A torque control method, device, vehicle and medium in a deceleration and slow driving mode

Abstract

The embodiment of the present invention provides a torque control method, device, vehicle and medium for a deceleration mode, the method comprising: obtaining a first model, the first model is used to indicate the changing relationship between the vehicle torque and the braking strength and vehicle speed in the deceleration mode, wherein the braking strength in the deceleration mode is determined based on the pedal opening and battery state of charge in the deceleration mode; obtaining training data based on the changing relationship, and obtaining a second model according to the training data; wherein the second model is obtained by training based on the pedal opening, vehicle speed and battery state of charge as state variables, the vehicle torque as an action variable, and the torque change rate as a loss function; generating a torque control strategy according to the second model. When the deceleration mode is turned on, the consideration of the braking strength and the battery state of charge is introduced, so that the torque output is continuous and smooth, ensuring that the fluctuation of acceleration and deceleration is small, and achieving the effect of no shaking of the vehicle body and good driving experience.

Description

Technical Field

[0001] The present invention relates to the field of vehicle technology, and in particular to a torque control method for a deceleration mode, a torque control device for a deceleration mode, a corresponding vehicle, and a corresponding computer-readable storage medium. Background Art

[0002] With the improvement of environmental awareness, pure electric vehicles have become one of the main development directions of new energy vehicles. In addition, in terms of carbon emission reduction and energy saving, brake energy recovery technology has become one of the important technologies of pure electric vehicles. As an emerging driving mode, the deceleration driving mode can control the vehicle's acceleration, deceleration and other vehicle states only by operating the accelerator pedal.

[0003] Among them, deceleration and slow-moving braking can not only meet the braking energy recovery function of pure electric vehicles, avoid the continuous pedaling operation of the traditional brake pedal, make the braking process simpler and more convenient, simplify the driver's operation process, and lay the foundation for fully automatic driving; at the same time, it can also reduce the frequency of use of the brake pedal, reduce the number and intensity of braking triggers, thereby reducing the burden of hydraulic braking, and thus extending the service life of the friction plate to a certain extent. However, in the torque control technology related to the deceleration and slow-moving driving mode, there is still a high torque output discreteness and large fluctuations in braking intensity, which leads to large fluctuations in acceleration and deceleration and poor driving experience during the vehicle energy recovery braking process and the switching process with the sliding energy recovery mode. Summary of the invention

[0004] In view of the above problems, embodiments of the present invention are proposed to provide a torque control method for a deceleration mode, a torque control device for a deceleration mode, a corresponding vehicle, and a corresponding computer-readable storage medium that overcome the above problems or at least partially solve the above problems.

[0005] The embodiment of the present invention discloses a torque control method in a deceleration and slow driving mode, the method comprising:

[0006] Acquire a first model, wherein the first model is used to indicate a relationship between a vehicle torque and a braking intensity and a vehicle speed in a deceleration and creeping mode, wherein the braking intensity in the deceleration and creeping mode is determined based on a pedal opening and a battery state of charge in the deceleration and creeping mode;

[0007] Acquire training data obtained based on the change relationship, and obtain a second model according to the training data; wherein the second model is obtained by training based on using the pedal opening, vehicle speed and battery state of charge as state variables, the vehicle torque as an action variable, and the torque change rate as a loss function;

[0008] A torque control strategy is generated according to the second model.

[0009] The embodiment of the present invention further discloses a torque control device in a deceleration and slow driving mode, the device comprising:

[0010] A first model acquisition module, used to acquire a first model, wherein the first model is used to indicate a relationship between a vehicle torque and a braking intensity and a vehicle speed in a deceleration mode, wherein the braking intensity in the deceleration mode is determined based on a pedal opening and a battery state of charge in the deceleration mode;

[0011] a second model acquisition module, configured to acquire training data obtained based on the change relationship, and obtain a second model according to the training data; wherein the second model is obtained by training by taking the pedal opening, vehicle speed and battery state of charge as state variables, taking the vehicle torque as an action variable, and taking the torque change rate as a loss function;

[0012] A torque control strategy generating module is used to generate a torque control strategy according to the second model.

[0013] An embodiment of the present invention further discloses a vehicle, comprising: a processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, any one of the torque control methods of the deceleration and slow driving mode is implemented.

[0014] An embodiment of the present invention further discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, any of the torque control methods for the deceleration and creeping mode is implemented.

[0015] The embodiments of the present invention include the following advantages:

[0016] In an embodiment of the present invention, the first model can be used to indicate the changing relationship between the vehicle torque and the braking intensity and vehicle speed in the deceleration mode, and the braking intensity in the deceleration mode can be determined based on the pedal opening and the battery state of charge in the deceleration mode, that is, the changing relationship between the braking intensity, the pedal opening and the battery state of charge can be introduced into the changing relationship, and the consideration of the battery energy recovery capability is increased based on the introduction of the battery state of charge, thereby realizing the combination of the vehicle data in the deceleration mode with the braking intensity of the oil vehicle and the coasting energy recovery braking intensity; then, the data obtained based on the aforementioned changing relationship can be obtained as the training data for the second model, so that the torque control can ensure that the braking intensity is stable under different vehicle speeds, pedal openings and battery states of charge. The second model can be trained based on the trained second model, wherein the pedal opening, vehicle speed and battery state of charge can be used as state variables, vehicle torque can be used as action variable, and torque change rate can be used as loss function to generate a torque control strategy. When the deceleration mode is turned on, the generated torque control strategy can determine the optimal output torque change rate as the goal, so that the torque change rate is small and the target torque output is continuous and smooth, thereby ensuring that the fluctuation of acceleration and deceleration is small, and realizing smooth transition of switching between the deceleration mode and the gliding energy recovery mode, thereby achieving the effect of no shaking of the vehicle body and good driving experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flowchart of a torque control method in a deceleration and creeping mode according to an embodiment of the present invention;

[0018] Figure 2 is a flowchart of another torque control method embodiment of a deceleration and creeping mode of the present invention;

[0019] Figure 3 is a schematic diagram of building a first model provided by an embodiment of the present invention;

[0020] Figure 4 is a schematic diagram of the braking intensity output of the energy recovery model provided by an embodiment of the present invention;

[0021] Figure 5 is a schematic diagram of building a second model provided by an embodiment of the present invention;

[0022] Figure 6 It is a structural block diagram of an embodiment of a torque control device for a deceleration and creeping mode of the present invention. DETAILED DESCRIPTION

[0023] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] In the torque control technology related to the deceleration and slow driving mode, as an example, the torque control related to the deceleration and slow driving mode can be achieved by taking a more detailed correction and calibration approach to the pedal opening-vehicle speed-torque Map (Pedal Map, which refers to a relationship diagram in which the powertrain output torque changes with the accelerator pedal opening and the vehicle speed). However, the aforementioned method will lead to a significant increase in calibration work due to the need for more detailed calibration, and due to the limitations on the factors considered in the Map, it cannot cover the remaining factors in the deceleration and slow driving mode, such as braking intensity, battery state of charge, etc.; as another example, the driving intention can be deeply integrated with the single-pedal control technology to establish a nonlinear model predictive control method for a single pedal of a pure electric vehicle, so as to achieve the torque control related to the deceleration and slow driving mode. However, the aforementioned method has many conversions from the pedal opening to the output target torque, which may cause large errors and extended time, and may be difficult to calculate due to the complexity of the objective function.

