METHOD FOR CONTROLLING AN ELECTRIC MACHINE TORQUE IN A DRIVE CHAIN ​​OF AN ELECTRIC VEHICLE

Abstract

The invention relates to a method for controlling an electric machine torque in a powertrain of an electric vehicle, using a control loop servocontrol, the powertrain comprising the electric machine engaged with a differential, right and left wheel drive shafts and right and left wheels, the control loop comprising:- an acquisition of a desired machine torque,- an anticipatory filtering, acting on the desired machine torque- a feedback loop, the control loop delivering a corrected machine torque, resulting from the anticipatory filtering and the feedback loop, characterized in that the anticipatory filtering uses a flatness property. Figure 3

Description

Title of the invention: METHOD FOR CONTROLLING AN ELECTRIC MACHINE TORQUE IN A DRIVE CHAIN ​​OF AN ELECTRIC VEHICLE

[0001] The invention relates to a method for controlling an electric machine torque in an electric powertrain of a motor vehicle.

[0002] We are interested here in a powertrain in a hybrid vehicle or in a 100% electric vehicle. This powertrain is installed, in the case of interest, on the front or rear axle of the vehicle which includes steered wheels, in addition to being driven.

[0003] As known per se, the drive train comprises an electric machine, which can operate as a motor or as a generator, depending on the circumstances of use. The mechanical output of the electric machine (shaft with pinion) is engaged, if necessary via a reducer, with a differential. At the output of the differential, a left wheel shaft and a right wheel shaft are arranged on the one hand, each being connected to a respective wheel hub. The tire mounted on each wheel rim completes the drive train.

[0004] In the case of a train with steered wheels, the wheel shafts comprise a cardan-type joint, also called in practice a homokinetic joint, to allow angular movement of the steered wheels for the steering function.

[0005] All along the transmission line between the electric machine and the wheels, it is noted that there are mechanical clearances and that, moreover, certain elements have a certain elasticity; the drive train is thus the site of undesirable oscillations in the rotation kinematics, whether during changes in torque setpoint at the level of the electric machine or during the feedback of disturbances which occur on the road traveled by the vehicle wheels (e.g. potholes, cobblestones, and imperfections in the road).

[0006] In order to minimize the phenomenon, it is known to use a servo control law with a feedback loop which, from a real-time measurement of the rotation speed of the electric motor, makes it possible to reduce the amplitude of these oscillations so that the perception, from the point of view of the driver or the occupants of the vehicle, is minimized.

[0007] It is also known to apply a filtering of the torque setpoint. This can however introduce a delay time and / or a tracking deviation compared to the driver's request.

[0008] The inventors therefore sought to propose new solutions in order to reduce drastically reduce oscillations in the drive train while having responsive control that is faithful to the desired torque setting.

[0009] To this end, the present invention proposes a method for controlling an electric machine torque in a powertrain of an electric vehicle, using a control loop servocontrol, the powertrain comprising the electric machine engaged with a differential, right and left wheel drive shafts and right and left wheels, the control loop comprising: - an acquisition of a desired machine torque, - anticipatory filtering, acting on the desired machine torque - a feedback loop, the control loop delivering a corrected machine torque, resulting from the anticipatory filtering and the feedback loop, characterized in that the anticipatory filtering uses a flatness property.

[0010] The flatness property refers to the use of particular state variables representing a generally non-linear complex system.

[0011] More precisely, a system is generally defined using a vector of state variables noted x, and a vector of control variables noted u, the output of the system being formed of quantities in the form of a vector noted z.

[0012] The behavioral equation of the system is written in a general way:

[0013] x = f ( x. u ) and z = / z(x,u)

[0014] In these formulas, the functions f and h are multidimensional functions.

[0015] The system meets the flatness criteria if it is possible to identify a number m of so-called flat variables written as follows y = X (x,u,*X ü, • • • ) with an expression of the state vectors x, input / control u and output z as a function of the flat variables y and their time derivatives as follows:

[0016] x = <Dx(y,X 5- ■••)

[0017] u = <Du(y,X ÿ, ...)

[0018] z = <Dz(y,M Ÿ, ...)

[0019] In matrix terms, if we write -t ​​= Ax + Bm, the flat variables being written y = Cx, the flatness property exists if the following conditions are met on the following matrix products: C*B = 0, C*A*B = 0 and C*A*A*B 0 for a system of order 2.

