You are on page 1of 6

Proceedings of the 7th Asian Control Conference, Hong Kong, China, August 27-29, 2009

SaB1.4

Novel Clamping Force Control for Electric Parking Brake Systems


Minseok Jang, Young O. Lee, Student Member, IEEE, Wongoo Lee, Choong W. Lee, Student Member, IEEE, Chung C. Chung, Member, IEEE, and Youngsub Son
without using the sensor was proposed for EMB systems [4]. In [4], the motor rotor position and the motor current were used for clamping force estimation. The characteristic curves of the relation between motor angle/ motor current and clamping force of an EMB caliper were used to estimate the clamping force. The clearance between the brake pads and the brake disk needed to be adjusted when the brake is released. And the stiffness quotient was evaluated to detect the contact point between the brake pads and the disk independent of the clamping force setting signal. In this paper, we propose a novel clamping force control method independent of the clearance between the brake pads and the brake disk unlike the conventional method [4]. This control method consists of three steps: (1) detecting of the contact point between the brake pads and the brake disk, (2) clamping force estimation, (3) a novel on-off control method considering inertia effects. First, the proposed control method does not need the clearance management since we can approximately detect the initial contact point between brake pads and the brake disk with the angular velocity of motor. To detect the contact point, we derive the relationship between the angular velocity of motor and the contact point. From this relationship, we will show the contact happens when the angular velocity of motor reaches near its maximum. Second, we will show that the clamping force can be estimated with a function of the angular displacement of the motor from the maximum angular velocity. The function is approximated as a second order polynomial. Third, a novel on-off control method considering the inertia effects is employed for low-cost control. Since the motor continues to rotate about several tens milliseconds even after power-off, we propose a novel clamping force prediction method to decrease the inertia effect of DC motor. The control method brings the smaller error between the target clamping force and the final clamping force than the simple on-off control method. The proposed control method is validated by experimental results. From the experimental results we observed that the proposed method meets the clamping force specifications. This paper is organized as follows: In Section II, the structure and specifications of EPB system are introduced. Also detecting method of the contact point and the estimation method of the clamping force are introduced. In Section III, the inertia effect is analyzed to decrease the error between the target clamping force and the final clamping force. In Section IV, the experiments are executed to validate the proposed estimation method. The conclusions are given in Section V.

AbstractIn this paper we present a novel claming force control method for electric parking brake systems without using force sensors. The control method consists of three steps for a simple control structure. First, we show how to approximately detect the initial contact point between brake pads and a brake disk with the angular velocity of motor. Second, the clamping force is estimated with the function of angular displacement from the maximum angular velocity. Third, since the motor continues to rotate about several tens milliseconds even after power-off we propose a novel on-off control method to decrease the inertia effect of DC motor. The proposed control method is validated by experiments. It enables low cost manufacturing of electric parking brake systems due to its simple control structure.

Keywords-electric parking brake, clamping force, estimation, DC motor, inertia effect

I. INTRODUCTION

electric parking brake (EPB) system that generates clamping force for parking by controlling a DC motor has been introduced to improve safety and convenience for drivers. At the press of a button on the dashboard a driver can easily apply or release the parking brake. The EPB systems provide some advantages. They allow for a larger interior space than conventional parking brake systems. And drivers with physical difficulties can activate the parking brake easily just by pushing the button. The EPB systems also support dynamic braking and anti-lock braking system (ABS) as well as the conventional parking brake function [1], [2]. Because of these advantages, the production of vehicles equipped with EPB systems is increasing. Force sensors have been used to measure clamping force in EPB or electric mechanical brake (EMB) systems, which are brake-by-wire systems. However, it is difficult to install the force sensor due to the limited mounting space and the cost of the sensor [3]. Therefore, a clamping force estimation method

ECENTLY,

Manuscript received February 15, 2009. This work was supported in part by the Minster of Knowledge and Economy, Republic of Korea under Grant 10014728. M. Jang, Y. O. Lee, W. Lee, C. W. Lee are with Dept. of Electrical Engineering, Hanyang University, Seoul 133-791, KOREA (e-mail: Skyler.Jang06@gmail.com, e-mail: foryou5252@hotmail.com, e-mail: hillfolk219@gmail.com, and chungwoo.lee@gmail.com) C. C. Chung is with Div. of Electrical and Biomedical Engineering, Hanyang University, Seoul 133-791, KOREA (+82-2-2220-1724; fax:+82-2-2291-5307; e-mail: cchung@hanyang.ac.kr) Y. Son is with Central R&D Center, MANDO Corporation, Kyonggi-Do 446-901, KOREA (e-mail: ysson@mando.com)

