CN101624063B - An Automobile Electric Power Steering System Without Torque Sensor - Google Patents

Abstract

The invention provides an automobile Electric Power Steering (EPS) system without a torque sensor, which comprises: 1. a steering column torque sensor in the existing automobile Electric Power Steering (EPS) system is eliminated; 2. the electric power-assisted steering of the automobile is realized by adopting a brand new control method. By establishing a state space mathematical model of a non-torque sensor automobile Electric Power Steering (EPS) system, observing other system state components by adopting an improved Kalman observer algorithm and taking a power-assisted motor rotor angle (system state component) as measurement input, and calculating to obtain the torque quantity of an automobile steering column in the EPS system, the power-assisted torque of the control motor of the non-torque sensor EPS closed-loop control system is established. The invention cancels a steering column torque sensor in the prior EPS system, simplifies the structure and the manufacturing difficulty of the steering gear, increases the reliability of the system, reduces the complexity of the EPS system and greatly reduces the cost of the EPS system.

Description

A kind of automobile electric booster steering system of non-torque sensor

Technical field

The present invention relates to automobile electric booster steering system, be specifically related to a kind of automobile electric power-assisted steering (EPS) system that does not have the Steering gear torque sensor.

Background technology

Automobile electric power-assisted steering EPS system has than traditional hydraulic power-assist steering system remarkable advantages in each side such as fuel efficiency, modularization, road feel adjustability and environment friendly.Present automobile electric power-assisted steering EPS system mainly contains following four types, is respectively Steering gear booster type, miniature gears booster type, tooth bar booster type and double pinion booster type.All these four kinds of EPS systems all have three basic elements of character: control unit (ECU), assist motor be installed in the torque sensor on the Steering gear.This has just determined that a common feature of existing EPS control policy is that they all depend on the Steering gear torque sensor.The popular control method of existing product EPS adopts traditional proportional-integral-differential (PID) controller to handle the input that the torque sensor data produce current assist motor.Used torque sensor is the special-purpose parts of EPS, and price is higher.In addition, existing torque sensor technology all realizes the measurement to moment of torsion through the angle deformation of its torsion bar.In order to increase sensitieness, its torsion bar rigidity is less than Steering gear rigidity, and this has increased its output noise; Also there is the rigidity softening problem to Steering gear in it in addition, has influenced the stability of chaufeur road feel and control.

Summary of the invention

Above-mentioned deficiency to the existence of existing automobile electric power-assisted steering EPS system; The purpose of this invention is to provide a kind of Steering gear torque sensor that do not need; Reduce the complexity of EPS system, increase system reliability, reduce the automobile EPS system of the no Steering gear torque sensor of EPS system cost.

The objective of the invention is to realize like this: a kind of automobile electric booster steering system of non-torque sensor comprises control unit (ECU), angular transducer and assisted electric machine; Said control unit (ECU) is according to the space mathematical model of the non-torque sensor automobile electric booster steering system of setting up; Carry out improved Kalman's observer algorithm; With assist motor rotor angle (state of the system component) serves as to measure input, observes the Steering gear rotational angle theta _c, other state of the system component such as current of electric i and rack displacement p, by formula

$T_c=K_c({\mathrm{\θ}}_c-\frac{p}{r_p})$ (K _cBe Steering gear stiffness coefficient, r _pBe the miniature gears radius) obtain Steering gear torque capacity T _c, set up the booster torquemoment that non-torque sensor EPS closed loop control system is controlled motor thus.

Compare prior art, the present invention has following beneficial effect:

1, through setting up the space mathematical model of non-torque sensor EPS system; With assist motor rotor angle (state of the system component) serves as to measure input; Adopt improved Kalman's observer algorithm to observe other unknown state of the system component of system; Steering column for vehicles torque capacity in calculating the EPS system; Set up the booster torquemoment of non-torque sensor EPS closed loop control system control motor thus, thereby cancelled essential torque sensor in the existing EPS system architecture, simplified the EPS system architecture; Saved the loaded down with trivial details operation of the installation and the debugging of torque sensor in the EPS system; Saved processing to the output noise of torque sensor; And the torque sensor among the former automobile electric booster steering system EPS is a special-purpose member, and price is higher, its elimination can be reduced greatly the cost of EPS system.