[0025] In summary, in the torque control technology related to the deceleration driving mode, there are still problems such as high torque output discreteness and large fluctuations in braking intensity, which lead to large fluctuations in acceleration and deceleration and poor driving experience for the driver during the vehicle energy recovery braking process and the switching process with the gliding energy recovery mode.

[0026] The embodiments of the present invention can indicate the changing relationship between the vehicle torque and the braking intensity and the vehicle speed in the deceleration mode, and the braking intensity in the deceleration mode can be determined based on the pedal opening and the battery state of charge in the deceleration mode. The changing relationship between the braking intensity, the pedal opening and the battery state of charge can be introduced into the changing relationship, and based on the introduction of the battery state of charge, the consideration of the battery energy recovery capability can be increased. The aforementioned braking intensity can be used to indicate the braking intensity of a gasoline vehicle and / or the coasting energy recovery braking intensity, thereby realizing the combination of the vehicle data in the deceleration mode with the braking intensity of a gasoline vehicle and the coasting energy recovery braking intensity. Furthermore, the pedal opening, vehicle speed and battery state of charge can be used as state variables, the vehicle torque can be used as an action variable, and the torque change rate can be used as a loss function to train a second model, and a torque control strategy can be generated based on the trained second model. When the deceleration mode is turned on, under different vehicle speeds, pedal openings and battery states of charge, the torque change rate is determined to be optimal, so that the torque change rate is smaller and the target torque output is continuous and smooth, thereby ensuring that the fluctuation of acceleration and deceleration is small, and achieving a smooth transition between the deceleration mode and the gliding energy recovery mode, thereby achieving the effect of no vehicle body shaking and good driving experience.

[0027] Specifically, an energy recovery model can be established based on the Map in the deceleration mode, the brake pedal opening - vehicle speed - torque Map of the fuel vehicle, and the coasting energy recovery Map data. Combined with the vehicle dynamics model, battery model, and motor model, a torque control model of the deceleration mode of the pure electric vehicle with the accelerator pedal opening, vehicle speed, and battery state of charge as input and the target torque as output can be obtained. In the deceleration mode, the brake torque output is smooth for different pedal openings and vehicle speeds of the driver, so as to achieve a smooth transition between the deceleration mode and the coasting energy recovery mode; and, in the torque When the control model continuously outputs, the transition and connection between the coasting energy recovery braking intensity and the deceleration energy recovery braking intensity below 5% of the pedal opening can be considered, so that the braking intensity fluctuation is slowed down, thereby reducing the acceleration and deceleration fluctuation of the vehicle. The braking intensity when the brake pedal opening of the oil vehicle is 0-25% can also be considered to fully consider the driving habits of road drivers, increase the driver's driving consistency experience, make driving behavior easier to use, and provide a good driving experience; and, by proposing a deep reinforcement learning control strategy that can solve the problem of continuous torque output, a deep Q-Learining algorithm (Deep Q-Network, DQN for short) is adopted to use the pedal opening, vehicle speed, and battery state of charge as state variables, output torque as the control variable, and torque change rate as the loss function. On the basis of determining the optimal output torque change rate as the goal, a torque output with the goal of continuous and smooth torque output and high driving experience is achieved, so that the torque change rate is small and the target torque output is continuous and smooth.

[0028] Reference Figure 1 , shows a flowchart of a torque control method in a deceleration mode according to an embodiment of the present invention, which may specifically include the following steps:

[0029] Step 101, obtaining a first model, the first model is used to indicate the relationship between the vehicle torque and the braking intensity and the vehicle speed in the deceleration mode;

[0030] The first model may refer to a torque control model of a deceleration mode of a pure electric vehicle.

[0031] In the relationship between the vehicle torque indicated by the first model and the braking intensity and vehicle speed change in the deceleration mode, the braking intensity in the deceleration mode can be determined based on the pedal opening and the battery state of charge (State of Charge, SOC) in the deceleration mode. At this time, based on the introduction of the battery state of charge, the consideration of the battery energy recovery capability can be increased, and based on the introduction of the braking intensity, the aforementioned braking intensity can be used to indicate the braking intensity of the oil vehicle and / or the coasting energy recovery braking intensity, thereby realizing the combination of the vehicle data in the deceleration mode with the braking intensity of the oil vehicle and the coasting energy recovery braking intensity.

[0032] In practical applications, the deceleration mode can be a specific application of a single-pedal vehicle, and the single pedal can mainly be an accelerator pedal. The determination of the braking intensity in the deceleration mode is reflected in the determination of the braking deceleration, which can be specifically manifested as the identification of the acceleration intention based on the pedal opening of the accelerator pedal, and the output of the corresponding braking deceleration. For example, the smaller the pedal opening, the weaker the acceleration intention, and energy recovery is required at this time, and the output braking deceleration may be greater. It should be noted that the torque control model may involve a combination of Map data in different modes, and the braking intensity indicated by it in the deceleration mode may be reflected as a combination of the braking intensity of the oil vehicle and the braking intensity of the coasting energy recovery in the deceleration mode, that is, the combination of the deceleration mode with the braking form of the oil vehicle and the coasting energy recovery mode can be achieved; the changing relationship between the vehicle torque indicated by it and the braking intensity and vehicle speed in the deceleration mode can not only increase the consideration of the battery energy recovery capability and the braking intensity, but also the changing relationship can make the torque control model continuously output torque in the deceleration mode. The braking torque output is smooth and the driving experience is smooth at different pedal openings and vehicle speeds, and it can be combined with the braking form of the oil vehicle to make the driving experience familiar, and at the same time, the deceleration mode and the energy recovery mode can be switched smoothly to achieve the purpose of no sense of frustration.

[0033] Among them, the first model can be mainly built based on the energy recovery model, the vehicle dynamics model, the battery model and the motor model. Specifically, it can be manifested as combining the deceleration mode with the braking form of the oil vehicle and the Map data under the gliding energy recovery mode according to the Map diagram under the deceleration mode, combined with the brake pedal opening of the fuel vehicle-vehicle speed-torque Map, and combined with the gliding energy recovery Map data, to achieve the energy recovery model; and then combine the established energy recovery model with the vehicle dynamics model, the battery model, and the motor model to obtain the torque control model. The specific first model building process is not limited in the embodiment of the present invention.

[0034] Step 102, obtaining training data based on the change relationship, and obtaining a second model according to the training data;

[0035] The second model may be used to obtain a DQN torque control strategy, that is, the second model may be substantially a torque control model, such as a DQN model for outputting the aforementioned DQN torque control strategy.

[0036] In one embodiment of the present invention, the input data and output data of the aforementioned torque control model can be used as training data for training the second model, so that the first group of relevant data can be used as relevant control variables of the second model to train and generate the corresponding torque control strategy.