978-89-956056-9-1/09/2009 ACA

1588

7th ASCC, Hong Kong, China, Aug. 27-29, 2009

SaB1.4

In Fig. 2, we see that the final clamping forces are different even for the same angular displacement due to different clearances. To resolve this problem, we studied the relationship between the angular velocity of motor and the contact time.
5 4.5

Comparison between 3 signals' output


Angular velocity Motor current Clamping force

II. CLAMPING FORCE ESTIMATION A. Structure and specifications of EPB system In this paper, we study a stretch type EPB system depicted in Fig. 1. It includes a controller, parking cables, a DC motor, a gear box, an assembly of screw, a nut, a controller, a motor driver, and a current sensor. The parking cables of EPB system are connected to the brake pads. There are two operating modes: applying and releasing force mode. In the applying force mode, the clamping force is increased by pulling the parking cables using the DC motor until it reaches the target force [13]. In the releasing force mode, the clamping force is decreased by releasing the parking cables reversely. In order to measure the angular displacement of motor, the current sensor is used by using current ripples as [5], [6]. In Table I, specifications of EPB systems are listed [7].
TABLE I SPECIFICATIONS OF EPB SYSTEM Target force Settling time Permitted error bound 80 100 [kgf] Less than 1 [s] 10[%] of target force

magnitude[not to scale]

Fig. 1. Structure of EPB system.

3.5 3

2.5 2

1.5 1

0.5 0 The initial contact point 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

-0.5 0

time [s] Fig. 3. The clamping force, the angular velocity and the current in applying force mode

Fig. 3 shows the motor current, the angular velocity, and the clamping force in applying force mode. The results show that the three signals have correlations. The angular velocity decreases and the motor current increases when the clamping force increases. Thus we will derive the relationship between the angular velocity and the clamping force from now on. The dynamics of the DC motor may be represented by (1) - (3) [8].

va (t ) = Rim (t ) + Ldim (t ) / dt + vemf (t )


vemf (t ) = K b (t )

(1) (2) (3)

Tm = K m im (t )

B. Detecting the contact point The clearance between the brake pads and the brake disk is different whenever the brake is released. To overcome this problem, the conventional method requires clearance management [4]. Three experimental results for different clearances are plotted in Fig. 2.
120

where R is the resistance, L is the inductance, and Kb is the back emf constant of the motor. The torque developed by the motor is Tm and the motor torque constant is Km. The motor current is im(t), the applied input voltage is va(t) and the back emf is vemf(t). The torque relation in EPB system is given by Tm = Ta + Ti + T f (4)

Clamping force vs. angular displacement


Release for long time Release for mid time Release for short time

100

80

force [kgf]

where Ta is the applying torque of EPB system, Ti is the inertia torque, and Tf is the friction torque. Then Ta is proportional to the clamping force, f(eff), which is defined by Ta = t f ( eff ) (5)

60

40

20

where eff (t ) := (t ) (tmax ). t is a constant determined by the screw gain and gear ratio. Details on the modeling are referred to [9]. The clamping force, f(eff), is a function of the angular displacement from the contact point. This function will be derived in Section . C. And Ti is proportional to the

0 0

20

40

60

80

100

120

angular displacement [rev] Fig. 2. Clamping force versus angular displacement for different clearances.

1589

7th ASCC, Hong Kong, China, Aug. 27-29, 2009

SaB1.4

(t ) , as defined by (6) angular acceleration,

(t ) = (ta )e( K K
b

/ RJt )( t ta ) Kb K m ( t ta ) RJ t

(t ) Ti = J t

(6)

R( t + ) K v R f (eff ) m a 1 e Kb K m Kb K m

(13)

where Jt is the total inertia constant. In (4), Tf is a sum of the viscous friction, Tv, and coulomb friction, Tc as follows (7). T f = Tc + Tv (7)

The coulomb friction dominates the viscous friction torque in the EPB system so that we have Tc >>Tv. Thus (7) can be approximated such as Tf Tc [4]. Then we can rewrite Tc as a function of f(eff) with two constants and defined by [10], [11] T f Tc = f ( eff ) + . Equation (4) can be rewritten as (9).
(t ) + f ( eff ) + Tm = t f ( eff ) + J t
Fig. 4. Comparison between (ta) and (tmax)

(8)

(9)

In (1), L can be disregarded because the response of the armature circuit is much faster than the mechanical response. And (10) can be derived by rearranging (1), (2), (3), and (9).