2, advanced state space designs model rather than traditional input-output design mock-up have been adopted; The rotor angular position that has obtained with assist motor is output, and has combined the non-torque sensor EPS system state space math modeling of assist motor kinetics equation; Improved the Kalman and observed algorithm can not effectively handle the shortcoming of Non-zero Mean signal, made it be suitable for the observation of the Steering gear moment of torsion in the non-torque sensor EPS system; Used improvement Kalman moment of torsion observation algorithm can effectively be eliminated the influence of vibration random noise, and the estimation of moment of torsion is had accuracy height, characteristics that cumulative errors are little.

3, eliminated because of of the influence of torque sensor fault, strengthened the reliability of EPS system greatly the EPS system; Increase the mechanical rigid of Steering gear in the EPS system, improved the stability of chaufeur road feel and control thus.

4, reduce the complexity of EPS system, be easy to the installation and the debugging of EPS system.

Description of drawings

Fig. 1 is non-torque sensor automobile electric power-assisted steering of the present invention (EPS) system architecture scheme drawing.

Fig. 2 is non-torque sensor automobile electric power-assisted steering of the present invention (EPS) system closed loop control system scheme drawing.

Fig. 3 is applied to Kalman's observer recursive algorithm flow process of the present invention.

The specific embodiment

Referring to Fig. 1, a kind of automobile electric booster steering system of non-torque sensor comprises control unit (ECU) 1, angular transducer 2 and assisted electric machine 3; Said control unit (ECU) 1 is according to the space mathematical model of the non-torque sensor automobile electric booster steering system of setting up; Carry out improved Kalman's observer algorithm; With assist motor rotor angle (state of the system component) serves as to measure input, observes the Steering gear rotational angle theta _c, other state of the system component such as current of electric i and rack displacement p, by formula

$T_c=K_c({\mathrm{\θ}}_c-\frac{p}{r_p})$ (K _cBe Steering gear stiffness coefficient, r _pBe the miniature gears radius) obtain Steering gear torque capacity T _c, set up the booster torquemoment that non-torque sensor EPS closed loop control system is controlled motor thus.

Innovation of the present invention is:

1, the torque sensor in the cancellation EPS system

Existing four types EPS system, the installation site of torque sensor is not quite similar.Wherein, the torque sensor of Steering gear booster type is installed on the Steering gear; The torque sensor of miniature gears booster type, double pinion booster type and tooth bar booster type EPS is installed in and turns to the miniature gears place.The present invention cancels the torque sensor in the EPS system, and its non-torque sensor EPS system architecture scheme drawing is as shown in Figure 1; Among the figure, the 1st, control unit (ECU), the 2nd, angular transducer, the 3rd, assisted electric machine, the 4th, steering handwheel, the 5th, Steering gear, the 6th, power-transfer clutch and speed reduction gearing, the 7th, Steering gear, the 8th, tooth bar, the 9th, miniature gears.

2, set up the space mathematical model of non-torque sensor EPS system

The kinetic model of EPS is by the quality in the system and rotor inertia and each spring and damping element mutual action and constitute.System is mainly low frequency motion, can ignore the influence of high stiffness elements.Fig. 1 non-torque sensor EPS system is made up of three essential parts: steering rack 8; Be coupled to the Steering gear 7 of tooth bar through miniature gears 9; Be connected with the assist motor 3 of independent rotation axle, it is connected with Steering gear 5 with speed reduction gearing 6 through power-transfer clutch, and intermediate rod connects tooth bar and tire.The inertia of having ignored tire pipe link and drive disk assemblies such as tire quality, tire motion, friction and gear in the model.The elastic constant that has added the tire pipe link in the model.

