with the dynamic threshold value Tth of the torque in the steady state. If the torque estimate T M If the value exceeds the dynamic threshold value Tth of the torque in steady state, the anti-pinch judgment will take effect. If the opening and closing process of the electric vehicle tailgate is blocked, the motor will reverse and fall back to protection. If the torque estimation value T M If the value is lower than the dynamic threshold value Tth of the torque in the steady state, the anti-pinch judgment is not effective, and the data signal is acquired again in real time to update the data.

[0123] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An anti-pinch control method for the tailgate of an electric vehicle based on the driving motor current and speed, characterized in that: The following steps are involved: Step S1: signal acquisition; The armature current signal of the drive motor is obtained by adopting the phase current sampling method, and the torque signal calculated by the angular velocity is collected by the Hall sensor. The motor current signal and Hall signal are updated according to a fixed time, and the data is collected in real time. Step S2: establishing a data calculation model; By building a learning-based electric vehicle network data model, the relationship between motor torque and armature current is analyzed and obtained, and the motor torque is selected as the state variable for anti-pinch detection. Step S3: data calculation; The torque state is estimated by using a steady-state Kalman filter algorithm to obtain a torque estimate value, and the sliding square average of the estimated torque is calculated; Step S4: Anti-pinch determination By comparing the sliding square average of the torque with the torque dynamic threshold, it is determined whether the opening and closing process of the electric vehicle tailgate encounters resistance. Before the anti-pinch judgment takes effect, the anti-pinch prediction is performed first, and the anti-pinch force is detected at the same time. If the anti-pinch judgment is effective, the abnormal signal is identified. If it is determined that the anti-pinch prediction is caused by the abnormal signal, the calculated value in the anti-pinch force detection is cleared, and the data signal is obtained again in real time for data update. If the anti-pinch judgment is effective, the motor is reversed and the protection is retracted.

2. The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed according to claim 1 is characterized in that: In step S1, the electric vehicle tailgate drive motor adopts a permanent magnet DC brushed motor, and the electric vehicle tailgate drive motor adopts PWM pulse width modulation technology to realize DC motor speed regulation. By configuring the timer to the PWM output mode, PWM signals with different duty cycles are generated as inputs of the H-bridge power drive circuit, wherein the H-bridge power drive circuit is composed of a MOS tube driver chip IR2104 and an H-bridge composed of four MOS tubes. When the upper left and lower right MOS tubes in the H-bridge are turned on, the motor current flows from point A to point B, and the motor rotates forward; when the lower left and upper right MOS tubes in the H-bridge are turned on, the motor current flows from point B to point A, and the motor rotates reversely; when the PWM signal duty cycle is set to 0, the MOS tube is cut off and the motor stops, thereby realizing motor control. The DC motor drive system has the following four basic electromagnetic equations: Voltage balance equation: LdI d / dT=U d -I d RE a Among them, U d The armature voltage of the motor can be controlled by adjusting the PWM duty cycle generated by the timer. d is the motor armature current, obtained by phase current sampling, E a is the back electromotive force of the motor, which hinders the change of the armature current, L is the motor inductance, and R is the motor resistance; Back EMF equation: E a =K e *ω Among them, K e is the motor back electromotive force constant, ω is the motor angular velocity, Torque balance equation: Jdω / dT=T C -T L -T μ +μ V Among them, J is the moment of inertia, Tc is the electromagnetic torque of the DC motor, T L is the total external load torque of the DC motor, T μ is the friction torque generated by the friction between the motor components, μ V is the vibration torque generated by the vibration of the motor body and its components. Due to the uncertainty of the vibration torque, it is assumed to be a white noise input with a variance mean of 0. Electromagnetic torque equation: T c =K t *Id,T L =T load +T p ,T μ =B μ *ω Among them, K t is the torque constant, B μ is the viscous friction coefficient, T load is the load torque generated by the tailgate of the electric vehicle, T p is the clamping torque generated when the tailgate encounters an obstacle when it is raised or lowered. Using the above equation, the relationship between the motor's total load torque and armature current is: T L =T load +T p =K t *Id-J*ω-B μ *ω+μ V .