(t ) = va (t ) / K b R( t + ) f ( eff ) /( K b K m )
(t ) + R ) /( K b K m ) ( RJ t

(10)

It is possible to detect the point, tmax, when the motors angular velocity is the maximum using current ripples [6]. However, it is not possible to detect the point, ta, when the parking cable force begins increasing without using a force sensor. Thus in this paper we use (tmax) instead of (ta) for clamping force estimation. It is essential to investigate the influence on the force control from this position difference, = (tmax) - (ta).
3.5 3

Let ta be the time of the contact point between the brake pads and the brake disk. Then (10) can be rewritten by
K m va K b K m (t ) R , t ta RJ t (11) (t ) = K m v (t ) K b K m (t ) t + f ( ), t > t a eff a RJ t Jt Jt RJ t

Comparision vs. t a

2.5

[rev]

2 1.5 1 0.5 0 0

In (11), va(t) is a constant, whose magnitude is va , for t>0 because the controller uses a on-off control for low-cost control in this paper. In t ta case, (12) is a solution of the differential equation for (t).

0.05

0.1

t [s]
a

0.15

0.2

0.25

(t ) =

( K m va R ) (1 e ( K K
b

/ RJ t ) t

Fig. 5. Comparision of =(tmax) - (ta) vs. different ta

Kb K m

(12)

In (12), if the input voltage, va(t), is large enough such as Kmva(t) - R > 0, the angular velocity, (t), will increase at least until the contact point as depicted in Fig. 4. In t > ta case, a solution of the differential equation for (t) is (13).

Fig. 5 is the simulation results with the parameters used in [7]. In Fig. 5, is plotted as a function of the time of contact point, ta. Here, ta=0 is the case the clearance is zero. For ta>0.2, is nearly zero. In other words, we see that ta tmax for a large clearance. Further, the worst case of is less than 3.3[rev]. The possible clamping force error due to the maximum for each target force is listed in Table II. In the operating target force range, these errors are acceptable based on the specification in Table I.

1590

7th ASCC, Hong Kong, China, Aug. 27-29, 2009

SaB1.4

TABLE II THE FORCE ERROR DUE TO AT EACH TARGET FORCE [UNIT: KGF] Target force 80 90 100 Force error 2.7 3.8 4.2

i = (T W ) 1 T WY a

(i = 1, 2,3)

(15)

C. The clamping force vs. the angular displacement


140 120 100

i s are coefficients, T is the input matrix, W is a where a diagonal matrix with the weights in the diagonal, and Y is the output matrix. We used the higher weighting factor for the larger clamping force. In Fig. 6 the measured clamping forces by the force sensor and the estimated clamping force using (14) are plotted. The estimated clamping forces are plotted by the green dotted lines. The maximum estimation error is less than 3.5[kgf]. III. ANALYSIS OF THE INERTIA EFFECT First we tried a simple on-off control for the EPB system. The simple on-off control means that it cuts off the input signal when the clamping force reaches the target force. Such a simple switch off control makes the motor continue to rotate further due to its momentum even without power. Although this simple on-off control meets the specifications, there is a large error. If the prediction how much the inertia effect makes the motor rotate is possible, the error can be reduced. We propose a novel prediction method analyzing the inertia effect and validate it with experiments. When the power is off, the input voltage, va, becomes zero, (10) may be rearranged as follows

angular displacement vs. force


1st 2nd 3rd 4th 5th estimated force

force [kgf]

80 60 40 20 0 0

10

20

30

40

50

60

70

80

90

angular displacement [rev] Fig. 6. Comparison between the approximated function, f(eff), and the absolute clamping force measured by the force sensor.