To Steering gear, motor shaft and tooth bar, can set up the kinetics equation of following non-torque sensor EPS system by the Lagrange's dynamical equations and the moment of momentum theorem respectively:

$J_c{\stackrel{\·\·}{\mathrm{\θ}}}_c+B_c{\stackrel{\·}{\mathrm{\θ}}}_c+K_c({\mathrm{\θ}}_c-\frac{p}{r_p})=T_d---\left(1\right)$

$J_m{\stackrel{\·\·}{\mathrm{\θ}}}_m+B_m{\stackrel{\·}{\mathrm{\θ}}}_m+K_m({\mathrm{\θ}}_m-\frac{\mathrm{pG}}{r_p})=\mathrm{ki}---\left(2\right)$

$M_r\stackrel{\·\·}{p}+B_r\stackrel{\·}{p}+K_{t}p=\frac{K_c}{r_p}({\mathrm{\θ}}_c-\frac{p}{r_p})+\frac{K_{m}G}{r_p}({\mathrm{\θ}}_m-\frac{\mathrm{pG}}{r_p})---\left(3\right)$

In the formula, J _cBe Steering gear rotor inertia, B _cBe Steering gear damping coefficient, K _cBe Steering gear stiffness coefficient, θ _cBe Steering gear corner, T _dBe steering-wheel torque; J _mBe motor shaft rotor inertia, B _mBe motor shaft damping coefficient, K _mBe motor shaft stiffness coefficient, θ _mBe the rotor corner, G is the speed reduction gearing reduction ratio, and k is the assist motor torque factor, and i is a current of electric; M _rBe tooth bar quality, B _rBe tooth bar damping coefficient, K _tBe the elasticity modulus of tooth bar, p is a rack displacement, r _pBe the miniature gears radius.

The kinetics equation of assist motor is:

$L\stackrel{\·}{i}+\mathrm{Ri}+k{\stackrel{\·}{\mathrm{\θ}}}_m=v---\left(4\right)$

In the formula, L is an assist motor stator winding inductance value, and R is the assist motor stator winding resistance, and v is a motor terminal voltage.

Can set up the state space description of non-torque sensor EPS system linearity by set of equations ((1)-(4)):

$\left\{\begin{array}{c}\stackrel{\·}{x}=\mathrm{Ax}+\mathrm{Bu}\\ y=\mathrm{cx}+n\left(t\right)\end{array}\right.---\left(5\right)$

N in the formula (t) is the random measurement noise, $x={\left[\begin{array}{ccccccc}{\mathrm{\θ}}_c& {\stackrel{\·}{\mathrm{\θ}}}_c& {\mathrm{\θ}}_m& {\stackrel{\·}{\mathrm{\θ}}}_m& p& \stackrel{\·}{p}& i\end{array}\right]}^T$ State for system; Steering handwheel input torque and motor terminal voltage are imported as system, u=[T _dV] ^TThe rotor rotational angle theta _mFine resolution angular transducer by on the motor shaft is measured, and as the output of this two single-input single-output system (SISO system).System matrix A, input matrix B, output matrix C is provided by following (6) formula.

$A=\left[\begin{array}{ccccccc}0& 1& 0& 0& 0& 0& 0\\ -\frac{K_c}{J_c}& -\frac{B_c}{J_c}& 0& 0& -\frac{K_c}{K_c{r}_p}& 0& 0\\ 0& 0& 0& 1& 0& 0& 0\\ 0& 0& -\frac{K_m}{J_m}& -\frac{B_m}{J_m}& -\frac{k_{m}G}{J_m{r}_p}& 0& \frac{k}{J_m}\\ 0& 0& 0& 0& 0& 1& 0\\ \frac{K_c}{M_r{r}_p}& 0& \frac{K_{m}G}{M_r{r}_p}& 0& -(\frac{K_t}{M_r}+\frac{K_c}{M_{\mathrm{<figure-callout\; id="2"\; label="r\; r\; p"\; filenames="H2009101045297E00011.png"\; state="\{\{state\}\}">r</figure-callout>}}{r}_p^{}{}_2^{}}+\frac{K_m{G}^2}{M_r{r}_p^2})& -\frac{B_r}{M_r}& 0\\ 0& 0& 0& -\frac{k}{L}& 0& 0& -\frac{B}{L}\end{array}\right],$