[0037] Among them, the training data obtained based on the changing relationship may include the input data of the first model and the output data obtained by the first model based on the input data. Specifically, the input data may include pedal opening, vehicle speed and battery state of charge, and the output data may include vehicle torque. Specifically, it can be expressed as using the torque control model constructed as mentioned above. When a set of data of pedal opening, vehicle speed and battery state of charge is input, the torque control model can output torque. At this time, the data obtained as mentioned above can be used as training data for the DQN control strategy.

[0038] In practical applications, the second model can be mainly obtained by adopting the deep Q-Learining algorithm (i.e. DQN), based on the pedal opening, vehicle speed and battery state of charge as state variables, vehicle torque as action variable, and torque change rate as loss function training, so that on the basis of determining the optimal output torque change rate as the goal, the torque change rate can be made smaller, and the target torque output can be made continuous and smooth, thereby ensuring small fluctuations in acceleration and deceleration.

[0039] Step 103: Generate a torque control strategy according to the second model.

[0040] After obtaining the second model, ie, the DQN model, a DQN torque control strategy may be obtained based on the model to implement corresponding torque control of the vehicle based on the output torque control strategy, thereby ensuring smooth control of the vehicle torque.

[0041] In one embodiment of the present invention, the converged state variables and action variables output by the second model can be used as a torque control strategy for the current vehicle state, wherein the converged state variables can be used to indicate the pedal opening, vehicle speed and battery state of charge obtained after convergence, and the converged action variables can be used to indicate the vehicle torque obtained after convergence. At this time, the vehicle can be controlled accordingly based on the pedal opening, vehicle speed and battery state of charge obtained after convergence, and the vehicle torque obtained after convergence, and the embodiment of the present invention is not limited to this.

[0042] In an embodiment of the present invention, the first model can be used to indicate the changing relationship between the vehicle torque and the braking intensity and vehicle speed in the deceleration mode, and the braking intensity in the deceleration mode can be determined based on the pedal opening and the battery state of charge in the deceleration mode, that is, the changing relationship between the braking intensity, the pedal opening and the battery state of charge can be introduced into the changing relationship, and the consideration of the battery energy recovery capability is increased based on the introduction of the battery state of charge, thereby realizing the combination of the vehicle data in the deceleration mode with the braking intensity of the oil vehicle and the coasting energy recovery braking intensity; then, the data obtained based on the aforementioned changing relationship can be obtained as the training data for the second model, so that the torque control can ensure that the braking intensity is stable under different vehicle speeds, pedal openings and battery states of charge. The second model can be trained based on the trained second model, wherein the pedal opening, vehicle speed and battery state of charge can be used as state variables, vehicle torque can be used as action variable, and torque change rate can be used as loss function to generate a torque control strategy. When the deceleration mode is turned on, the generated torque control strategy can determine the optimal output torque change rate as the goal, so that the torque change rate is small and the target torque output is continuous and smooth, thereby ensuring that the fluctuation of acceleration and deceleration is small, and realizing smooth transition of switching between the deceleration mode and the gliding energy recovery mode, thereby achieving the effect of no shaking of the vehicle body and good driving experience.

[0043] Reference Figure 2 , shows a flowchart of another torque control method embodiment of the deceleration and slowing mode of the present invention, which may specifically include the following steps:

[0044] Step 201, constructing a first model based on an energy recovery model, a vehicle dynamics model, a battery model, and a motor model;

[0045] The first model may refer to a torque control model of a pure electric vehicle in a deceleration mode. Specifically, the first model may be constructed based on an energy recovery model, a vehicle dynamics model, a battery model, and a motor model.

[0046] Among them, the energy recovery model can be mainly used to describe the conversion of waste or energy wasted in the process into reusable energy; the vehicle dynamics model can be mainly used to analyze the smoothness and handling stability of the vehicle; the battery model can be mainly used to predict battery performance; the motor model can be mainly used to describe the relationship between electromagnetic quantities such as voltage, current, magnetic flux and mechanical quantities such as torque and speed of the synchronous motor.

[0047] It should be noted that the change relationship can be reflected by using Map data or Map diagram.

[0048] In one embodiment of the present invention, it can be specifically manifested as combining the deceleration mode with the braking form of the fuel vehicle and the Map data under the coasting energy recovery mode according to the Map diagram under the deceleration mode, combining the brake pedal opening-vehicle speed-torque Map of the fuel vehicle, and combining the coasting energy recovery Map data, to establish an energy recovery model; and then combining the established energy recovery model with the vehicle dynamics model, battery model, and motor model to obtain a torque control model.

[0049] In the process of combining various models, Figure 3 As shown, the battery model can be mainly used to provide the battery state of charge at the previous moment to the energy recovery model; the energy recovery model can be mainly used to provide the braking intensity to the vehicle dynamics model based on the battery state of charge at the previous moment provided by the battery model and the input pedal opening; the vehicle dynamics model can be mainly used to provide the first required power to the battery model based on the braking intensity and vehicle speed provided by the energy recovery model; the first required power can be mainly used to indicate the driving required power, and the battery model can be mainly used to provide the second required power to the motor model based on the driving required power provided by the vehicle dynamics model; the second required power can be used to indicate the motor input power and the motor output efficiency, and the motor model can be mainly used to output the vehicle torque based on the motor input power and the motor output efficiency provided by the battery model, thereby realizing the construction of the torque control model.

[0050] Among them, in the process of combining the Map data under the deceleration mode with the braking form of the oil vehicle and the coasting energy recovery mode, the energy recovery model can mainly add consideration of the battery energy recovery capability based on the existing Map data under the original deceleration model, and establish an energy recovery model with the pedal opening, vehicle speed and SOC value as input and the braking intensity as output. In practical applications, the braking intensity output by the energy recovery model can provide parameters for the dynamics of the car (vehicle).

[0051] The automobile (vehicle) dynamics model can be used to simulate the vehicle, which can be achieved by performing a dynamic analysis on the vehicle and then establishing the relationship between the required braking intensity and the driving power. Specifically, in the deceleration mode, the driving required power is composed of the power of rolling resistance, air resistance, and acceleration resistance. When the slope resistance is ignored, the vehicle dynamics model can be expressed through the following formula:

[0052] P req =(-F f -F w +F j )

[0053] F f =fmg

[0054]

[0055]

[0056] In formula 1, P req It can provide the required power for the vehicle drive, F f Can be the rolling resistance, F w can be the air resistance, F j can be the acceleration resistance, v can be the vehicle speed, z can be the braking strength, m can be the vehicle weight, f can be the vehicle friction coefficient, C D can be the lift coefficient of the whole vehicle, and A can be the frontal area of ​​the whole vehicle.