[0020] The reader may refer to the following reference article: FLATNESS AND DEFECT OF NONLINEAR SYSTEMS: INTRODUCTORY THEORY AND EXAMPLES, by authors Michel Fliess, Jean Lévine, Philippe Martin and Pierre Rouchon, CAS internai report A-284, January 1994.

[0021] Advantageously, the use of a high-performance anticipatory filter, based on a modeling as defined above, makes it possible to use a lighter feedback loop than in the known art, that is to say with a little less gain and more stability.

[0022] The use of flat variables corresponds to a change of state variables towards a linear and canonical system which is easier to control via conventional techniques.

[0023] Expressed in another way, knowledge of the behavioral model of the transmission line makes it possible to construct an anticipatory term that is very favorable to torque control, without having to wait for the feedback given by the feedback loop.

[0024] More precisely, the closed loop is robustified via the performance contribution of the anticipation action. A structural self-centering of the torque correction is guaranteed, not modifying the driver's wishes in stabilized mode, and a simplification of the development of the anti-oscillation function thanks to the physical approach.

[0025] It is noted that the desired machine torque can come from a press on the accelerator pedal made by the driver, the angle of the pedal being transformed into torque via the use of a calibration as known per se. The machine torque can also come from a speed regulation function, without the driver pressing the accelerator pedal, then 0 pedal remains at 0.

[0026] It should be noted that the expression “transmission line” encompasses all of the elements involved in the transmission of torque between the electric machine and the wheels.

[0027] According to one embodiment, the traction chain exhibits kinematic behavior that the method models by: - a moment of inertia Jp of the rotating parts upstream of the differential, - a moment of inertia Jr of the rotating parts downstream of the differential, - a torsional stiffness Kr, characterizing at least the torsional stiffness of the wheel drive shafts, and more generally the transmission line, - a viscous friction coefficient Kv of the transmission line.

[0028] As will be seen later, the stiffness and the viscous friction coefficient will be used in the calculation of the torsional torque Ct.

[0029] We can write Ct — Kr Ot + Kv , with œt torsion speed and 0t torsion angle.

[0030] We can also write:

[0031] Jp C p = Cp Ch Cp being the primary torque at the output of the electric machine, and cop being the primary rotation speed at the output of the electric machine, and

[0032] Jr r — Cr + Ch Cr being the torque at the wheel, and cor being the rotational speed at the wheel.

[0033] According to one embodiment, the basic state variables of the system œt and Ct are defined as follows: œt: torsional velocity, time derivative of 0t which is the torsion angle Ct: torsion torque, the torsion torque Ct can be written Ct = Kr 0t + Kv œt.

[0034] Using these state variables, we can write the equations governing the behavioral mechanics of the powertrain.

[0035] According to an advantageous embodiment, a flat variable is defined as follows: y = Ct - Kv œt.

[0036] The use of this variable makes it possible to express the system in a linear and canonical form as will be seen below.

[0037] According to one embodiment, the anticipatory filtering delivers a corrected setpoint C'pU expressed as a linear combination of the flat variable, and its first and second derivatives: ^Ant _ Jp / ... Kv Kr \ p ~ k,. + j. y+ jt y /

[0038] with Jt = (Jp Jr) / (Jp + Jr)

[0039] As a result, said correction introduced by the anticipatory filtering thus proves to be very easy to calculate.

[0040] According to one embodiment, the base and output state variables are expressed as a function of the flat variable, and its first and second derivatives:

[0041] w, = Jv ' Ar -• Ct — y + y £p- Kt. ( ?+ j, y+ jt y + j2 Cf )

[0042] According to one embodiment, to construct a torsion torque trajectory, a state variable filter is used by giving a desired damping and pulsation aft, with the formula Ct+2f(A%Ct+(P^ the notation * characterizing the desired behavior, and the notation denoting the value w in the steady state (oscillations canceled).

[0043] This is an easy to implement 2nd order filter.

[0044] According to one embodiment, the anticipatory filtering delivers a corrected instruction CpnT expressed as a linear combination of the flat variable corresponding to the desired trajectory, and its first and second derivatives —

[0045] According to one embodiment, the feedback loop takes the form of a proportional regulator expressed from the derivative of the flat variable as follows: Cpel = - y) 'the output value C^1 being added to the corrected setpoint Cp”{.

[0046] The corrective value of the feedback loop is thus easily calculated from the first derivative of the flat variable.