$B=\left[\begin{array}{cc}0& 0\\ 0& 0\\ \frac{1}{J_c}& 0\\ 0& 0\\ 0& 0\\ 0& 0\\ 0& \frac{1}{J_c}\end{array}\right],C=\left[\begin{array}{ccccccc}0& 0& 1& 0& 0& 0& 0\end{array}\right]---\left(6\right)$

3, the rotor angle with assist motor is input, adopts improved Kalman's observer algorithm to observe system's unknown state by system's known state, through calculating the EPS moment of torsion, sets up the closed loop control system of non-torque sensor EPS;

System state space shown in the formula (6) is described, and (A, B) controlled to fully, (A is C) to observing fully.Can see system as single v input, θ _mMeasure the system of output, and with T _dThe interference of system is regarded in input as.With system outlet θ _mBe input with motor terminal voltage v, adopt Kalman's observer to obtain the system state estimation amount

By formula

$T_c=K_c({\mathrm{\&theta;}}_c-\frac{p}{r_p})---\left(7\right)$

Can obtain Steering gear torque capacity T _cIt is attached to speed V s the power-assisted curve is tabled look-up, and assist motor current deviation value is adopted controller (PID or other controller) control assist motor terminal voltage, can set up non-torque sensor EPS closed loop control system as shown in Figure 2.

The Kalman's observer that adopts in the system is a kind of online minimum variance recursive algorithm, and it is image data on one side, calculate on one side, realize real-time monitored to state of the system.Its stepping type calculates can be by micro controller system (MCU) or the online completion of digital signal processor (DSP).Also accurate observation system state under zero-mean randol noise and noise.The equation of state of Kalman's observer does

$\stackrel{\&CenterDot;}{\hat{x}}=A\hat{x}+\mathrm{Bu}+k_o(y-C\hat{x})---\left(8\right)$ In the formula, k _oBe the estimator gain matrix.Can be made as at random to Gaussian white noise disturbs disturbing in the EPS system, and adopt its noise covariance to be used for the design of Kalman's observer from the measurement noise of sensor and road surface.But to steering handwheel input torque T _d, more than mention its interference as system handled, but when design Kalman observer, can not it be regarded as the zero-mean signal.For this reason, need do further to improve to Kalman's observer.As shown in Figure 2, for non-torque sensor automobile electric power-assisted steering of the present invention (EPS) system closed loop control system scheme drawing, with the estimated valve T of Kalman's observer _cT as its recursion calculating next time _dEstimation.

Kalman's observer is applied to micro controller system (MCU) or digital signal processor (DSP) before, needs discretization to handle.To coefficient matrices A, B, C adopt following approximate formula

A′＝e ^AT≈I+AT

$B^{\&prime;}=\underset{0}{\overset{T}{\&Integral;}}{e}^{\mathrm{At}}\mathrm{Bdt}\&ap;\mathrm{BT}$

C′＝C (9)

T is the employing time in the formula, T=t _K+1-t _kIn order to obtain satisfied observation effect, the sampling time is littler than the electric time of EPS system.Kalman's state estimation is divided into two stages, is respectively forecast period and calibration phase.At forecast period, at first calculate that by the k time estimated result predictor

of estimation is next time provided by following formula

${\tilde{x}}_{k+1}=A^{\&prime;}{\hat{x}}_k+B^{\&prime;}{u}_k---\left(10\right)$

The cooresponding output of this premeasuring

does

${\tilde{y}}_{k+1}=C{\tilde{x}}_{k+1}---\left(11\right)$

The equation of state of Kalman's observer of corresponding discretization does

${\hat{x}}_{k+1}=A^{\&prime;}{\hat{x}}_k+B^{\&prime;}{u}_k+k_{\mathrm{ok}+1}(y_{k+1}-{\tilde{y}}_{k+1})---\left(12\right)$

