Tesla drive motor control system

Energy regeneration is a very core technology for electric vehicles - vehicle deceleration causes the vehicle's drive motor to act as a generator to charge the battery. Charging is essential to improve range.

Regeneration requires deceleration, which generally requires depressing the brake pedal. But if regeneration is controlled solely by the brake pedal, it will be inefficient. Because the frequency of use of the brake pedal is very low compared to the accelerator pedal, the energy recovery brought about by stepping on the brake pedal is far from being able to achieve the purpose of extending the cruising range. Deliberately depressing the brake pedal to recover energy would appear unnatural, and the driver would easily tire and create danger for road traffic.

Therefore, Tesla's drive motor system was designed in the form of one pedal driving (OPD, one pedal driving). The so-called single-pedal drive refers to controlling acceleration and energy regeneration at the same time by changing the opening of the accelerator pedal.

This color diagram clearly shows how Tesla's one-pedal drive system works. There are roughly four situations here:

• When the accelerator pedal is pressed to the floor (so-called floor oil), the driver will receive 100% acceleration

• As the accelerator pedal is gradually released, the acceleration obtained by the driver will become smaller.

• When the accelerator pedal is further released, there is a balance point between acceleration and deceleration, which is the "best position" for the vehicle to "coast".

• When the accelerator pedal is fully released, the driver will achieve maximum deceleration and maximum battery charge levels for "maximum" regeneration.

The maximum regenerative deceleration value is about 0.2g for older Teslas and about 0.3g for newer Teslas.

The deceleration level of 0.3g makes people feel like decelerating in the first gear of a diesel locomotive with a gearshift lever.

The characteristics of single-pedal driving make driving a Tesla completely different from driving a traditional fuel vehicle - as long as you keep your foot on the accelerator pedal, the torque can be between the maximum positive torque and a certain level of negative braking torque. Continuous adjustment. As a result, the brake pedal will only be used in two situations:

• bring the car to a complete stop

• Emergency braking at levels above 0.3g

The picture above shows the first step of the Tesla drive motor control system - through this calibrated pedal map, the accelerator pedal position set by the driver is converted into the requested motor torque value. In the blank area, the torque value will be obtained by interpolation. The pedal map in the system includes a two-dimensional lookup table that can be accessed using the accelerator pedal position signal.

The accelerator pedal position signal is an 8-bit signed integer obtained through the dual accelerator pedal position sensor.

The torque value is an 8-bit signed integer in Tesla Model 3, and a 16-bit signed integer in Tesla Model S and Model X. Torque values ​​are generated every 100 milliseconds.

The above figure shows the second step performed by the Tesla drive motor control system: In this step, the torque Map of the drive motor (i.e., torque-speed TN map) is given, which changes the requested torque value from The pedal map is converted into torque and "magnetic field flux command" to drive the motor. The map also includes a two-dimensional lookup table and is accessed via the pedal map torque map and vehicle speed. Areas without numerical values ​​can be obtained by two-dimensional interpolation.

• Forward motor torque commands in the "positive quadrant" of acceleration allow the vehicle to draw power from the battery while moving forward. At 100% motor torque, the magnitude of acceleration may vary from 0g to 1g. The dashed line in the green area shows a typical acceleration curve in the acceleration quadrant. More generally, it can be any compound curve.

  
  

• Negative motor torque commands in the braking "negative quadrant" enable regenerative braking by charging the drive battery. This is restricted by many factors. Only the red area will be used, and the red area outside the negative quadrant will not be used. The dotted line in the red portion of the diagram shows a typical brake regeneration path in the braking quadrant.

  
  

Zooming in on the red part of the torque-speed diagram, we get the graph above. There are several limiting factors here:

• Taking into account the sudden intervention of energy regeneration when the road friction coefficient decreases, in order to avoid vehicle instability, the maximum braking torque will be limited by the maximum deceleration level of -0.2g~-0.3g. (The straight line formed by the lower left square)

• Considering the existence of back electromotive force inside the motor (the size of the back electromotive force is proportional to the speed, its increase will reverse the bus voltage, and when the bus current is cut off, the motor will reach the speed limit), the motor is constrained by the maximum speed , the tail is truncated. In order to further increase the maximum speed, technology such as field weakening control can be used to sacrifice torque in exchange for high speed. (speed cutoff on the right)

• Due to the limitation of motor power, the product of torque and speed is constrained, and the power remains constant in the field weakening control area. (Curve formed by the squares on the right)

• At any given torque below the maximum braking torque, regenerative power and braking torque decrease in proportion to the motor speed. This is limited by the maximum regenerative power - which must remain within certain maximum values ​​determined by the battery charging circuit. (a straight line formed by dots)

• If the maximum regenerative power is optimized it is no longer an issue, it will eventually be limited by the electromotive force emf - staying above some minimum value for the battery to be effectively charged. (a straight line formed by a triangle)

Below the minimum emf voltage, braking torque is only possible by drawing power from the battery, which is called plug braking in the technical literature.