[0057] Among them, the vehicle driving demand power refers to the power required by the vehicle to meet the current driving state and demand. When the vehicle needs to accelerate, climb or overcome driving resistance, the driving demand power will increase accordingly; the above parameter symbols distinguish between positive and negative, "+" can be used to indicate driving, and "-" can be used to indicate braking feedback, which is not limited by the embodiment of the present invention. In practical applications, the braking intensity z can be obtained through the output of the energy recovery model. At this time, the vehicle dynamics model can calculate the acceleration resistance based on the braking intensity provided by the energy recovery model, and then substitute the calculated acceleration resistance, rolling resistance and air resistance into the above formula to obtain the vehicle driving demand power, which is used to provide the required power to the battery model.

[0058] The battery model may refer to a power lithium-ion battery model, which may be simplified to a first-order approximate internal resistance model. The specific SOC calculation formula may be as follows:

[0059]

[0060] SOC(k)=SOC(k-1)-∫Idt Formula 2

[0061] In formula 2, SOC(k) can be the battery SOC value at the current moment, SOC(k-1) can be the battery SOC value at the previous moment, U can be the open circuit voltage value, I can be the current at the current moment, and P can be the current current value. b can be the battery output power, and R can be the battery equivalent internal resistance.

[0062] In practical applications, the battery output power used for calculation by the battery model can be matched based on the driving demand power provided by the energy recovery model, wherein the battery output power refers to the maximum power transmission rate that the battery can provide. In order to ensure that the vehicle can run normally and efficiently, the battery output power usually meets or exceeds the current driving demand power; and the battery state of charge at the current moment output by the battery model based on the battery output power can be used to provide the energy recovery model as the battery state of charge at the previous moment at the next moment, so as to be used as a parameter for the calculation of the energy recovery model.

[0063] The motor model can be simplified by using a motor-speed-torque model based on experimental data. The motor output and input power formulas can be shown as follows:

[0064]

[0065] In formula 3, P out Can be used to output power to the motor, P in The input power for the motor is T in Can be the vehicle braking torque, T out can be the vehicle driving torque, e m can be the motor efficiency, e pc can be the power converter efficiency, and n can be the motor speed. Among them, torque and efficiency can be obtained through experiments, and the model can be directly established by looking up the table; the motor input power can refer to the electrical energy actually received by the motor, and it is converted into mechanical energy to drive the vehicle; the motor output power can refer to the mechanical power actually generated by the motor in the working state; the braking torque is the torque generated by the brake, and the driving torque is the torque that drives the prime mover to rotate. Specifically, torque is the basic power parameter of the engine output, and the braking torque and driving torque can be the specific application of this power in the vehicle driving process.

[0066] In practical applications, the required power provided by the battery model to the motor model, that is, the battery output power, can be used to match the motor input power and the motor output efficiency, so that the motor model can output the vehicle braking torque and the vehicle driving torque based on the motor input power and the motor output efficiency, respectively, thereby realizing the output of vehicle torque.

[0067] Step 202, obtaining the vehicle torque output by the first model by inputting the pedal opening, the vehicle speed and the battery state of charge into the first model;

[0068] like Figure 3 As shown, the first model constructed, namely the torque control model, can realize input based on the pedal opening, vehicle speed and battery state of charge, thereby outputting the vehicle torque.

[0069] In practical applications, the input data and output data of the first model can be used as a set of training data for training the second model.

[0070] Specifically, the battery state of charge at the previous moment can be obtained through the battery model as the battery state of charge training data. At this time, the braking intensity training data can be obtained through the energy recovery model based on the pedal opening training data and the battery state of charge training data. Then, the vehicle torque training data can be calculated based on the braking intensity training data and the vehicle speed training data through the vehicle dynamics model, battery model and motor model.

[0071] Specifically, the energy recovery model can be used to output the braking intensity training data based on the pedal opening training data and the battery state of charge training data; then, the vehicle dynamics model can be used to output the first required power based on the braking intensity training data and the vehicle speed training data according to the above formula 1. The outputted first required power can be used to indicate the driving required power. At this time, the battery model can be used to output the second required power based on the first required power according to the above formula 2. The second required power can be used to indicate the motor input power and the motor output efficiency. Specifically, the battery output power can be obtained based on the driving required power matching, and the motor input power and the motor output efficiency can be obtained based on the battery output power matching. The embodiment of the present invention does not limit the specific matching process; finally, the motor model can be used to output the vehicle torque training data based on the second required power. Specifically, it can be expressed as follows: according to the above formula 3, the motor model outputs the vehicle braking torque and the vehicle driving torque based on the motor input power and the motor output efficiency, respectively, so as to achieve the output of the vehicle torque.

[0072] It should be noted that the battery state of charge at the current moment obtained by the battery model based on the battery output power output can be used to provide the battery state of charge at the previous moment to the energy recovery model at the next moment, so as to be used as a parameter for the calculation of the energy recovery model. This is not limited in this embodiment of the present invention.

[0073] In practical applications, the battery state of charge training data specifically used in the energy recovery model can be the battery state of charge at the previous moment. In the process of obtaining real-time and continuous braking intensity output under different vehicle speeds and pedal openings, the deceleration mode can be combined with the oil vehicle braking form and the Map data under the gliding energy recovery mode. Specifically, it can be manifested as adding consideration of the battery energy recovery capability based on the existing Map data under the original deceleration model, and establishing a system with pedal opening, vehicle speed and SOC value as input and braking intensity as output.

[0074] In the embodiment of the present invention, the braking intensity output of the energy recovery model can consider the transition and connection between the coasting energy recovery braking intensity and the deceleration energy recovery braking intensity when the pedal opening is below 5%, so as to reduce the fluctuation of the braking intensity and thus reduce the acceleration and deceleration fluctuation of the vehicle; and, the braking intensity when the brake pedal opening of the oil vehicle is 0-25% can also be considered, so as to fully consider the driving habits of road drivers, increase the driver's driving consistency experience, make the driving behavior easier to use, and provide a better driving experience.

[0075] Specifically, after the deceleration mode of the pure electric vehicle is turned on, different braking intensities can be determined based on different pedal opening ranges and different battery charge states.

[0076] In one case, when the pedal opening training data is less than a preset first threshold, such as 25%, if the battery state of charge at the previous moment is greater than or equal to a preset second threshold, such as 90%, then the current braking intensity training data can be determined to be 0; in another case, when the pedal opening training data is less than the preset first threshold, such as 25%, if the battery state of charge at the previous moment is less than the preset second threshold, such as 90%, then the initial value of the braking intensity can be obtained, and the initial value of the braking intensity can be controlled to change to obtain the braking intensity training data.

[0077] The initial value of the braking intensity can be obtained by calibrating the Map data in the original deceleration mode. Specifically, the first change relationship can be obtained, which can be used to indicate the change relationship between different vehicle speeds and different pedal openings and braking intensity in the deceleration mode. At this time, the first braking intensity value calibrated corresponding to the pedal opening training data can be obtained based on the first change relationship, and the first braking intensity value is used as the initial value of the braking intensity.

[0078] The control strategy for changing the initial value of the braking intensity can be executed based on different opening ranges of the pedal opening.