(t ) R K b K m (t ) = R( t + ) f ( eff ) RJ t
In (16), we linearize f(eff) such as f(eff)=kl

(16)

We showed that the parking cables begin applying the clamping force near the time of the maximum angular velocity of the motor in Section II. B. In this section, we will validate that the clamping force can be estimated from the angular displacement of the motor after the maximum angular velocity. Fig. 6 shows the experimental results about correlation between the clamping force and the angular displacement, which is counted after the maximum velocity point, tmax, for different clearances. The clamping force is measured by the force sensor. Unlike Fig. 2, Fig. 6 shows that f(eff)s agree within a bound which is tolerable. We calculated an approximated function of f(n) as f (n) = a1n2 + a2 n + a3 ; n = / 2 (14)

at the desired clamping force. The linearized spring constant, kl is obtained from the clamping force function (14).
kl ( desired ) = f (eff ) /
desired

= 2a1 desired + a2

(17)

Taking the Laplace transform of (16) and rearranging its result gives

(s) = (t f )

1 (18) J t s s 2 + ( Kb Km / RJ t )s + ( t + )kl / J t

where n is the rotational number of the motor after the maximum velocity point. The rotational number is counted using the current ripples. This method estimates the angular displacement and angular velocity of the motor by detecting periodical oscillations of the armature current caused by rotor slots. The method has higher resolution than cheap digital position sensor like an encoder. Moreover, the method has less computational effort than model based estimation methods than state observer [5], [6]. In order to get coefficients, ais of (14), the weighed least square estimation method (15) is used [12].

where tf is the time when the input signal is zero. From the final value theorem, the steady-state values of is given by

() = /{( t + )kl },

( < 0)

(19)

Since it is assumed that (tf)=0 in (18). () is the additional angular displacement of the motor after the power is cut off due to the inertia effect. Thus we can calculate the final clamping using (14) and (19). It is possible to reduce the error by shutting off the power before the estimated force reaches

1591

7th ASCC, Hong Kong, China, Aug. 27-29, 2009

SaB1.4

the target force. In Fig. 7, the additional angular displacement and the final clamping forces are plotted when the power is cut off at each target force.
6 additional angular displacement [rev] 70 75 final clamping force [kgf] 5.8 80 85 5.6 90 95 5.4

The simple on-off control applies the maximum input signal until the estimated clamping force reaches the target force for apply action. And the input signal becomes zero when the estimated force reaches or over the target force as (20),
u u (t ) = max, 0, e>0 e0 (20)

where e = f target f est. , u(t) is input voltage for the motor, umax is maximum voltage of the battery, and e is the error between the estimated clamping force and the target force. C. Experimental results We controlled the EPB system using the clamping force function (14) with the simple on-off controller (20). In order to validate the proposed method for different clearances, we performed experiments for three different target clamping forces under different clearances and the results are shown in Fig. 9.
110

100 105

5.2 74

76

78

80

82

84 86 88 90 target force [kgf]

92

94

96

110 98 100

Fig. 7. The additional angular displacement and the final clamping force under each target force

The left y axis is the additional angular displacement of the DC motor and the right y axis is the final clamping force. IV. VALIDATION OF THE PROPOSED METHOD A. Experiment environment

EPB control with estimated force


80kgf from force sensor 80kgf from estimated force 90kgf from force sensor 90kgf from estimated force 100kgf from force sensor 100kgf from estimated force

100 90 80

force [kgf]

70 60 50 40 30 20 10 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

time[s] Fig. 9. Experimental results with the simple on-off control.

Fig.8. Experiment environment of the EPB system.

In order to control the clamping force of the EPB system, a stretch type EPB system is used shown in Fig. 8. The system consists of a force indicator, a load cell measuring the absolute tension of parking cables, a controller board with a digital signal processor (TMS320F2812), a motor driver, and a module box including the motor, a gear box, and an encoder. B. Control method Three control methods were evaluated in [13]: On-off, linear Proportional (P), and nonlinear P. The nonlinear P controller had the best robustness and uniform tracking performance among them. However, the nonlinear P needs a PWM driver. For a low-cost control, we use an on-off control.

In order to compare the estimated forces with the measured forces, which is gained from the force sensor embedded in the module of the EPB system, the measured forces are plotted by solid lines in Fig. 9. We see that there is a good agreement. The initial point where the estimated force follows the force sensors output is different since the experiment is performed under different clearance. And the maximum error between the estimated forces and the measured forces is about 3.5[kgf] which is within the error bound. We observed that the estimated force is much less than the measure force after the power is cut off. The reason is that the current ripple is not reliably detected after the power cut off. The EPB system has the maximum apply time of less than 0.8[s]. It thus satisfies the specification of the applying time. The average and the deviation of the errors are plotted in Fig. 10. We repeated 30 times applying and releasing for five different target clamping forces using the two on-off control methods (the simple on-off and the novel on-off).