With (11) formula substitution (12) formula, can get

${\hat{x}}_{k+1}={\tilde{x}}_{k+1}+k_{\mathrm{ok}+1}(y_{k+1}-C{\tilde{x}}_{k+1})---\left(13\right)$

Y in the formula _K+1Be measured value, represent assist motor rotor angle θ _mThe correction of subordinate phase is mainly reflected in (13) formula, promptly utilizes the deviation of actual measurement output and prediction output that predicted state is carried out feedback compensation, to obtain satisfied state estimation.The result of feedback compensation also depends on gain matrix k _Ok+1Effect.k _Ok+1Selection principle be to make

Mean square error matrix minimalization.Usually, utilize covariance matrix P _K+1K derives _Ok+1 The mean square error matrix get the minimum P of being equal to _K+1Get minimumly, make P _K+1To k _Ok+1Derivative be zero, can derive k _Ok+1

Gradient matrix G _K+1And H _K+1, can obtain by following two formulas respectively:

$G_{k+1}=\frac{\&PartialD;}{\&PartialD;x}(A^{\&prime;}x+B^{\&prime;}u){|}_{x={\tilde{x}}_{k+1}}---\left(14\right)$

$H_{k+1}=\frac{\&PartialD;}{\&PartialD;x}\left(C^{\&prime;}x\right){|}_{x={\tilde{x}}_{k+1}}---\left(15\right)$

Finally, can get Kalman's recursion formula as follows,

${\tilde{x}}_{k+1}=A^{\&prime;}{\hat{x}}_k+B^{\&prime;}{u}_k---\left(16\right)$

${\tilde{P}}_{k+1}=G_{k+1}{\hat{P}}_k{G}_{k+1}^T+Q---\left(17\right)$

$K_{\mathrm{ok}+1}=\frac{{\tilde{P}}_{k+1}{H}_{k+1}^T}{H_{k+1}{\tilde{P}}_{k+1}{H}_{k+1}^T+R}---\left(18\right)$

${\hat{x}}_{k+1}={\tilde{x}}_{k+1}+k_{\mathrm{ok}+1}(y_{k+1}-{\tilde{y}}_{k+1})---\left(19\right)$

${\hat{P}}_{k+1}={\tilde{P}}_{k+1}-K_{\mathrm{ok}+1}{H}_{k+1}{\tilde{P}}_{k+1}---\left(20\right)$

Q and R are covariance matrix.The key of Kalman's state estimation is to confirm gain matrix k _Ok+1, and designing gain matrix k _Ok+1Key be the selection of initial value of Q, R and P.Usually, Q and R are unknown, can only be according to the qualitative selection of noise random character.

Fig. 3 is the diagram of circuit that utilizes Kalman's observer algorithm recursion estimation Steering gear torque capacity, comprises the steps:

At first initialized card Germania observer algorithm is promptly given initial condition and initial variance assignment;

Measure k+1 assist motor rotor angle value y constantly earlier by testing circuit _K+1(promptly surveying value);

Utilized the estimated valve in a last moment

Substitution (16) formula calculates this predictor constantly

Again by

Calculate gradient matrix G _K+1And H _K+1, and obtain the two transposed matrix G _K+1 ^TAnd H _K+1 ^T

Use the covariance matrix in a moment

Calculate the covariance matrix of current time

Predictor ((17) formula) calculates gain matrix k afterwards _Ok+1((18) formula);

Estimate the state estimation value of current time with (19) formula

Calculate current Steering gear torque capacity T by (7) formula _Ck+1

Calculate out covariance matrix

estimated valve of current time with (20) formula.

