KR102649360B1 - Wheelslip control method for electric vehicles - Google Patents

Abstract

The present invention relates to a wheel slip control method for electric vehicles. More specifically, the friction coefficient of the road surface is more accurately estimated using machine learning techniques, the upper limit value of the motor torque is calculated through this, and the motor driver is operated according to the calculated upper limit value. This relates to a wheel slip control method for electric vehicles that allows more accurate control of wheel slip of electric vehicles, where wheelslip is more likely to occur on low-friction roads compared to internal combustion engine vehicles.

Description

{Wheelslip control method for electric vehicles}

The present invention relates to a wheel slip control method for electric vehicles. More specifically, the friction coefficient of the road surface is more accurately estimated using machine learning techniques, the upper limit value of the motor torque is calculated through this, and the motor driver is operated according to the calculated upper limit value. This relates to a wheel slip control method for electric vehicles that allows more accurate control of wheel slip of electric vehicles, where wheelslip is more likely to occur on low-friction roads compared to internal combustion engine vehicles.

In general, an electric vehicle refers to an eco-friendly vehicle powered by electricity, that is, electricity collected in the battery provided in the vehicle. Since it does not use fossil fuels, it produces almost no noise and exhaust gas emissions.

Since the 1990s, as problems of environmental pollution and shortage of fossil fuel resources have emerged, competition among automobile companies around the world to develop electric vehicles has become fierce.

In addition, recently, the development of connected cars that enable two-way communication by connecting information and communication technology with cars has been actively conducted. These connected cars are designed to solve vehicle problems caused by bad weather such as heavy snow or heavy rain. This technology was designed to solve this problem and is being used to reduce the incidence of accidents by allowing cars to update and communicate weather information in real time.

Meanwhile, wheelslip refers to a phenomenon in which the wheel spins due to the driving force being excessive than the adhesion between the tire and the road surface. Since electric vehicles can produce greater torque than internal combustion engines, wheel slip is likely to occur on low-friction roads. The disadvantage is that the possibility is higher than that of internal combustion engine vehicles.

In other words, unlike existing internal combustion engines, electric vehicles use a motor instead of an engine and a battery instead of fuel, so they have the advantage of being able to maximize the initial starting torque compared to an internal combustion engine due to the nature of controlling the motor. When starting and accelerating with Open Throttle (Open Throttle), the maximum torque of the motor can be produced, so there is a vulnerability to wheel slip when starting or accelerating the vehicle compared to an internal combustion engine vehicle.

In order to solve these problems and realize safer driving of electric vehicles, the road surface condition or friction coefficient is estimated in real time using information that can be obtained while driving the vehicle, and wheel slip is controlled through this. Technologies or methods that allow this to be done are being developed.

For example, Republic of Korea Patent Publication No. 10-2022-0008952 discloses a vehicle wheel slip reduction control method, the main technical configuration of which is that the controller generates wheel slip on low-friction road surfaces from vehicle driving information collected from the vehicle. It is determined whether the set conditions representing the state are satisfied, and when it is determined that the set conditions are satisfied, the controller accumulates and counts the number of wheel slip occurrences on the low-friction road surface, which is the number of times the set conditions are satisfied, and the counted low-friction road surface When the number of wheel slip occurrences reaches a set number of times, the controller reduces the starting torque of the vehicle. However, conventional wheel slip control methods, including the above-described prior art, are based on the estimation of road surface conditions or road surface friction coefficient. Since the information used is limited, there is a problem of low accuracy.

1. Republic of Korea Patent Publication No. 10-2022-0008952 (published on January 24, 2022)

The present invention was created to solve the problems of the prior art as described above. The purpose of the present invention is to calculate the friction coefficient of a running electric vehicle in real time using a pre-learned deep learning model, and use this to determine the upper limit of motor torque. The aim is to provide a wheel slip control method for electric vehicles that allows the wheel slip of a running electric vehicle to be accurately controlled in real time by calculating .

In addition, the present invention can improve the accuracy of friction coefficient estimation by enabling the friction coefficient to be estimated using various information including weather information, vehicle information, and vehicle weight in dictionary learning using a deep learning model, thereby improving the accuracy of friction coefficient estimation. Another purpose is to provide a wheel slip control method for electric vehicles that can improve the accuracy of wheel slip control.

The present invention to achieve the above objectives,

A data collection step that collects information to be used for estimating the friction coefficient of the road surface, and a deep learning model that can process time series information that uses the information collected in the data collection step as input and defines the output as the road surface friction coefficient. A deep learning learning step of performing repetitive learning, a friction coefficient estimation step of calculating a real-time road surface friction coefficient using the model learned in the deep learning step and information collected in real time in the data collection step, and estimating the friction coefficient It may include a wheel slip control step of controlling wheel slip of the vehicle using the real-time road surface friction coefficient calculated in the step.

