Real-time torque control method based on model predictive control

Abstract

The invention discloses a real-time torque control method based on model predictive control, which reduces gear shifting impact and effectively improves comfort in a driving process. The method is characterized in that a dynamical model is established to predict the working condition of the next moment, the dynamical model comprises a working condition recognition module, a vehicle speed and state module matched with the working condition and a model prediction control module, and the model prediction and control module obtains an optimal solution after obtaining an objective function and constraints through a prediction model. The output control quantity corresponding to the optimal solution is used for eliminating the expected value deviation and predicting the output execution quantity at the next moment in real time to be executed by an execution mechanism of the driving transmission system, and the working condition identification module is used for collecting the corresponding parameters of the control quantity of each working condition in the current and historical input information; the corresponding parameters of the control quantity comprise a target vehicle speed vref, a target motor torque tau ref, a target gearbox gear gearref and a target torque J; the future working condition is predicted according to current and historical input information collected in real time.

Description

Technical Field

[0001] The invention relates to the technical field of automobile torque control, and in particular to a real-time torque control method based on model predictive control. Background Art

[0002] With the development of clean energy and driven by national policies, the new energy vehicle industry has developed rapidly. New energy passenger vehicles have high technical maturity, and there are many mature cases in electric drive systems, etc. The infrastructure is complete, the market share is large, the acceptance is high, and the development is relatively rapid. Passenger vehicles usually do not need to work for a long time under high load, and the driving conditions are relatively simple. In terms of shifting strategies, there are many intelligent control methods to achieve the adaptability of different driving scenarios and driving modes. In the continuous development, there are gradually cases of adopting more intelligent and advanced systems such as deep learning or reinforcement learning.

[0003] Commercial vehicles, as the basic carrier for the rapid development of the national economy, play an important role. In the field of commercial vehicles, the technical maturity of shifting strategies is relatively low. Compared with passenger cars, commercial vehicles usually have more complex and changeable driving conditions and more complex driving routes. In addition, heavy trucks and logistics vehicles have more complex power and load requirements, especially in heavy load and long-distance transportation, as well as frequent starting, acceleration, deceleration and parking. An adaptive drive system is required to provide corresponding torque output that meets the requirements under the relevant requirements of energy saving and smoothness. Therefore, the shifting strategy needs to be adjusted in combination with multiple factors such as load conditions to ensure efficient power distribution and adapt to changing conditions for adjustment and prediction.

[0004] The existing pure electric commercial vehicle shifting strategy is based on static rules and cannot cope with complex and changing driving conditions (such as road slope changes, load fluctuations, traffic conditions, etc.). The disadvantage of this static strategy is that the feedback is delayed, which makes the vehicle unable to match the current driving needs in real time, affecting driving performance and comfort.

[0005] If pure electric commercial vehicles are not properly controlled when shifting gears, it is easy to cause torque mutation, gear shift shock and driving discomfort. Especially in electric commercial vehicles with multi-speed gearboxes, the smoothness of torque changes and precise control of gear shift timing are crucial to comfort and power performance.

[0006] In actual operation, pure electric commercial vehicles need to balance power performance and energy efficiency, especially in long-distance transportation and under heavy load conditions. Existing shift strategies usually aim to maximize power output while ignoring energy efficiency optimization, resulting in unnecessary energy loss. Summary of the invention

[0007] In response to the above problems, the present invention provides a real-time torque control method based on model predictive control, which controls the output and thus the power state of the vehicle by balancing the objectives and constraints. It can control the rate of change of torque and the amount of change of speed, avoid smooth gear shifting and sudden changes in related control quantities, reduce gear shifting shock, and effectively improve comfort during driving.

[0008] A real-time torque control method based on model predictive control, characterized in that:

[0009] It establishes a dynamic model to predict the working condition at the next moment. The dynamic model includes a working condition identification module, a vehicle speed and state module matching the working condition, and a model prediction control module. The model prediction and control module obtains the optimal solution after obtaining the objective function and constraints through the prediction model. The output control amount corresponding to the optimal solution eliminates the expected value deviation and predicts the output execution amount of the next moment in real time to the actuator of the drive transmission system for execution. The working condition identification module collects the corresponding parameters of the control amount of each working condition in the current and historical input information. The corresponding parameters of the control amount include the target vehicle speed vref, the target motor torque τref, the target gearbox gear gearref, and the target torque J;

[0010] It predicts future operating conditions based on current and historical input information collected in real time. During the control process, the reference control quantity is designed as the target vehicle speed vref, target motor torque τref, target gearbox gear gearref, and the real-time torque target value that matches the current operating condition for subsequent operating condition identification. At the same time, the real-time torque target value that matches the current operating condition is the reference value for subsequent model predictive control, so that the actual torque target value of the next subsequent operating condition is as close as possible to the actual torque target value in the calculation process of the control output.