This takes turns end, can repeat state of the system and the estimation of Steering gear torque capacity that said process carries out a new round.

Explanation is at last; Above embodiment is only unrestricted in order to technical scheme of the present invention to be described; Although the present invention is specified with reference to preferred embodiment; Those of ordinary skill in the art should be appreciated that the improvement of in aim that does not break away from technical scheme of the present invention and scope, being done, and it all should be encompassed among the claim scope of the present invention.

Claims (1)

1. the automobile electric booster steering system of a non-torque sensor comprises control unit (ECU), angular transducer and assist motor; It is characterized in that; Said control unit (ECU) is according to the space mathematical model of the non-torque sensor automobile electric booster steering system of setting up; Carrying out improved Kalman's observer algorithm, serves as to measure input with state of the system component assist motor rotor angle, observes the Steering gear rotational angle theta _c, current of electric i and rack displacement p, by formula

K _cBe Steering gear stiffness coefficient, r _pBe the miniature gears radius; Obtain Steering gear torque capacity T _c, set up the booster torquemoment that non-torque sensor EPS closed loop control system is controlled motor thus;

Said improved Kalman's observer algorithm is a kind of online minimum variance recursive algorithm, on one side it can one side image data calculate, realizes the real-time monitored to state of the system, and execution in step comprises:

At first initialized card Germania observer algorithm is promptly given initial condition and initial variance assignment;

Measure k+1 assist motor rotor angle value y constantly earlier by testing circuit _K+1

Utilized the estimated valve in a last moment

Substitution

In calculate the predictor of carving this moment

U=[T _dV] ^T, again by

Calculate gradient matrix G _K+1And H _K+1, And obtain the two transposed matrix With T _dAs the steering handwheel input torque, v is a motor terminal voltage; System matrix A, input matrix B and output matrix C are respectively:

$A=\left[\begin{array}{ccccccc}0& 1& 0& 0& 0& 0& 0\\ -\frac{K_c}{J_c}& -\frac{B_c}{J_c}& 0& 0& -\frac{K_c}{J_c{r}_p}& 0& 0\\ 0& 0& 0& 1& 0& 0& 0\\ 0& 0& -\frac{K_m}{J_m}& -\frac{B_m}{J_m}& -\frac{k_{m}G}{J_m{r}_p}& 0& \frac{k}{J_m}\\ 0& 0& 0& 0& 0& 1& 0\\ \frac{K_c}{M_r{r}_p}& 0& \frac{K_{m}G}{M_r{r}_p}& 0& -(\frac{K_t}{M_r}+\frac{K_c}{M_r{r}_p^2}+\frac{K_m{G}^2}{M_r{r}_p^2}& -\frac{B_r}{M_r}& 0\\ 0& 0& 0& -\frac{k}{L}& 0& 0& -\frac{B}{L}\end{array}\right],$

$B=\left[\begin{array}{cc}0& 0\\ 0& 0\\ \frac{1}{J_c}& 0\\ 0& 0\\ 0& 0\\ 0& 0\\ 0& \frac{1}{J_c}\end{array}\right],$ C＝[0010000]；

J wherein _cBe Steering gear rotor inertia, B _cBe Steering gear damping coefficient, K _cBe Steering gear stiffness coefficient, J _mBe motor shaft rotor inertia, B _mBe motor shaft damping coefficient, K _mBe the motor shaft stiffness coefficient, G is the speed reduction gearing reduction ratio, and k is the assist motor torque factor, M _rBe tooth bar quality, B _rBe tooth bar damping coefficient, K _tBe the elasticity modulus of tooth bar, r _pBe the miniature gears radius;

Use the covariance matrix in a moment

Calculate the covariance matrix of current time Predictor,

Calculate gain matrix k afterwards _Ok+1,

Use

Estimate the state estimation value of current time

By

Calculate current Steering gear torque capacity T _Ck+1

With

calculate the covariance matrix of the current moment estimate.

Images