At this time, the data collection step includes a weather information acquisition step of acquiring real-time weather information including outside temperature, humidity, and precipitation using connected car technology, and wheel torque and wiper use measured through sensors provided in the vehicle. It may include a vehicle information acquisition step of acquiring driving vehicle information including whether and wheel speed, and a vehicle weight estimation step of calculating the weight of the vehicle using the vehicle information and vehicle specifications obtained in the vehicle information acquisition step. there is.

Here, the weight m of the vehicle in the vehicle weight estimation step is,

It is characterized by being calculated by (here, , , , ego, is the torque input to the transmission when in EV mode. , N _transmission and N _differential are the gear ratios of the transmission and differential, respectively, f _efficiency is the efficiency of the electric vehicle, and r _tire is the radius of the wheels.)

In addition, the wheel slip control step is a motor torque calculation that calculates the upper limit of the motor torque to be supplied to the motor driver using the road surface friction coefficient calculated in the friction coefficient estimation step and the vehicle weight calculated in the vehicle weight estimation step. May include steps.

Additionally, in the wheel slip control step, the motor driver is controlled not to drive beyond the upper limit value of the motor torque calculated in the motor torque calculation step.

According to the present invention, the friction coefficient of a running electric vehicle is calculated in real time using a pre-trained deep learning model, the motor torque upper limit is calculated using this, and the calculated motor torque upper limit is applied to the motor driver to prevent wheel slip of the electric vehicle. It has an excellent effect of accurately and stably controlling.

In addition, according to the present invention, the accuracy of friction coefficient estimation can be improved by using various information including weather information, vehicle information, and vehicle weight in dictionary learning using a deep learning model, thereby improving the accuracy of wheel slip control. It has additional effects that can be used.

1 is a diagram sequentially showing a wheel slip control method for an electric vehicle according to the present invention.
Figure 2 is a diagram conceptually showing a wheel slip control method for an electric vehicle according to the present invention.
Figure 3 is a diagram schematically showing the geometric relationship of the electric vehicle used in the vehicle weight estimation step of the present invention shown in Figure 2.
Figures 4 (a) and (b) are diagrams showing wheel slip control in the wheel slip control stage of the present invention shown in Figure 2, compared with the prior art.

Hereinafter, preferred embodiments of the wheel slip control method for an electric vehicle according to the present invention will be described in detail with reference to the attached drawings.

Figure 1 is a diagram sequentially showing a wheel slip control method for an electric vehicle according to the present invention, Figure 2 is a diagram conceptually showing a wheel slip control method for an electric vehicle according to the present invention, and Figure 3 is a diagram showing the wheel slip control method for an electric vehicle according to the present invention. It is a diagram schematically showing the geometric relationship of the electric vehicle used in the vehicle weight estimation step, and Figures 4 (a) and (b) show wheel slip control in the wheel slip control step of the present invention shown in Figure 2 compared to the prior art. This is the drawing shown.

The present invention uses machine learning techniques to more accurately estimate the friction coefficient of the road surface, calculates the upper limit value of the motor torque through this, and controls the motor driver according to the calculated upper limit value, thereby reducing wheel slip (wheel slip) on low friction roads compared to internal combustion engine vehicles. This relates to a wheel slip control method for electric vehicles that allows more accurate control of wheel slip of an electric vehicle (10), which is likely to cause wheel slip. As shown in Figure 1, the structure is largely comprised of a data collection step (S10), It may include a deep learning learning step (S20), a friction coefficient estimation step (S30), and a wheel slip control step (S40).

First, the data collection step (S10) is a process for collecting information to be used for wheel slip control, and can collect information to be used for deep learning learning, which will be described later, and for estimating the road surface friction coefficient using the learned model.

In more detail, the data collection step (S10) may include a weather information acquisition step (S12), a vehicle information acquisition step (S14), and a vehicle weight estimation step (S16), wherein the weather information acquisition step (S12) As shown in FIG. 2, is a process for acquiring external weather information related to the friction coefficient of the road surface on which the vehicle runs through the weather information acquisition unit 20 provided in the electric vehicle 10, and the weather information includes Temperature (outside temperature), humidity, precipitation, etc. may be included.

At this time, the weather information acquisition unit 20 may use an external temperature sensor, a humidity sensor, etc. provided in the vehicle. However, by using connected car technology, weather information of the surrounding area where the vehicle is driving is collected in real time. It is desirable to make it possible to obtain it.

Next, the vehicle information acquisition step (S14) is a process for acquiring information about the running vehicle, including wheel torque, whether wipers are used, and wheel speed, which are related to the friction coefficient of the road surface, and includes sensors provided in the vehicle. It can be obtained through the vehicle information acquisition unit 30.