[0011] It is further characterized by:

[0012] First, a vehicle dynamics model is established, that is, a discretized prediction model containing information at each moment, which includes a vehicle speed model and a gearbox gear model. To reduce the complexity of computational control, a simplified linear model is used:

[0013] Vehicle speed model based on motor torque:

[0014] v(t+1)=v(t)+Δt·a(t)

[0015]

[0016] Where v(t) is the vehicle speed at time t, Δt is the time step, and τmotor(t) is the motor torque. In this process, you can customize the prediction duration.

[0017] Transmission gear model:

[0018] u gear (t) = f (v (t), a (t), SOC (t))

[0019] Discrete gear information is selected based on the current vehicle speed and motor torque and other related information. The current vehicle speed, accelerator pedal position information, current motor speed and output torque information, and current gear position information of the transmission are collected in real time, and combined with the corresponding working condition identification module, the current working condition information is obtained, and the vehicle speed and acceleration requirements at the next moment are predicted as input information;

[0020] The control problem is then transformed into an optimization problem under given constraints, and the system performance is optimized using the minimum change in control input. The objective function is defined as:

[0021]

[0022] v(t+k) is the speed of the vehicle at the kth point in the prediction time domain, v ref (t+k) is the desired reference speed, utorque(t+k) is the control output of the motor torque, ugear(t+k) is the prediction of the gearbox gear output, δ(t+k) is the predicted information of the accelerator pedal position, δref(t+k) is the desired pedal position, and the control is close to the reference position;

[0023] Q1 is the weighted coefficient of speed error, which controls the accuracy of speed tracking;

[0024] Q2 is the weighted coefficient for controlling the motor torque output;

[0025] Q3 is the weighting coefficient of the speed line shift;

[0026] Q4 is the smooth response of the accelerator pedal;

[0027] Adding constraints related to smoothness to the objective function, the constraints limit the changes in motor torque and gearbox gear position, avoiding large mutations in the control quantity of the calculated output, ensuring smoothness, and further matching the current working conditions;

[0028]

[0029] Add vehicle speed constraints:

[0030] v min ≤v(t+k)≤v max

[0031] MPC adjusts the control amount in real time through rolling optimization, ensuring that the control strategy can be adjusted according to the current working conditions at every moment to optimize the driving experience, performance and energy efficiency.

[0032] It is further characterized by:

[0033] Q1 is the weighted coefficient of speed error, which controls the accuracy of speed tracking. In order to improve the stability of gear shift control, some speed fluctuations are allowed appropriately. The empirical value is set between 10^2 and 10^3.

[0034] Q2 is the weighted coefficient for controlling the motor torque output. A larger coefficient is used to control the algorithm to output the motor torque smoothly, reducing the unevenness and frustration during acceleration. It is set to 8 in the empirical value of 1-10.

[0035] Q3 is the weighting coefficient of the gear shift line. A larger value is used to avoid a large gear shift amplitude and a sense of frustration. It is set to 800 in the empirical value of 100-1000;

[0036] Q4 is the smooth response of the accelerator pedal and is a moderate value of 5.

[0037] The output control quantity calculated at the next moment, i.e., the motor torque, the gearbox gear information, and the motor torque, is output to the motor execution drive module. If the gearbox calculates and evaluates the need to shift up or down, it is transmitted to the drive system TCU, and the control information is transmitted to the shift motor to control the corresponding shift fork to perform the shift action.