[0079] In one case, if the pedal opening training data is within a preset first numerical range, such as 5%-25%, a second change relationship can be obtained, and the second change relationship can be used to indicate the change relationship between the oil vehicle at different vehicle speeds and different pedal openings and the braking intensity. At this time, the initial value of the braking intensity can be controlled based on the second change relationship to obtain a first braking intensity target value, which is subsequently used as the first braking intensity training data. The obtained first braking intensity training data can be used to indicate the braking intensity of the oil vehicle, that is, the braking intensity target value obtained under the preset change strategy can be the final braking intensity obtained by controlling the braking intensity of the deceleration mode according to the vehicle speed, pedal opening, and SOC with reference to the fuel vehicle brake pedal opening-vehicle speed-torque Map, thereby realizing the combination of the deceleration mode and the Map data of the oil vehicle braking form; in another case If the pedal opening training data is within a preset second numerical range, such as 0-5%, a third change relationship can be obtained, and the third change relationship can be used to indicate the change relationship between different vehicle speeds and different pedal openings and braking intensity in the gliding energy recovery mode. At this time, the initial value of the braking intensity can be controlled based on the third change relationship to obtain a second braking intensity target value, which is subsequently used as the second braking intensity training data. The obtained second braking intensity training data can be used to indicate the gliding energy recovery braking intensity, that is, the braking intensity target value obtained after the initial value of the braking intensity is controlled to change. It can be the final braking intensity obtained by controlling the braking intensity of the deceleration mode according to the vehicle speed, pedal opening, and SOC after taking the gliding energy recovery Map data as a reference, thereby realizing the combination of the Map data in the deceleration mode and the gliding energy recovery mode.

[0080] In practical applications, in the process of controlling the initial value of braking intensity based on Map data indicating the braking form of an oil vehicle and / or Map data indicating the gliding energy recovery mode, it can be expressed as assigning different weight coefficients to different modes, thereby achieving change control of the initial value of braking intensity.

[0081] As an example, in the process of controlling the initial value of braking intensity based on the second change relationship, a second braking intensity value calibrated corresponding to the pedal opening training data can be obtained from the second change relationship, and then the first braking intensity training data is calculated by multiplying the second braking intensity value with the preset first weight, and the product of the initial value of braking intensity with the preset second weight; as another example, in the process of controlling the initial value of braking intensity based on the third change relationship, a third braking intensity value calibrated corresponding to the pedal opening training data can be obtained from the third change relationship, and then the second braking intensity training data is calculated by multiplying the third braking intensity value with the preset third weight, and the product of the initial value of braking intensity with the preset fourth weight.

[0082] The braking intensity may be used to indicate the braking deceleration, which may mainly refer to the deceleration required at the current speed.

[0083] For example, Figure 4 As shown, after the deceleration mode is turned on, when the pedal opening is less than a preset first threshold, such as 25%, a Map of the pedal opening, vehicle speed and braking intensity can be obtained as an initial value, which can be obtained through actual calibration. At this time, if the battery SOC value is lower than a preset second threshold, such as 90%, the maximum deceleration of the electric brake can gradually change from 0.3g (g is the acceleration of gravity) to 0 with the numerical range of the pedal opening, such as 0-25%; considering that the braking intensity of the battery will be limited at a high SOC, if the battery SOC value is greater than or equal to the preset second threshold, such as 90%, the electric braking intensity is reduced to 0%.

[0084] It should be noted that the change relationship can be reflected by using Map data or Map diagram.

[0085] like Figure 4As shown, for the case where the pedal opening is less than the preset first threshold and the battery SOC value is lower than the preset second threshold, further, as an example, in order to realize the combination of the Map diagram in the deceleration mode and the brake pedal opening-vehicle speed-torque Map of the fuel vehicle, the brake pedal opening-vehicle speed-torque data of the fuel vehicle can be collected at this time, for example, only the data of the brake pedal opening of 0-25% is collected, but because the pedal opening below 5% belongs to the sliding energy recovery mode, in fact, the preset first numerical range, such as (5%, 25%] of the fuel vehicle brake pedal opening-vehicle speed-torque data can be collected, and the braking torque of this group of data can be used as the braking torque reference value of the deceleration mode, so that the driver can have a consistent braking feel when driving the fuel vehicle and the electric vehicle, and increase the driving consistency. Specifically, it can be expressed as the product of the braking intensity value calibrated corresponding to the pedal opening and the preset second weight, such as 0.5, obtained from the first change diagram of the pedal opening-vehicle speed-torque data in the deceleration mode, and the brake strength value calibrated corresponding to the pedal opening and the preset second weight, such as 0.5, and the brake pedal opening of the fuel vehicle. -The braking intensity value calibrated corresponding to the pedal opening obtained from the second change diagram of the vehicle speed-torque data and the preset first weight, such as 0.5, is accumulated to obtain the output braking intensity value; as another example, in order to realize the combination of the Map diagram in the deceleration mode and the sliding energy recovery Map data, the sliding energy recovery map diagram can be collected at this time as a reference for the braking torque in the deceleration mode where the pedal opening is within the preset second numerical range, such as 0-5%, so that the switching between the deceleration mode and the sliding capacity recovery can be smooth and the vehicle body does not shake. Specifically, it can be manifested as the braking intensity value calibrated corresponding to the pedal opening obtained from the first change diagram of the pedal opening-vehicle speed-torque data in the deceleration mode and the preset third weight, such as 0.3, and the braking intensity value calibrated corresponding to the pedal opening obtained from the third change diagram of the sliding energy recovery Map data and the preset fourth weight, such as 0.7, are accumulated to obtain the output braking intensity value.

[0086] It should be noted that the specific values ​​of the preset first threshold, preset second threshold, preset first numerical range, preset second numerical range, preset first weight, preset second weight, preset third weight and preset fourth weight mentioned in the embodiments of the present invention can be determined through actual needs or experiments, and the embodiments of the present invention are not limited to this.

[0087] Step 203, using the input data and output data of the first model as training data to train a second model;

[0088] The second model may be used to obtain a DQN torque control strategy, that is, the second model may be substantially a torque control model, such as a DQN model for outputting the aforementioned DQN torque control strategy.

[0089] In one embodiment of the present invention, the input data and output data of the aforementioned torque control model can be used as training data for training the second model, so that the first group of related data can be used as the related control variables of the second model to train and generate the corresponding torque control strategy. Specifically, the second model can be mainly obtained by adopting a deep Q-Learining algorithm, based on the pedal opening training data, vehicle speed training data and battery state of charge training data as state variables, the vehicle torque training data as action variables, and the torque change rate as a loss function training, so that the torque change rate can be smaller on the basis of determining the optimal output torque change rate as the goal, and the target torque output is continuous and smooth, thereby ensuring that the fluctuation of acceleration and deceleration is small.

[0090] In a specific implementation, the pedal opening training data, vehicle speed training data and battery state of charge training data can be used as state variables to construct a state space, and the vehicle torque training data can be used as an action variable; and the torque change rate can be obtained, and the minimum torque change rate can be used as an iteration condition, and then the objective function for the second model can be constructed based on the state space, action variables and the aforementioned iteration conditions, so as to optimize the objective function and obtain the second model.