In addition, in the vehicle information acquisition step (S14), vehicle information to be used for calculating the vehicle weight in the vehicle weight estimation step (S16) to be described later, that is, the propulsion force (F _x ), air resistance (F _aero ) of the electric vehicle 10, You can obtain vehicle information necessary for calculating rolling resistance (F _rolling ) and grade resistance (F _grade ).

Next, the vehicle weight estimation step (S16) is a process of calculating the weight of the vehicle using the vehicle information and vehicle specifications obtained in the vehicle information acquisition step (S14). The calculated vehicle weight is described later. It can be used for deep learning learning to estimate the road surface friction coefficient and to calculate the upper limit of motor torque using the estimated road surface friction coefficient.

In more detail, in the vehicle weight estimation step (S16), the weight (m) of the running vehicle is calculated by equation (1) below using the geometric structure of the vehicle and the formula for the force acting on the vehicle shown in FIG. 3. It can be calculated.

... (One)

_{Here , F ​} Each is represented and can be calculated using equations (2) to (5) below.

... (2)

... (3) ... (4)

...(5)

At this time, the variables used in equations (2) to (5) are all obtained from the vehicle information obtained in the vehicle information acquisition step (S14), vehicle specifications, and geometric relationships including the slope angle (θ) of the road. This is information that can be used, and the air resistance, rolling resistance, and gradient resistance acting on the vehicle corresponding to equations (3) to (5) correspond to already known calculation equations, so detailed explanations will be omitted.

Additionally, the one used in equation (2) above is refers to the torque input to the transmission (TM), in the case of EV mode applied to the electric vehicle (10) It can be expressed as follows.

And, N _transmission and N _differential mean the gear ratio of the transmission and differential, respectively, f _efficiency means the efficiency of the electric vehicle 10, and r _tire is the radius of the wheel.

To calculate the vehicle weight as described above, a vehicle weight estimation unit 40 programmed with calculation equations including equations (1) to (5) may be provided in the electric vehicle 10.

Next, the deep learning learning step (S20) is a process of performing prior learning to estimate the road surface friction coefficient using the information collected in the data collection step (S10). The deep learning learning unit ( 50).

The deep learning learning unit 50 for prior learning as described above inputs weather information including outside temperature (air temperature), humidity, and precipitation, vehicle information including wheel torque, whether wipers are used, and wheel speed, and vehicle weight. A deep learning model that can process time series information whose output is defined as the road surface friction coefficient can be used.

That is, in the deep learning learning step (S20), real-time information provided from a running vehicle through repeated learning of a deep learning model that allows calculating the road surface friction coefficient using the information collected in the data collection step (S10). It is possible to estimate the road surface friction coefficient, and by diversifying the information input to the deep learning model, the accuracy of road surface friction coefficient estimation can be improved.

Next, the friction coefficient estimation step (S30) calculates the friction coefficient of the road surface on which the vehicle runs in real time using the information collected in real time while the vehicle is driving, that is, the information collected in the data collection step (S10). As a process, the road surface friction coefficient can be calculated in real time through a deep learning model pre-trained through the deep learning learning step (S20).

At this time, the electric vehicle 10 may be equipped with a database (not shown) that stores information collected in real time while the vehicle is driving, and calculates the road surface friction coefficient by updating the deep learning model using the information stored in the database. accuracy can be continuously improved.

Next, the wheel folding control step (S40) is a process for controlling the wheel folding of a running vehicle using the real-time road surface friction coefficient calculated in the friction coefficient estimation step (S30), which involves driving the motor 16 of the vehicle. Wheel slip can be controlled by controlling the motor torque supplied to the motor driver 14.

That is, the wheel slip control step (S40) may include a motor torque calculation step (S42) and a motor torque control step (S44). First, the motor torque calculation step (S42) is performed in the friction coefficient estimation step (S30). This is a process for calculating the upper limit value of the motor torque to be supplied to the motor driver 14 using the calculated real-time road surface friction coefficient and the weight of the vehicle calculated in the vehicle weight estimation step (S16), and the motor torque to be supplied to the electric vehicle 10 A calculation unit 60 may be provided to calculate the upper limit value of .

At this time, since the method of calculating the upper limit value of the motor torque using the road surface friction coefficient and vehicle weight performed in the calculation unit 60 is already known, detailed description thereof will be omitted.

Next, the motor torque control step (S44) is a process of controlling the motor torque supplied to the motor driver 14 so as not to exceed the upper limit of the motor torque calculated in the motor torque calculation step (S42), through which the motor driver 14 ) can be controlled so that it is not driven beyond the upper limit of the calculated motor torque.