[0038] After adopting the present invention, model predictive control can be used to dynamically adapt to the working conditions. Through some sensors to identify or monitor relevant vehicle real-time data, the torque and gear can be adaptively adjusted within the range allowed by the computing power of the vehicle-mounted platform. Since the model predictive control itself can set multiple control quantities, and the output of the next control quantity is performed within the settable prediction range, the control quantity can be slowly corrected and output, and the next long output can be predicted and controlled in advance. The motor output can be adjusted in advance and the gear shifting timing can be controlled to match the driving state. The gear shifting strategy of the motor and the gearbox can be coordinated through a more sophisticated control method, and the output torque is smoother. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a flow chart of the control process of the present invention;

[0040] Figure 2 The vehicle speed state diagram obtained by using the method of the present invention and the existing control method is shown. DETAILED DESCRIPTION

[0041] A real-time torque control method based on model predictive control, see Figure 1, which establishes a dynamic model to predict the working condition prediction at the next moment. The dynamic model includes a working condition identification module, a vehicle speed and state module matching the working condition, and a model prediction control module. The model prediction and control module obtains the optimal solution after obtaining the objective function and constraints through the prediction model. The output control amount corresponding to the optimal solution eliminates the expected value deviation and predicts the output execution amount of the next moment in real time to the actuator of the drive transmission system for execution. The working condition identification module collects the corresponding parameters of the control amount of each working condition in the current and historical input information. The corresponding parameters of the control amount include the target vehicle speed vref, the target motor torque τref, the target gearbox gear gearref, and the target torque J;

[0042] It predicts future operating conditions based on current and historical input information collected in real time. During the control process, the reference control quantity is designed as the target vehicle speed vref, target motor torque τref, target gearbox gear gearref, and the real-time torque target value that matches the current operating condition for subsequent operating condition identification. At the same time, the real-time torque target value that matches the current operating condition is the reference value for subsequent model predictive control, so that the actual torque target value of the next subsequent operating condition is as close as possible to the actual torque target value in the calculation process of the control output.

[0043] In the specific implementation, we first establish the vehicle dynamics model, that is, the prediction model containing the information at each moment after discretization, including the vehicle speed model and the gearbox gear model. In order to reduce the complexity of calculation and control, a simplified linear model is used:

[0044] Vehicle speed model based on motor torque:

[0045] v(t+1)=v(t)+Δt·a(t)

[0046]

[0047] Where v(t) is the vehicle speed at time t, Δt is the time step, and τmotor(t) is the motor torque. In this process, you can customize the prediction duration.

[0048] Transmission gear model:

[0049] u gear (t) = f (v (t), a (t), SOC (t))

[0050] Discrete gear information is selected based on current vehicle speed, motor torque and other related information.

[0051] By collecting the current vehicle speed, accelerator pedal position information, current motor speed and output torque information, and current gear position information of the transmission in real time, combined with the corresponding working condition identification module, the current working condition information is obtained, and the vehicle speed and acceleration requirements at the next moment are predicted as input information;

[0052] The control problem is then transformed into an optimization problem under given constraints, and the system performance is optimized using the minimum change in control input. The objective function is defined as:

[0053]

[0054] v(t+k) is the speed of the vehicle at the kth point in the prediction time domain, v ref (t+k) is the desired reference speed, utorque(t+k) is the control output of the motor torque, ugear(t+k) is the prediction of the gearbox gear output, δ(t+k) is the predicted information of the accelerator pedal position, δref(t+k) is the desired pedal position, and the control is close to the reference position;

[0055] Q1 is the weighted coefficient of speed error, which controls the accuracy of speed tracking. In order to improve the stability of shift control, some speed fluctuations are allowed appropriately. The empirical value is set to 10. ^ 3;

[0056] Q2 is the weighted coefficient for controlling the motor torque output. A larger coefficient is used to control the algorithm to output the motor torque smoothly, reducing the unevenness and frustration during acceleration. The empirical value is set to 8.

[0057] Q3 is the weighting coefficient of the gear shift line. A larger value is used to avoid a large gear shift amplitude and a sense of frustration. The empirical value is set to 800.

[0058] Q4 is the smooth response of the accelerator pedal, which is a moderate value of 5;

[0059] Adding constraints related to smoothness to the objective function, the constraints limit the changes in motor torque and gearbox gear position, avoiding large mutations in the control quantity of the calculated output, ensuring smoothness, and further matching the current working conditions;

[0060]

[0061] Add vehicle speed constraints:

[0062] v min ≤v(t+k)≤v max

[0063] MPC adjusts the control amount in real time through rolling optimization, ensuring that the control strategy can be adjusted according to the current working conditions at every moment to optimize the driving experience, performance and energy efficiency.