[0047] The invention further relates to a motor vehicle comprising a powertrain comprising an electric machine engaged with a differential, drive shafts right and left wheel transmission and right and left wheels, the drive train being controlled at least in part by a control unit configured to implement the method as described above.

[0048] The invention will be further detailed by the description of non-limiting embodiments, and on the basis of the appended figures illustrating variants of the invention, in which: [Fig.l] schematically illustrates an electric powertrain of a motor vehicle in which the present invention is implemented; [Fig.2] shows a schematic diagram of the torsional modeling of the traction chain; [Fig.3] shows a flowchart illustrating an example of a control loop applied within the framework of the present invention.

[0049] In the various figures, the same references designate identical or similar elements. For reasons of clarity of the description, certain elements are not necessarily represented to scale.

[0050] In [Fig.l], the traction chain comprises an electric machine marked 2, a reducer marked 3, a differential marked 4.

[0051] At the output of the differential, the drive train comprises two wheel transmission shafts 5, each being connected to a wheel 59. It should be noted that only one transmission shaft 5 and one wheel 59 are shown in [Fig.l].

[0052] In the illustrated example, we have shown a front axle equipped with steered wheels. However, the drive train could be installed on an axle with non-steered wheels. The suspension system allows for vertical wheel travel, in particular, and angular travel in the case of steered wheels.

[0053] Thus, each wheel transmission shaft 5 comprises a cardan joint, also called a constant velocity joint in the art. Due to its design and construction, the cardan joint introduces functional play as well as torsional elasticity, which proves to be the main source of oscillations, the amplitudes and duration of which the present invention will attempt to reduce.

[0054] In the example illustrated, the vehicle comprises an accelerator pedal 8. A pedal position sensor delivers accelerator pedal angle information, noted Oaccel, to give a desired machine torque Cp.

[0055] A calibration table 18 is provided to translate the pedal angle Oaccel into the desired machine torque Cp.

[0056] A speed regulation function 24 is also provided, either conventional CC (for Cruise Control) or adaptive ACC (for Adaptive Cruise Control). The regulation function makes it possible to provide a desired machine torque Cp without the driver pressing the pedal.

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072] The drive train / The system includes a control unit rated 10 which receives the desired machine torque Cp. The control unit 10 includes parameters 25 which relate to the modeling of the torsional behavior of the transmission line as will be seen later. Furthermore, the system includes a wheel speed information socket 19. It should be noted that, in certain cases, this information socket may be part of the vehicle's ABS braking control system. cor is a real-time rotation speed, delivered at most every 10 ms. As schematically illustrated in [Fig.2], the torsional behavior of the traction chain is modeled by: - a moment of inertia Jp of the rotating parts upstream of the differential, - a moment of inertia Jr of the rotating parts downstream of the differential, - a torsional stiffness Kr, characterizing at least the torsional stiffness of the wheel transmission shafts, - a viscous friction coefficient Kv of the transmission line. Furthermore, the basic state variables of the system œt and Ct are defined as follows: œt: torsional velocity, time derivative of 0t which is the torsion angle Ct: torsion torque, the torsion torque Ct can be written Ct = Kr 0t + Kv œt. Using these basic state variables, the system dynamics is reduced to a second-order system. The twisting speed can be expressed as a function of the primary rotational speed cüp and the wheel rotational speed cor as follows (Ot =(Op — (Or. If we take into account the torsion torque Ct, by applying Newton's second law we obtain: Jp 6? p — Cp Ct Jr™> = Cr + C, The dynamic equations of the state variables are written as follows. . G cP c,. œt ~ " jt + jp ' jr Ct = Krœr^Ct+^Cp-^Cr where Jt is an aggregate moment of inertia: Jt = (Jp Jr) / (Jp + Jr) A flat variable is defined as follows: y = Ct - Kv œt. It turns out that this variable has the flatness properties stated in the introductory part of this presentation.

[0073] The base and output state variables are expressed as a function of the flat variable, and its first and second derivatives:

[0074] = 1 d, 1 A ; • -7 Z-* u - C^y+cy c p = t; (>' + ~y+~y+— Cr)

[0075] The anticipatory filtering delivers a corrected instruction C^îf expressed as a linear combination of the flat variable, and its first and second derivatives: f-<Ant _ Jp / K. Kr \ + —> +77? J

[0076] We use a state variable filter by giving ourselves a desired damping g* and pulsation iv„, with the formula the notation * characterizing the desired behavior, and the notation w denoting the value w in the steady state (oscillations canceled).