[0091] For example, assuming that the state variable is S t , the vehicle speed v, pedal opening k and SOC can be selected as state variables, and the state space S is constructed by the above parameters, which can be specifically expressed as S t ={v,k,SOC}; Assume the action variable is A t In order to achieve the purpose of continuous and smooth torque output, the action variable can be expressed as output torque, such as A t ={T, ω1, ω2, ω3}; In order to use the continuous and smooth torque output and high driving experience as the loss function of DQN, the torque change rate can be used as the reward function, which can be specifically expressed as r=T^=dT / dt; Then the expectation of the minimum cumulative reward can be used as the objective function, which can be specifically expressed as follows:

[0092]

[0093] In Formula 4, J can be the optimization objective function, Q*(s,a) can be the optimal state-action value function, γ can be the discount factor, and r can be the immediate reward.

[0094] In practical applications, the deep Q learning algorithm can obtain the optimal strategy by continuously training and iterating the Q network. Specifically, the greedy algorithm can be used to select actions according to the current state of the vehicle (select the action value corresponding to the largest Q value with probability ε, and select other actions with probability 1-ε) to avoid falling into the local optimum; secondly, the current reward value and the state at the next moment can be calculated through the current state and the selected action, and stored in the experience pool. When the experience pool is stored to a certain number, in order to avoid the correlation of training samples and improve training efficiency, samples can be randomly selected from the experience pool for network training.

[0095] Specifically, in addition to the DQN loss function, the second model may also involve an estimated value network (ValueNetwork), a target value network (Target Network) and a recall memory unit (Experience Replay).

[0096] Among them, the estimated value network in the DQN model can mainly play the role of estimating the state value. The network can accept the state as input and output the Q value corresponding to each possible action. The Q value can represent the expected return of performing different actions in the current state. It can be mainly through continuous updating and optimization of this network that DQN can gradually learn how to select the optimal action according to the current state; the role of the target value network in the DQN model is mainly to improve the stability and convergence of training. The calculation of the network target value can usually use the existing Q value, and the target value network, as a network with a slower update, can be used to provide the Q value required for calculation. By introducing the target value network, DQN can alleviate the problem of training instability caused by changes in data distribution to a certain extent; the recall memory unit can be mainly used in the DQN model to store and replay past experiences (such as state transitions, actions and rewards). In the network training process, it involves iteration and optimization of the estimated value network, the target value network and the recall memory unit.

[0097] In one embodiment of the present invention, the process of network training can be reflected in the optimization of the constructed objective function, so as to obtain the second model after training. Specifically, it can be expressed as optimizing the objective function, obtaining the optimal action variable and reward value under the current state variable corresponding to the state space, and the reward value can be used to indicate the reward value for the loss function. Then, the state variable at the next moment can be obtained, and the current state variable corresponding to the state space, the optimal action variable and reward value under the current state variable, and the state variable at the next moment are stored in the preset recall unit; when the storage amount of the preset recall unit reaches the preset third threshold, the preset estimated value network can be reversely updated using the loss function gradient descent method, and the weight parameters of the preset estimated value network are copied to the preset target value network according to the preset step length, and the current weight parameter copy number is obtained, and the weight parameter copy number can be used to indicate the number of iterations. When the number of iterations reaches the preset fourth threshold, if the difference between the cumulative evaluation values ​​of adjacent iterations is less than the preset fifth threshold, it means that the network training is completed, thereby obtaining the second model obtained after training.

[0098] For example, Figure 5 As shown, the basic parameters of DQN can be initialized. Assume that the number of randomly selected samples is n, the number of iterations is 5000, the evaluation network parameter of the estimated value network is ζ, the target network parameter of the target value network is ζ-, the exploration rate is ε (a parameter between 0 and 1. At each time step, the algorithm will decide whether to choose the action with the highest current estimated value, that is, greedy selection, or randomly select an action based on the value of ε), and the discount factor is γ. At this time, the ε-greedy strategy (a strategy commonly used in reinforcement learning to balance exploration and utilization) can be used according to the selected state and action to obtain the current state s t The best action a under t , and the current state s t 、Action a t , return value r t and the state s at the next moment t+1 Stored in the recall memory unit; when the experience pool, that is, the recall memory unit stores samples of the preset third threshold value, n samples can be randomly selected from the experience pool for network training. The specific network training method can use the loss function gradient descent method L(θ)=E{[r+γmin Q(s t+1 , a a+1 θ - )-Q(s t ,a t ;θ)] 2}Reversely update the evaluation network, and copy the weight parameters of the estimated value network to the target value network every N=50 steps. At this time, it can be determined whether t has reached the predetermined number of iterations. If t has not reached the predetermined number of iterations, the iteration is repeated. If t has reached the predetermined number of iterations, it can be determined whether the difference between the cumulative evaluation Q values ​​of adjacent iterations is less than the preset fourth threshold. If the difference between the cumulative evaluation Q values ​​of adjacent iterations is less than the preset fourth threshold, it can indicate that the iteration is over, otherwise iterative training will need to continue.

[0099] Step 204: Generate a torque control strategy according to the second model.

[0100] After obtaining the second model, ie, the DQN model, a DQN torque control strategy may be obtained based on the model to implement corresponding torque control of the vehicle based on the output torque control strategy, thereby ensuring smooth control of the vehicle torque.

[0101] In one embodiment of the present invention, the converged state variables and action variables output by the second model can be used as the torque control strategy for the current vehicle state. For example, the strategy corresponding to the minimum value of Q(s,a) after convergence can be used as the optimal strategy. The converged state variables can be used to indicate the pedal opening, vehicle speed and battery state of charge obtained after convergence, and the converged action variables can be used to indicate the vehicle torque obtained after convergence. At this time, the vehicle can be controlled accordingly based on the pedal opening, vehicle speed and battery state of charge obtained after convergence, and the vehicle torque obtained after convergence, and the embodiment of the present invention is not limited to this.

[0102] In a preferred embodiment of the present invention, the first model (i.e., the torque control model) can be used to provide the second model with the training data required by the deep reinforcement learning algorithm. After the second model (i.e., the DQN model) is built, the DQN model can be input into the vehicle controller to realize the output of the torque control strategy under different vehicle speeds and pedal openings, and then the vehicle torque is controlled accordingly based on the real-time continuous output braking intensity. This is not limited by the embodiment of the present invention.