[0064] The output control quantity calculated at the next moment, i.e., the motor torque, the gearbox gear information, and the motor torque, is output to the motor execution drive module. If the gearbox calculates and evaluates the need to shift up or down, it is transmitted to the drive system TCU, and the control information is transmitted to the shift motor to control the corresponding shift fork to perform the shift action.

[0065] Figure 2 In order to adopt the vehicle speed state diagram reflected in the control process of the method of the present invention and the existing method, the simulation time is preset to 250s, a stable load is given in the first 50s, and wind resistance and friction-related resistance are continuously given. At this stage, the existing method does not adjust the torque and gear position in time, resulting in vehicle speed jitter. Under continuous working conditions, the loss of sufficient driving force causes a certain decline in vehicle speed. At this stage, the corresponding model prediction method of the present invention is relatively stable;

[0066] During the 50-150s process, a sudden load is input, and it is set that the vehicle suddenly enters a downhill section and then enters an uphill section. At this time, the load on the downhill section suddenly decreases and the vehicle speed cannot be controlled in time, resulting in jitter. When the load increases on the uphill section, the traditional method fails to adjust the torque in time due to no prediction, resulting in a decrease in vehicle speed jitter. In this process, the model prediction control output corresponding to the present invention is relatively stable, and the fluctuation amplitude and frequency affected by the working conditions are significantly reduced;

[0067] A deceleration condition for obstacle avoidance is set in 150-250s. It can be seen that the model predictive control method adopted by the present invention can accurately control the torque and then control the vehicle speed, quickly respond to the braking demand, and the vehicle speed is stable and without jitter. The existing method has a lag in torque and speed control, and there may be an obvious sense of frustration during the deceleration process.

[0068] The existing methods cannot fully consider sudden changes in operating conditions and dynamic operating conditions. The adjustment and control of the motor torque and the gearbox gear are not accurate and timely enough, and the matching is not precise enough, resulting in vehicle speed jitter. The model predictive control method of the present invention can predict and optimize the control in advance, and can achieve a smooth transition when the operating conditions change. By accurately matching the motor torque and the gearbox gear, possible torque mutations and gear shift shocks can be reduced, thereby making the vehicle and speed control smoother and more stable.

[0069] After adopting the present invention, model predictive control can be used to dynamically adapt to the working conditions. Through some sensors to identify or monitor relevant vehicle real-time data, the torque and gear can be adaptively adjusted within the range allowed by the computing power of the vehicle-mounted platform. Since the model predictive control itself can set multiple control quantities, and the output of the next control quantity is performed within the settable prediction range, the control quantity can be slowly corrected and output, and the next long output can be predicted and controlled in advance. The motor output can be adjusted in advance and the gear shifting timing can be controlled to match the driving state. The gear shifting strategy of the motor and the gearbox can be coordinated through a more sophisticated control method, and the output torque is smoother.

[0070] Under the control of motor torque and gear, the target control quantity is not met rigidly, but the output is controlled through the balance of target and constraint to control the power state of the vehicle. Under the setting of constraints, the rate of change of torque and the amount of change of speed can be controlled to avoid smooth gear shifting and sudden changes in related control quantities, reduce gear shifting shock, and effectively improve comfort during driving.

[0071] During the control process, weighting can be used to adjust the weights in the target parameters according to vehicle type and power requirements to obtain different effects.

[0072] This can not only be applied to the shifting strategy of pure electric commercial vehicles, but also can control the matching and weighted control of the relevant control quantities of the engine and motor of hybrid commercial vehicles.

[0073] This control method can also improve energy efficiency as much as possible and reduce unnecessary energy loss.