3. The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed according to claim 1 is characterized in that: In step S2, the network data model is to first configure the Linux+GPU server environment, create a deep learning sandbox environment based on the Keras framework, adopt the ResNet network learning method of feature extraction and residual learning, and complete the configuration of the network environment and the construction of the model based on the GitHub code framework.

4. The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed according to claim 1 is characterized in that: In step S2, the specific process of using the steady-state Kalman filter algorithm to estimate the torque state is to establish the state space equation of the torque based on the electromagnetic equation of the DC motor: X=[T d ωT L ] T ,u=U d ,Y=I d ,W=[0μ V μ T ] T Among them, X is the state variable matrix of the system, Y is the output variable matrix of the system, u is the input variable of the system, F is the system parameter matrix, G is the input matrix, H is the output matrix, μ T is the white noise input with zero mean assumed for the uncertainty of the vibration torque, W is the process noise during the operation of the electric vehicle tailgate, and its covariance is Q and its mean is 0. v is the measurement noise during signal acquisition, and its covariance is R and its mean is 0.

5. The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed according to claim 1 is characterized in that: In step S3, the steady-state Kalman torque prediction iterative equation is: In the formula, the angle is α, the angular velocity is ω, Assuming that the state quantity is s(n) = [α(n)ω(n)], the state equation can be obtained from the kinematic equation: The measured value can be regarded as the state value plus a measurement noise to obtain the measurement equation: x(n)=Cs(n)+v(n)=[10][α(n)ω(n)]+v(n) x(n) is the posterior estimation matrix of the nth state variable, which is the best estimate of the torque. The specific method for calculating the sliding square average of the estimated torque is: The sliding square average formula is as follows: T_rms=√(1 / T_total_time)∫(T(t)dt) Where T-total time represents the total time of torque measurement, and T(t) represents the torque value at time t.

6. The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed according to claim 1, characterized in that: In the step S4, during the anti-pinch prediction process, a current change threshold I1 and an angular velocity change threshold N1 are set. When the current increases and the change exceeds the threshold I1, and the angular velocity decreases and the change exceeds the threshold N1, the anti-pinch flag F=1 is set, and the current value before the signal change is selected as the current reference value I ref , the angular velocity value before the signal changes is the velocity reference value N ref , and switch to anti-pinch force detection at the same time. Otherwise, clear the calculated value in the anti-pinch force detection, and then re-acquire the data signal in real time to update the data. During the anti-pinch force detection process, the current reference value I ref Based on the current integration, I S , with the angular velocity reference value being N ref Basic difference calculation N S When the result difference exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2, the final anti-pinch judgment is made in combination with the torque threshold.

7. The electric vehicle tailgate anti-pinch control method based on the driving motor current and speed according to claim 1, characterized in that: In step S4, during the anti-pinch determination process, the dynamic threshold of the estimated torque is set to: Tth=μT+3*σT Wherein, μT is the estimated torque average value within the monitoring period T, σT is the estimated torque standard deviation within the monitoring period T, and the confidence interval of the estimated torque is determined to be 95% of the estimated torque value. The estimated torque data within the monitoring period T are compared with the confidence interval of the estimated torque. If all the data fall within the confidence interval, the motor reaches a stable state. Otherwise, it is not stable. When the motor is stable, the dynamic threshold value Tth of the torque in the steady state is obtained. The torque estimation value T when the difference between the above results exceeds the current integral difference threshold I2 or the angular velocity difference threshold N2 is calculated. M After smoothing, it is compared with the dynamic threshold value Tth of the torque in the steady state. If the torque estimate T M If the value exceeds the dynamic threshold value Tth of the torque in steady state, the anti-pinch judgment will take effect. If the opening and closing process of the electric vehicle tailgate is blocked, the motor will reverse and fall back to protection. If the torque estimation value T M If the value is lower than the dynamic threshold value Tth of the torque in the steady state, the anti-pinch judgment is not effective, and the data signal is acquired again in real time to update the data.

Images