1592

7th ASCC, Hong Kong, China, Aug. 27-29, 2009

SaB1.4

10 8 6 4

Simple on-off control Novel on-off control

Error[kgf]

2 0 -2 -4 -6 -8 -10 80 85 90 95 100

Target force[kgf]
Fig. 10. The clamping force errors: Simple on-off control and novel on-off control method

applications, in IEEE Industry Applications Society Annual Meeting , pp. 1217-1224, 2004. [6] H. B. Chung, C. W. Lee, D. H. Lee, C. C. Chung, Y. S. Son, and P. Yoon, A Fault detection method for electric parking brake (EPB) systems with sensorless estimation using current ripples, in 14th Asia Pacific Automotive Engineering Conf., no. 2007-01-3660, Aug. 2007. [7] I. Yang, Y. Son, I. Park, K. Noh, and P. Yoon, Development and performance evaluation of electronic parking brake, in KASE Spring Conf., no. KSAE06-S0152, pp. 1836-1837, May 2006. [8] N. S. Nise, Control Systems Engineering (4th ed.), CA: John Wiley & Sons, 2004, ch. 2. [9] Y. O. Lee, C. W. Lee, C. C. Chung, Y. Son, P. Yoon, and I. Hwang, Stability analysis of electric parking brake(EPB) systems with a nonlinear proportional controller, in Proc. of 17th IFAC World Congress, Seoul, Korea, pp. 14247-14253, June 2008. [10] C. Line, C. Manzie and M. Good, Control of an electromechanical brake for automotive brake-by-wire systems with an adapted motion control architecture, SAE Tech. Paper, vol. 113, no. 2004-01-2050, May 2004. [11] H. Olsson, K. J. strm, C. C. d. Wit, M. Gfvert, and P. Lischinsky, Friction models and friction compensation, European Journal of Control, vol. 4, no. 3, pp. 176195, 1998. [12] K. J. strm, B. Wittenmark, Adaptive control (2nd ed.), New York: Wesley, 1995, ch. 2. [13] Y. O. Lee, C. W. Lee, H. B. Chung, C. C. Chung, Y. Son, P. Yoon and I. Hwang, A nonlinear proportional controller for electric parking brake(EPB) Systems, in 14th Asia Pacific Automotive Engineering Conf., no. 2007-01-3657, Aug. 2007.

In Fig. 10, negative error means that the over clamping force is applied. We see that every experiment satisfies the specification even no matter the inertia effect is considered or not. The novel on-off control, however, provides less averaged errors than the simple on-off control. Therefore, a low-cost control of the EPB system is possible using the novel on-off control. V. CONCLUSION This paper proposed the clamping force control method of EPB systems without using force sensors. This method used the relationship between the angular displacement of motor and the clamping force from the initial contact point between brake pads and a disk. We introduced how to approximately detect the initial contact point. The clamping force was estimated with the clamping force function of the angular displacement after the maximum angular velocity. We used on-off controller to make a simple control logics. We proposed the novel clamping force prediction method to decrease the inertia effect of DC motor. We executed experiments to validate the proposed method and showed that every experiment satisfied the specifications of the EPB systems. Due to its simple control structure, a low cost EPB system can be implemented with the proposed method. REFERENCES
[1] [2] [3] [4] [5] H. M. Jung, C. H. Park, J. W. Jeon, and H. Yeo, Development of electric parking brake system, in KSAE Autumn Conf., no. KSAE07-F0154, pp. 962-967, Nov. 2007. D. Hyun, I.-S. Kim, and H. Seo, Study on a dynamic braking algorithm for an electronic parking brake, in KSAE Autumn Conf., no. KSAE07-F0152, pp. 950-955, Nov. 2007. Y. Iwashita, Force control method based on disturbance load estimation, U.S. Patent 5 734 242, Mar. 31, 1998. R. Schwarz, R. Isermann, J. Bohm, J. Nell, and P. Rieth, Clamping force estimation for a brake-by-wire actuator, SAE Tech. Paper, vol. 108, no. 1999-01-0482, Mar. 1999. A. Consoli, G. Bottiglieri, R. Lector, R. Ruggeri, A. Testa, and S. De Caro, Sensorless position control of DC actuators for automotive

1593

You might also like