[0077] Considering the equation (y = Ct - Kv œt) in the steady state ü't — 0, and knowing the setpoint driver torque, we deduced from the expression of Cp as a function of the flat variable and its derivatives in the steady state (* zero) the stabilized torsion torque: ç* _ Af1*.

[0078] To construct a torque trajectory, a state variable filter is used, giving a desired damping and pulsation. It follows:

[0079]

[0080] In Figure 3, starting from the left, we begin by deducing the desired torsion torque in steady state from the desired primary torque Cp- This is represented by the box marked 11.

[0081] From the desired torsion torque c* in steady state, by means of the second-order filter defined above based on the desired damping and pulsation values ​​oA, the box marked 12 calculates the desired torsion torque C*, its first derivative q and its second derivative

[0082] The flat setpoint output and its derivatives can be obtained by filtering the torque setpoint C* and its derivatives with a low-pass filter having a pole at _ . It is the box marked 13 in [Fig.3] which represents this step which essentially contains first-order filters which are easy to implement.

[0083] In the box marked 14, the anticipatory filtering delivers a corrected instruction Cpnt expressed as a linear combination of the flat variable corresponding to the desired trajectory, and its first and second derivatives +^ / ')

[0084] Concerning the feedback loop 15, once the term linked to the anticipation is known, it remains to establish the feedback term. In order to have a stabilized correction centered at 0, it is wise to identify a measurable variable of the system allowing to have a zero static gain. We choose the torsion speed œ't. We can write in the Laplace domain.

[0085] .... A v Cri) ri'

[0086] We then note that the transfer function of the feedback loop has a zero static gain: _ q.

[0087] Thus, the feedback loop 16 takes the form of a proportional regulator expressed from the derivative of the flat variable as follows: C™ = k,if-y) the output value Cpet being added to the corrected setpoint Cprit-

Claims

Claims

1. Method for controlling an electric machine torque in a powertrain of an electric vehicle, using a control loop servocontrol, the powertrain comprising the electric machine engaged with a differential, right and left wheel drive shafts and right and left wheels, the control loop comprising: - an acquisition of a desired machine torque, - an anticipatory filtering, acting on the desired machine torque - a feedback loop, the control loop delivering a corrected machine torque, resulting from the anticipatory filtering and the feedback loop, characterized in that the anticipatory filtering uses a flatness property.

2. Method according to claim 1, characterized in that the drive train has a kinematic behavior that the method models by: - ​​a moment of inertia Jp of the rotating parts upstream of the differential, - a moment of inertia Jr of the rotating parts downstream of the differential, - a torsional stiffness Kr, characterizing at least the torsional stiffness of the wheel transmission shafts, - a viscous friction coefficient Kv of the transmission line.

3. Method according to claim 2, characterized in that the basic state variables of the system œt and Ct are defined as follows: œt: torsion speed, time derivative of 0t which is the torsion angle Ct: torsion torque, the torsion torque Ct being able to be written Ct = Kr 0t + Kv œt.

4. Method according to claim 3, characterized in that a flat variable is defined as follows: y = Ct - Kv œt

5. Method according to claim 4, characterized in that the anticipatory filtering delivers a corrected setpoint Cpnt expressed as a linear combination of the flat variable, and its first and second derivatives: C^'t - ÿ (ÿ + y + ^y javec Jt = (Jp Jr) / (Jp + Jr).

6. A method according to claim 4, characterized in that the base and output state variables are expressed as a function of the flat variable, and its first and second derivatives: a't ~ ^y CP=:^r(ÿ+y;y+^y+^C^

7. Method according to claim 4, characterized in that to construct a torsion torque trajectory, a state variable filter is used by giving a desired damping and pulsation, with the formula Cf (ü„Ct =œ„2Cl the notation * characterizing the desired behavior, and the notation denoting the value w in the stationary state (oscillations canceled).

8. Method according to claim 4, characterized in that the anticipatory filtering delivers a corrected instruction Cpnt expressed as a linear combination of the flat variable corresponding to the desired trajectory, and its first and second derivatives: +~yj

9. Method according to claim 8, characterized in that the feedback loop takes the form of a proportional regulator expressed from the derivative of the flat variable as follows: _ j) 'the output value being added to the corrected setpoint Cpnt-

10. A motor vehicle comprising a powertrain comprising an electric machine engaged with a differential, right and left wheel drive shafts and right and left wheels, the powertrain being controlled at least in part by a control unit (10) configured to implement the method according to one of claims 1 to 9.

Images