[0103] In an embodiment of the present invention, the first model can be used to indicate the changing relationship between the vehicle torque and the braking intensity and vehicle speed in the deceleration mode, and the braking intensity in the deceleration mode can be determined based on the pedal opening and the battery state of charge in the deceleration mode, that is, the changing relationship between the braking intensity, the pedal opening and the battery state of charge can be introduced into the changing relationship, and the consideration of the battery energy recovery capability is increased based on the introduction of the battery state of charge, thereby realizing the combination of the vehicle data in the deceleration mode with the braking intensity of the oil vehicle and the coasting energy recovery braking intensity; then, the data obtained based on the aforementioned changing relationship can be obtained as the training data for the second model, so that the torque control can ensure that the braking intensity is stable under different vehicle speeds, pedal openings and battery states of charge. The second model can be trained based on the trained second model, wherein the pedal opening, vehicle speed and battery state of charge can be used as state variables, vehicle torque can be used as action variable, and torque change rate can be used as loss function to generate a torque control strategy. When the deceleration mode is turned on, the generated torque control strategy can determine the optimal output torque change rate as the goal, so that the torque change rate is small and the target torque output is continuous and smooth, thereby ensuring that the fluctuation of acceleration and deceleration is small, and realizing smooth transition of switching between the deceleration mode and the gliding energy recovery mode, thereby achieving the effect of no shaking of the vehicle body and good driving experience.

[0104] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0105] Reference Figure 6 , shows a structural block diagram of an embodiment of a torque control device in a deceleration and slowing mode of the present invention, which may specifically include the following modules:

[0106] A first model acquisition module 601 is used to acquire a first model, wherein the first model is used to indicate a relationship between a vehicle torque and a braking intensity and a vehicle speed in a deceleration mode, wherein the braking intensity in the deceleration mode is determined based on a pedal opening and a battery state of charge in the deceleration mode;

[0107] A second model acquisition module 602 is used to acquire training data based on the change relationship, and obtain a second model according to the training data; wherein the second model is obtained by training by taking the pedal opening, vehicle speed and battery state of charge as state variables, taking the vehicle torque as an action variable, and taking the torque change rate as a loss function;

[0108] The torque control strategy generating module 603 is used to generate a torque control strategy according to the second model.

[0109] In one embodiment of the present invention, the first model is constructed based on an energy recovery model, a vehicle dynamics model, a battery model and a motor model, and the second model acquisition module 602 may include the following submodules:

[0110] The training data acquisition submodule is used to acquire pedal opening training data and battery state of charge training data, and obtain braking intensity training data based on the pedal opening training data and the battery state of charge training data through the energy recovery model; the battery state of charge training data is the battery state of charge at the last moment output by the battery model; and obtain vehicle speed training data, and calculate vehicle torque training data based on the braking intensity training data and the vehicle speed training data through the vehicle dynamics model, battery model and motor model.

[0111] In one embodiment of the present invention, the training data acquisition submodule may include the following units:

[0112] A braking intensity output unit is used to determine that the braking intensity training data is 0 when the pedal opening training data is less than a preset first threshold value, if the battery state of charge training data is greater than or equal to a preset second threshold value; and / or, if the battery state of charge at the previous moment is less than the preset second threshold value, obtain an initial braking intensity value, and control the change of the initial braking intensity value to obtain braking intensity training data.

[0113] In one embodiment of the present invention, the braking intensity output unit may include the following subunits:

[0114] A braking intensity initial value acquisition unit is used to acquire a first change relationship; the first change relationship is used to indicate the change relationship between different pedal openings and braking intensity in a deceleration mode; based on the first change relationship, a first braking intensity value calibrated corresponding to the pedal opening training data is acquired, and the first braking intensity value is used as the braking intensity initial value.

[0115] In one embodiment of the present invention, the braking intensity output unit may include the following subunits:

[0116] A braking intensity output subunit is used to obtain a second change relationship when the pedal opening training data is within a preset first value range, and to obtain a second braking intensity value calibrated corresponding to the pedal opening training data from the second change relationship, and to calculate the first braking intensity training data by multiplying the second braking intensity value by a preset first weight, and by multiplying the initial value of the braking intensity by a preset second weight; the first braking intensity training data is used to indicate the braking intensity of a petrol vehicle; wherein the second change relationship is used to indicate the change relationship between different pedal openings and braking intensity of a petrol vehicle.

[0117] In one embodiment of the present invention, the braking intensity output subunit is also used to obtain a third change relationship when the pedal opening training data is within a preset second numerical range, and to obtain a third braking intensity value calibrated corresponding to the pedal opening training data from the third change relationship, and to calculate the second braking intensity training data by multiplying the third braking intensity value by a preset third weight, and the product of the braking intensity initial value by a preset fourth weight; the second braking intensity training data is used to indicate the braking intensity of gliding energy recovery; wherein the third change relationship is used to indicate the change relationship between different pedal openings and braking intensity in the gliding energy recovery mode.

[0118] In one embodiment of the present invention, the loss function is based on the minimum torque change rate as an iteration condition; the second model acquisition module 602 may include the following submodules:

[0119] The second model generation submodule is used to construct a state space using the pedal opening training data, vehicle speed training data and battery state of charge training data as state variables, and use the vehicle torque training data as action variables, and use the minimum torque change rate as an iteration condition; construct an objective function for the second model based on the state space, action variables and the iteration conditions; optimize the objective function to obtain the second model.

[0120] In one embodiment of the present invention, the second model generation submodule may include the following units:

[0121] A second model generation unit is used to optimize the objective function to obtain an optimal action variable and a reward value under a current state variable corresponding to the state space; the reward value is used to indicate a reward value for the loss function;

[0122] Obtain the state variables at the next moment, and store the current state variables corresponding to the state space, the optimal action variables and reward values ​​under the current state variables, and the state variables at the next moment in a preset recall unit; when the storage capacity of the preset recall unit reaches a preset third threshold, reversely update the preset estimated value network using the loss function gradient descent method; according to the preset step size, copy the weight parameters of the preset estimated value network to the preset target value network, and obtain the current number of weight parameter copies; the number of weight parameter copies is used to indicate the number of iterations; when the number of iterations reaches a preset fourth threshold, if the difference between the cumulative evaluation values ​​of adjacent iterations is less than the preset fifth threshold, obtain the second model.

[0123] In an embodiment of the present invention, in the torque control device provided in the embodiment of the present invention, the first model can be used to indicate the changing relationship between the vehicle torque and the braking intensity and vehicle speed in the deceleration mode, and the braking intensity in the deceleration mode can be determined based on the pedal opening and the battery state of charge in the deceleration mode, that is, the changing relationship between the braking intensity and the pedal opening and the battery state of charge can be introduced into the changing relationship, and based on the introduction of the battery state of charge, the consideration of the battery energy recovery capability is increased, thereby realizing the combination of the vehicle data in the deceleration mode with the braking intensity of the oil vehicle and the coasting energy recovery braking intensity; then, the data obtained based on the aforementioned changing relationship can be obtained as the training data of the second model, so that the torque control can ensure that the braking intensity is adjusted under different vehicle speeds, pedal openings and battery state of charge. The pedal opening, vehicle speed and battery state of charge are taken into consideration, wherein specifically, the pedal opening, vehicle speed and battery state of charge can be used as state variables, the vehicle torque can be used as an action variable, and the torque change rate can be used as a loss function to train a second model, so as to generate a torque control strategy based on the trained second model, so that when the deceleration mode is turned on, the generated torque control strategy, under the conditions of different vehicle speeds, pedal openings and battery state of charge, makes the torque change rate smaller on the basis of determining the optimal output torque change rate as the goal, and makes the target torque output continuous and smooth, thereby ensuring that the fluctuation of acceleration and deceleration is small, and realizing a smooth transition between the deceleration mode and the gliding energy recovery mode, thereby achieving the effect of no vehicle body shaking and good driving experience.