A crucial point to remember here is that the motor is still able to operate in all areas of the braking quadrant - but regenerative braking can only be achieved within a limited area (the red zone) by limiting the torque request at a given speed. Within the regenerative braking region described by many of the above limitations, any desired braking torque and regenerative power can be obtained by providing appropriate torque and field flux commands to the drive motor.

Consider another limitation: when the vehicle speed (proportional to the motor speed) decreases, such as below 5mph, the AC induction asynchronous motor will not be able to produce any effective braking torque or regenerative power. This is because the magnetic field of the rotor is limited by the induced current, which is too low for practical use and cannot form a strong enough magnetic field.

But for permanent magnet synchronous motors containing permanent magnets, at low speeds, the motor can still generate a large enough rotor magnetic field, allowing it to operate below 5mph. Tesla introduced this permanent magnet synchronous (PMSM) motor for the first time in their Model 3, replacing the previous AC induction asynchronous motor (ACIM), and plans to use it in all new versions of Model S and Model X production .

The graph above shows the operating points used in the drive motor torque diagram when following a typical urban driving cycle. The operating point is shown in red, allowing positive acceleration and negative braking torque to be clearly seen. Especially at low speeds in the lower left corner, the linear change of braking torque with vehicle speed is obvious.

As shown in the figure above, the driving cycle used here is based on the U.S. EPA's standard Urban Dynamometer Driving Schedule (UDDS).

As we all know, many Tesla cars have dual drive motors.

As shown in the figure above, the behavior of a Model 3 equipped with dual drive motors when decelerating from 100 kph to 0 kph is drawn when regenerative braking is involved. When vehicle speed is reduced, all regenerative torque and power is provided only by the rear drive motor, resulting in the greatest possible regeneration. For Model 3 with software update 2018.42 v9, the maximum braking torque corresponds to -0.3g acceleration.

As expected, at low speeds, regenerative torque and power decrease linearly with rpm.

Unlike the previous image, this new image simulates the slippage between the vehicle's tires and the road when there is snow on the road. When slip occurs, you will find that the braking torque of the rear motor is transferred to the front motor, and the sum of the two torques remains consistent near the original maximum value.

Why is it designed this way?

The reason for this behavior is easy to understand - regenerative torque is a braking operation that occurs when braking torque is applied to the rear wheels due to a lack of traction due to slippage due to snow, ice, rain or gravel. Hug to death. Without rear wheel traction, the vehicle could become unstable and enter dangerous uncontrollable spins around its vertical axis (z-axis). Therefore, vehicle manufacturers are required by government regulations to use a certain amount of front wheel braking to prevent this instability from occurring.

When any given vehicle is braked, deceleration causes the vehicle to lean forward and shift the center of gravity forward, changing the level of braking force on the wheels - as weight is removed from the rear wheels, traction is also reduced and when applied When braking, they tend to lock up faster at lower deceleration values. Of course, transferring too much weight to the front wheel can also cause the front wheel to lock.

As a function of the deceleration value, vehicle mass, center of mass position, wheelbase and road friction coefficient, the horizontal braking force on the front and rear wheels when the front and rear wheels are locked simultaneously can be calculated. If we plot these values ​​as a curve on a plane, with the front and rear braking forces as the orthogonal axes, we get the ideal I curve shown in the figure above.

• Above curve I, the rear wheels will lock first, which is an unsafe condition expressly prohibited by government regulations.

• Below curve I, any operating point is considered safe except the point under the M curve which defines the vehicle's minimum rear braking force.

A series of slashes shows the deceleration value. The beta-curve shows the operating point associated with the linear brake proportioning valve used in most ICE vehicles on the road today.

Using this diagram, we can explain the regenerative braking behavior of Tesla Model 3. Corresponding to point A on the figure, it is in a state of deceleration value 0.3g. All braking here is provided by the rear wheel. This is the dangerous area where the rear wheel locks first when skidding is expressly prohibited by regulations - when the rear wheel is not slipping, This is allowed when there is 100% traction on the road.

If any slip occurs, in order to keep the vehicle stable at the same deceleration value, the rear wheel braking force must be reduced while increasing the front wheel braking force to move the operating point to point B on the I curve in the figure (or below ). At point B, the front and rear braking forces are roughly equal, 0.2g and 0.1g respectively.

The next question is, how does the traction control system of Tesla Model 3 achieve brake regeneration distribution in the event of slippage?

Let’s look at the principle of traction control module. We know that the vehicle torque command generation function includes the accelerator pedal map shown previously, which converts the accelerator pedal value into the total vehicle torque request.

The optimal torque distribution function consists of two drive motor torque-speed maps that convert the total vehicle torque request into two pairs (one for the front and one for the front) of the motor torque and flux commands. Optimal torque distribution operates at the motor's most efficient operating point to conserve battery power.

For certain operating points, the use of only one motor certainly provides acceptable vehicle stability without wheel slip. When wheel slip occurs, the traction control system redistributes the two torques from the optimal torque distribution function into two new front and rear torques to work in or below the I-curve to meet the required vehicle stability.