In more detail, as shown in (a) of FIG. 4, conventionally, the amount of motor torque required by the motor driver 14 is generated through the target torque generator 12 and supplied to the motor driver 14, The motor driver 14 controls the driving of the electric vehicle 10 by supplying an amount of current corresponding to the supplied torque to the motor 16 to drive the vehicle.

In contrast, in the present invention, as shown in (b) of FIG. 4, a motor torque limiter 70 may be provided between the target torque generator 12 and the motor driver 14, and the motor torque limiter (70) may serve to control the motor torque supplied to the motor driver 14 so that it does not exceed the upper limit value of the motor torque calculated in the motor torque calculation step (S42).

In other words, the occurrence of wheel slip while driving the vehicle depends on the friction coefficient of the road surface on which the vehicle is driven, so the road surface friction coefficient is estimated in real time through a deep learning model previously learned using information acquired while the vehicle is driving. After calculating the upper limit value of the motor torque that can be supplied to the motor driver 14 according to the estimated road surface friction coefficient, the motor torque supplied to the motor driver 14 is controlled so as not to exceed the calculated upper limit value, thereby It can be accurately controlled to prevent slip.

Therefore, according to the wheel slip control method for an electric vehicle according to the present invention as described above, the friction coefficient of the running electric vehicle 10 is calculated in real time using a pre-learned deep learning model, and the upper limit value of the motor torque is determined using this. Not only can the wheel slip of the electric vehicle (10) be accurately and stably controlled by applying the calculated motor torque upper limit to the motor driver (14), but also weather information, vehicle information, and It has various advantages, such as being able to improve the accuracy of friction coefficient estimation by using various information including the vehicle weight, and thus improving the accuracy of wheel slip control.

Although the above-described embodiments describe the most preferred examples of the present invention, it is not limited to the above embodiments, and it is clear to those skilled in the art that various modifications are possible without departing from the technical spirit of the present invention.

The present invention relates to a wheel slip control method for electric vehicles. More specifically, the friction coefficient of the road surface is more accurately estimated using machine learning techniques, the upper limit value of the motor torque is calculated through this, and the motor driver is operated according to the calculated upper limit value. This relates to a wheel slip control method for electric vehicles that allows more accurate control of wheel slip of electric vehicles, where wheelslip is more likely to occur on low-friction roads compared to internal combustion engine vehicles.

10: Electric vehicle 12: Torque generator
14: motor driver 16: motor
20: Weather information acquisition unit 30: Vehicle information acquisition unit
40: Vehicle weight estimation unit 50: Deep learning learning unit
60: calculation unit 70: motor torque limiter
S10: Data collection step S12: Weather information acquisition step
S14: Vehicle information acquisition step S16: Vehicle weight estimation step
S20: Deep learning learning step S30: Friction coefficient estimation step
S40: Wheel slip control step S42: Motor torque calculation step
S44: Motor torque control stage

Claims (5)

A data collection step of collecting information to be used to estimate the friction coefficient of the road surface,
A deep learning learning step that uses the information collected in the data collection step as input and performs iterative learning using a deep learning model capable of processing time series information defined as the road surface friction coefficient as the output;
A friction coefficient estimation step of calculating a real-time road surface friction coefficient using the model learned in the deep learning learning step and the information collected in real time in the data collection step;
A wheel slip control step of controlling the wheel slip of the vehicle using the real-time road surface friction coefficient calculated in the friction coefficient estimation step,
The data collection step is,
A weather information acquisition step of acquiring real-time weather information including outdoor temperature, humidity, and precipitation using connected car technology;
A vehicle information acquisition step of acquiring driving vehicle information including wheel torque, wiper use, and wheel speed measured through sensors installed in the vehicle, and
A vehicle weight estimation step of calculating the weight of the vehicle using the vehicle information and vehicle specifications obtained in the vehicle information acquisition step,
The wheel slip control step is,
Wheel slip control for an electric vehicle including a motor torque calculation step of calculating an upper limit value of the motor torque to be supplied to the motor driver using the road surface friction coefficient calculated in the friction coefficient estimation step and the vehicle weight calculated in the vehicle weight estimation step. method.
delete According to clause 1,
In the vehicle weight estimation step, the weight m of the vehicle is,
A wheel slip control method for an electric vehicle, characterized in that it is calculated by.
(here, , , , ego, is the torque input to the transmission when in EV mode. , N _transmission and N _differential are the gear ratios of the transmission and differential, respectively, f _efficiency is the efficiency of the electric vehicle, and r _tire is the radius of the wheels.)
delete According to clause 1,
In the wheel slip control step,
A wheel slip control method for an electric vehicle, characterized in that the motor driver is controlled not to drive beyond the upper limit of the motor torque calculated in the motor torque calculation step.