[0124] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0125] An embodiment of the present invention also provides a vehicle controller, comprising: a torque control device for the deceleration and slow driving mode, a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, the various processes of the torque control method embodiment for the deceleration and slow driving mode described above are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0126] An embodiment of the present invention further provides a vehicle, comprising:

[0127] It includes a processor, a memory, and a computer program stored in the memory and capable of running on the processor. When the computer program is executed by the processor, the various processes of the torque control method embodiment of the deceleration mode are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0128] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the various processes of the torque control method embodiment of the deceleration mode are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0129] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0130] It will be appreciated by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0131] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0132] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0133] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0134] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0135] Finally, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0136] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0137] The above describes in detail a torque control method for a deceleration mode, a torque control device for a deceleration mode, a corresponding vehicle and a corresponding computer-readable storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A torque control method in a deceleration mode, characterized in that: The method comprises: Acquire a first model, wherein the first model is constructed based on an energy recovery model, a vehicle dynamics model, a battery model, and a motor model; the first model is used to indicate a relationship between a vehicle torque and a braking intensity and a vehicle speed in a deceleration mode, wherein the braking intensity in the deceleration mode is determined based on a pedal opening and a battery state of charge in the deceleration mode; Acquire braking intensity training data and vehicle speed training data, and calculate vehicle torque training data based on the braking intensity training data and the vehicle speed training data through the vehicle dynamics model, battery model and motor model; A second model is obtained according to the vehicle torque training data, wherein the second model is obtained by training based on the pedal opening, vehicle speed and battery state of charge as state variables, the vehicle torque as an action variable, and the torque change rate as a loss function, and the loss function is based on the minimum torque change rate as an iteration condition; A torque control strategy is generated according to the second model.

2. The method according to claim 1, characterized in that The step of obtaining the braking intensity training data comprises: Pedal opening training data and battery state of charge training data are obtained, and braking intensity training data is obtained through the energy recovery model based on the pedal opening training data and the battery state of charge training data; the battery state of charge training data is the battery state of charge at the last moment output by the battery model.

3. The method according to claim 2, characterized in that The obtaining of the braking intensity training data based on the pedal opening training data and the battery state of charge training data comprises: When the pedal opening training data is less than a preset first threshold, if the battery state of charge training data is greater than or equal to a preset second threshold, determining that the braking intensity training data is 0; And / or, when the pedal opening training data is less than a preset first threshold, if the battery state of charge training data is less than the preset second threshold, an initial braking intensity value is obtained, and the initial braking intensity value is controlled to change to obtain braking intensity training data.

4. The method according to claim 3, characterized in that The step of obtaining the initial value of the braking intensity includes: Acquire a first change relationship; the first change relationship is used to indicate the change relationship between different pedal openings and braking strengths in a deceleration and slow driving mode; A first braking intensity value calibrated corresponding to the pedal opening training data is acquired based on the first change relationship, and the first braking intensity value is used as an initial braking intensity value.

5. The method according to claim 4, characterized in that The step of controlling the initial value of the braking intensity to obtain the braking intensity training data includes: If the pedal opening training data is within a preset first value range, a second change relationship is obtained, and from the second change relationship, a second braking intensity value calibrated corresponding to the pedal opening training data is obtained, and the first braking intensity training data is calculated by multiplying the second braking intensity value by a preset first weight, and the product of the braking intensity initial value by a preset second weight; the first braking intensity training data is used to indicate the braking intensity of a petrol vehicle; wherein the second change relationship is used to indicate the change relationship between different pedal openings and braking intensity of a petrol vehicle; And / or, if the pedal opening training data is within a preset second numerical range, a third change relationship is obtained, and from the third change relationship, a third braking intensity value calibrated corresponding to the pedal opening training data is obtained, and the product of the third braking intensity value and a preset third weight, and the product of the braking intensity initial value and a preset fourth weight are used to calculate the second braking intensity training data; the second braking intensity training data is used to indicate the braking intensity of gliding energy recovery; wherein, the third change relationship is used to indicate the change relationship between different pedal openings and braking intensity in the gliding energy recovery mode.

6. The method according to claim 1, characterized in that The obtaining a second model according to the vehicle torque training data comprises: The pedal opening training data, the vehicle speed training data and the battery state of charge training data are used as state variables to construct a state space, and the vehicle torque training data is used as an action variable, and the minimum torque change rate is used as an iteration condition; Constructing an objective function for the second model based on the state space, the action variable, and the iteration condition; The objective function is optimized to obtain the second model.

7. The method according to claim 6, characterized in that The step of optimizing the objective function to obtain the second model includes: Optimizing the objective function to obtain an optimal action variable and a reward value under a current state variable corresponding to the state space; the reward value is used to indicate a reward value for the loss function; Obtaining the state variables at the next moment, and storing the current state variables corresponding to the state space, the optimal action variables and reward values ​​under the current state variables, and the state variables at the next moment in a preset recall unit; When the storage amount of the preset memory unit reaches a preset third threshold, reversely updating the preset estimated value network using a loss function gradient descent method; According to a preset step size, the weight parameters of the preset estimated value network are copied to the preset target value network, and the current number of weight parameter copies is obtained; the number of weight parameter copies is used to indicate the number of iterations; When the number of iterations reaches a preset fourth threshold, if the difference between the cumulative evaluation values ​​of adjacent iterations is less than a preset fifth threshold, a second model is obtained.

8. A torque control device in a deceleration mode, characterized in that: The device comprises: A first model acquisition module is used to obtain a first model based on an energy recovery model, a vehicle dynamics model, a battery model and a motor model, wherein the first model is used to indicate a relationship between a vehicle torque and a braking intensity and a vehicle speed in a deceleration mode, wherein the braking intensity in the deceleration mode is determined based on a pedal opening and a battery state of charge in the deceleration mode; a second model acquisition module, for acquiring vehicle torque training data obtained based on the change relationship, and obtaining a second model according to the vehicle torque training data; the second model acquisition module includes a training data acquisition submodule, the training data acquisition submodule is used to acquire braking intensity training data and vehicle speed training data, and the vehicle torque training data is calculated based on the braking intensity training data and the vehicle speed training data by the vehicle dynamics model, the battery model and the motor model; wherein the second model is obtained by training based on taking the pedal opening, the vehicle speed and the battery state of charge as state variables, taking the vehicle torque as an action variable, and taking the torque change rate as a loss function, and the loss function is based on the minimum torque change rate as an iteration condition; A torque control strategy generating module is used to generate a torque control strategy according to the second model.

9. A vehicle, characterized in that: include: A processor, a memory, and a computer program stored in the memory and capable of running on the processor, wherein when the computer program is executed by the processor, the torque control method of the deceleration and creeping mode as claimed in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the torque control method of the deceleration and creeping mode as described in any one of claims 1 to 7 is implemented.

Images