But there are two points to note here:

• First, providing front-to-rear torque redistribution for traction control under acceleration and regenerative deceleration does not serve other purposes - such as providing differential (i.e. lateral) wheel slip control for oversteer or understeer - which are more general The wheel slip control function is usually provided by the vehicle's braking system.

• Second, the vehicle (motor) torque command generation function has input from the Vehicle Stability Control System (VSC).

The second point seems a bit unexpected - this gives the VSC system the ability to control the vehicle's motor torque in addition to the accelerator pedal. This unique design is also included in Tesla's patent (this is not a mistake, but a unique structure). This is crucial, and we will discuss it in detail in the next section on Tesla’s braking system.

When wheel slip occurs in the front or rear wheels, the torque on the slipping axle is reduced and transferred to the other axle with less wheel slip. This logic can be expressed as a mathematical equation as:

These two equations are abstracted from the following Block Diagram:

This Block Diagram appears in Tesla's patent. It shows the summation operation everywhere, but it is obviously incorrect because the two inputs are in different units (apples plus oranges do not produce more oranges).

These four summation operations should be understood as modulation operations (i.e. multiplication operations), where the torque T is pulse width modulated (multiplication operation) as (1-δ)*T, and δ can be understood as the duty cycle, between 0 and 1 between, determined by the PID controller, which drives the current slip rate to the target slip rate:

• For a straight drive path, the target slip rate is zero

• For curved drive paths, the target slip rate is some normal minimum slip rate

The minimum slip rate can vary with vehicle speed and steering angle. Even without slipping, wheel speed often changes during a turn. Slip rate is calculated by dividing the difference between wheel speed and vehicle speed by the greater of the two. The input to the PID controller is the slip error, which is the difference between the current slip and the target slip, obtained from a lookup table. The PID controller drives this slip error to zero, and when wheel slip is present, the controller drives the torque from some reduced value (1-δ)*T to the full value T when wheel slip is not present.

When there is no slip on both axes, the δ of the front and rear axles is equal to 0:

As expected, this shows that the output torques C_Torque1 and C_torque2 are only the same as the input torques C_Torque1e and C_Torque2e.

If one axis has maximum slip, δ=1, and the other axis has no slip, δ=0, we get:

in this case:

• The non-slip shaft output torque C_torque2 increases from C_torque 2e to C_torque 2e+K2·C_Torque1e, which can be regarded as the torque transmitted from axle 1 to axle 2.

• The output torque C_torque1 of the slipping shaft is reduced from C_torque1e to the torque K1·C_torque1e, where the slip is reduced but the dynamically enhanced part of K1·C_torque1e is provided.

Next, use a high-pass filter and a second PID-based feedback controller to independently suppress rapid motor speed disturbances - when the wheel slips excessively, the load torque on the motor shaft suddenly decreases significantly, one or two A sudden and large increase in load torque on the motor shaft due to a stuck wheel may cause rapid disturbances in the motor speed.

In addition, between the first and second stages, there is a transient torque enhancement feedforward control circuit - that is, dynamic enhancement , which adds a certain amount of torque to each axis. The amount of increased torque is proportional to the difference between the driver's torque request and the combined torque command C_torque after the first stage of traction control.

The proportional constants K1<0 and K2<0 can be adjusted to different values ​​for the two axes. Feedforward torque improves vehicle performance, response to driver demands and drivability without affecting traction control or vehicle stability. When the torque requirement is fully met, the feedforward torque is zero, the effective wheel slip error is zero, and the maximum torque limit is inactive.

When wheels slip, causing torque to be reduced on one axle, feedforward control works by adding torque to the other axle that has better tire-to-road grip. Of course, feedforward control also has the side effect of adding torque to the axle where wheel slip occurs, but due to the relatively small gain in the feedforward path, the wheel slip error feedback loop still dominates, so the wheel slip error remains become smaller.

After traction control, the last step is to issue torque commands based on C_MaxTorque1 and C_MaxTorque2. This process ensures that the regenerative portion of the motor's retorque diagram operates (red portion) when torque is negative. What is output here are the torque commands C_torque1 and C_torque2.

Again, traction control is only available here in response to wheel slip that can be minimized longitudinally through front-to-rear torque redistribution. It doesn't offer stability control based on lateral wheel slip - these slips will be addressed by redistributing torque from left to right, as in oversteer and understeer . These additional stability control functions must be provided by the vehicle's braking system.

Before discussing Tesla's braking system, one additional exception must be mentioned: For some Tesla vehicles, including the Model 3, there is only one rear-drive motor. In this case, regenerative torque cannot be transferred from the rear drive wheels to the front drive wheels because there is no front drive motor.

Author: Ronald A. Belt

Translation: Xu Honghu

"Unlocking Tesla's Black Box" Model 3 Unexpected Acceleration Accident Investigation Report

Reply "EDR" in the background of the public account to obtain the original text of "Tesla Sudden Acceleration Investigation Report"