[0074] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0075] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A real-time torque control method based on model predictive control, characterized in that: It establishes a dynamic model to predict the working condition at the next moment. The dynamic model includes a working condition identification module, a vehicle speed and state module matching the working condition, and a model prediction control module. The model prediction and control module obtains the optimal solution after obtaining the objective function and constraints through the prediction model. The output control amount corresponding to the optimal solution eliminates the expected value deviation and predicts the output execution amount of the next moment in real time to the actuator of the drive transmission system for execution. The working condition identification module collects the corresponding parameters of the control amount of each working condition in the current and historical input information. The corresponding parameters of the control amount include the target vehicle speed vref, the target motor torque τref, the target gearbox gear gearref, and the target torque J; It predicts future operating conditions based on current and historical input information collected in real time. During the control process, the reference control quantity is designed as the target vehicle speed vref, target motor torque τref, target gearbox gear gearref, and the real-time torque target value that matches the current operating condition for subsequent operating condition identification. At the same time, the real-time torque target value that matches the current operating condition is the reference value for subsequent model predictive control, so that the actual torque target value of the next subsequent operating condition is as close as possible to the actual torque target value in the calculation process of the control output.

2. A real-time torque control method based on model predictive control according to claim 1, characterized in that: First, a vehicle dynamics model is established, that is, a discretized prediction model containing information at each moment, which includes a vehicle speed model and a gearbox gear model. To reduce the complexity of computational control, a simplified linear model is used: Vehicle speed model based on motor torque: v(t+1)=v(t)+Δt·a(t) Where v(t) is the vehicle speed at time t, Δt is the time step, and τmotor(t) is the motor torque. In this process, you can customize the prediction duration. Transmission gear model: u gear (t)=f(v(t),a(t),SOC(t)) Discrete gear information is selected based on the current vehicle speed and motor torque and other related information. The current vehicle speed, accelerator pedal position information, current motor speed and output torque information, and current gear position information of the transmission are collected in real time, and combined with the corresponding working condition identification module, the current working condition information is obtained, and the vehicle speed and acceleration requirements at the next moment are predicted as input information; The control problem is then transformed into an optimization problem under given constraints, and the system performance is optimized using the minimum change in control input. The objective function is defined as: v(t+k) is the speed of the vehicle at the kth point in the prediction time domain, v ref (t+k) is the desired reference speed, utorque(t+k) is the control output of the motor torque, ugear(t+k) is the prediction of the gearbox gear output, δ(t+k) is the predicted information of the accelerator pedal position, δref(t+k) is the desired pedal position, and the control is close to the reference position; Q1 is the weighted coefficient of speed error, which controls the accuracy of speed tracking; Q2 is the weighted coefficient for controlling the motor torque output; Q3 is the weighting coefficient of the speed line shift; Q4 is the smooth response of the accelerator pedal; Adding constraints related to smoothness to the objective function, the constraints limit the changes in motor torque and gearbox gear position, avoiding large mutations in the control quantity of the calculated output, ensuring smoothness, and further matching the current working conditions; Add vehicle speed constraints: v min ≤v(t+k)≤v max MPC adjusts the control amount in real time through rolling optimization, ensuring that the control strategy can be adjusted according to the current working conditions at every moment to optimize the driving experience, performance and energy efficiency.

3. A real-time torque control method based on model predictive control according to claim 2, characterized in that: Q1 is the weighted coefficient of speed error, which controls the accuracy of speed tracking. In order to improve the smoothness of gear shift control, some speed fluctuations are appropriately allowed. The empirical value is set between 10^2 and 10^3.

4. The real-time torque control method based on model predictive control according to claim 2, characterized in that: Q2 is the weighted coefficient for controlling the motor torque output. A larger coefficient is used to control the algorithm to output the motor torque smoothly, reducing the roughness and jerkiness during acceleration. It is set to 8 in the empirical value of 1-10.

5. The real-time torque control method based on model predictive control according to claim 2, characterized in that: Q3 is the weighting coefficient of the gear shifting. A larger value is used to avoid a large gear shift amplitude and a sense of frustration. It is set to 800 in the empirical value of 100-1000.

6. The real-time torque control method based on model predictive control according to claim 2, characterized in that: Q4 is the smooth response of the accelerator pedal and is a moderate value of 5.

7. The real-time torque control method based on model predictive control according to claim 2, characterized in that: The output control quantity calculated at the next moment, i.e., the motor torque, the gearbox gear information, and the motor torque, is output to the motor execution drive module. If the gearbox calculates and evaluates the need to shift up or down, it is transmitted to the drive system TCU, and the control information is transmitted to the shift motor to control the corresponding shift fork to perform the shift action.

Images