You are on page 1of 10

Chapter 7

Alternative Control Methods

The purpose of this chapter is to survey other modifications to the skyhook control
policy, known as the groundhook and hybrid control policies. These control policies are
generally designed for two-degree-of-freedom systems and do not directly apply to the
SDOF seat suspension used in this research. This chapter, however, will explore the
possibility of further modeling the seat suspension as a 2DOF system. This chapter will
also experimentally evaluate the merits of such control policies.

7.1 Groundhook Control Policy


As was shown for an SDOF system, the optimal control policy for reducing the motion of
the seat is the skyhook control policy. If we consider both the vehicle cab and seat
suspension, then the system becomes a 2DOF system, as shown in Fig. 7.1.

M1 X1, V1
SEAT

K1 C1

X2, V2
M2
CAB

K2 C2
Xin, Vin
CHASSIS

Figure 7.1. A 2DOF Model of the Cab and Seat Suspension.

70
This configuration also represents a single suspension of a vehicle. We would
like to apply the same concept as skyhook damping to form what is known as
groundhook damping. To do this, let's review the skyhook configuration for a 2DOF
suspension system, as shown in Fig. 7.2.

CSKY
M1 X1, V1 M1 X1, V1

K1 K1 ≈CSKY
X2, V2 X2, V2
M2 M2

K2 K2
Xin, Vin Xin, Vin

Ideal Skyhook Realistic Skyhook

Figure 7.2. Skyhook Formulation for the 2DOF System.

We immediately notice a potential problem. With the skyhook damper only


removing energy from M1, the motion of M2 becomes excessive due to the fact that no
damping is applied to it. Of course, in the realistic system, the semiactive damper will
still remove some of the energy from M2, since the semiactive damper is mounted
between the two masses. In vehicle suspensions, this problem is known as wheelhop,
and can cause excessive axle (mass 2) bounce and loss of contact with the road surface.
Now consider the case where we move the skyhook damper to mass 2, as shown
in Fig. 7.3. This is known as the groundhook configuration, since we are attempting to
"hook" the second mass to the ground.

71
M1 X1, V1 M1 X1, V1

K1 K1 ≈CGND
X2, V2 X2, V2
CGND M2 M2

K2 K2
Xin, Vin Xin, Vin

Ideal Groundhook Realistic Groundhook

Figure 7.3. Groundhook Formulation for the 2DOF System.

The potential problem with groundhook damping is that the motion of M1


becomes undamped. However, as in the skyhook case, this will not truly happen since
there is some residual damping between M1 and M2 due to the off-state and location of
the damper.
To derive the equations for the groundhook control policy, we proceed exactly as
described for Eq. (2.6). The relative velocity between mass 1 and mass 2 is defined to be
positive when separating. If mass 2 is moving with a positive velocity and mass 1 and
mass 2 are approaching each other (V12 is negative), the groundhook damping force
should be negative (i.e., in the downward sense). Thus,
V2V12<0 FSA = CGNDV12 (7.1)
where FSA is the semiactive damper force. Again, if V2 is positive but M1 and M2 are
separating (V12 is positive), the groundhook damping force is still in the negative
direction. The semiactive damper, however, cannot produce a force in this direction,
therefore the damping force must be made as small as possible, or ideally
V2V12>0 FSA = 0 (7.2)
We can show the same results for all combinations of V2 and V12. Summarizing Eqs.
(7.1) and (7.2) and rewriting them yields

72
-V2 V12 > 0 FSA = C GND V2
 (7.3)
-V2 V12 < 0 FSA = 0

Similar to the skyhook system, applying the damper model to Eq. (7.3) gives
-V2 V12 > 0 i = βV2
 (7.4)
-V2 V12 < 0 i = 0

where i is the damper current and β is a constant. Equation (7.4) is the functional
representation of how groundhook control is programmed in a semiactive controller.

7.2 Hybrid Control Policy


As observed earlier, the skyhook control policy minimizes seat motion (or vehicle body
motion for a primary suspension) at the cost of cab motion (or axle motion), whereas
groundhook control minimizes cab motion at the expense of seat motion. We would like
to take advantage of the desirable aspects of each control policy and choose a more
suitable compromise for a given application. This is the fundamental idea of hybrid
control, shown in Fig. 7.4.

CSKY
M1 X1, V1 M1 X1, V1

K1 K1 ≈CHYB
X2, V2 X2, V2
CGND M2 M2

K2 C2 K2 C2
Xin, Vin Xin, Vin

Ideal Hybrid Realistic Hybrid

Figure 7.4. Hybrid Control.

73
The control policy is a combination of the skyhook and groundhook damper
control signals:
V1V12 > 0 σ SKY = V1 
 
V1V12 < 0 σ SKY = 0 
{i = H[ασ SKY + (1- α )σ GND ] (7.5)
−V 2 V12 > 0 σ GND = V2 
 
−V 2 V12 < 0 σ GND = 0 

where σSKY and σGND are the skyhook and groundhook components of the current to the
damper, respectively, α is the relative ratio between skyhook and groundhook control,
and H is a constant gain. When α is 1, the control policy reduces to pure skyhook, while
when α is 0, the control is purely groundhook.
To practically implement hybrid control, we need to know V1, V2, and the relative
velocity V12. For ease of application in a research environment (i.e., proof of concept),
we will measure V1, and V2 using an accelerometer and analog integrator exactly as the
skyhook controller measures V1. The relative velocity, V12, is then computed using the
difference between V1 and V2. This entire system was implemented using the dSPACE
test system at Lord. Figure 7.5 shows a block diagram representation of the hybrid
controller, while Fig. 7.6 shows the SIMULINK code for the hybrid controller.

74
Analog
Integrator

Accelerometer Hybrid
Control
M X1, V1 Policy

dSPACE
Controller
MR K Analog
Damper
Integrator
Accelerometer

BASE X2, V2
Same as the
Lord Skyhook
Controller
Current
Driver

Figure 7.5. Block Diagram of the Hybrid Controller.

S x(n+1)=Ax(n)+Bu(n) -0.3
y(n)=Cx(n)+Du(n)
Seat Acc ADC
S Unit Abs Vel1 Scale
Seat Accel 'Rate' Filter factor + * Abs
Base Acc 1. *
-
V12V12 Gain Product Switch Abs Product2
DS1102AD x(n+1)=Ax(n)+Bu(n) -0.3 0
y(n)=Cx(n)+Du(n)
0.75
Abs Vel2 Scale Off State Current
Base Accel Rate Filter Alpha +
factor 50 S
- +
1 + Output DAC Current
Output Saturation
One 1-Alpha Sum Unit
Gain
-1.0 * DS1102DA
Invert Product3
* Abs
Product1 Switch1 Abs1

0
Off State
Current1

Figure 7.6. SIMULINK Code for the Hybrid Controller.

Using this system, any of the skyhook, groundhook, or hybrid control policies
can be implemented. All of the gains internal to the SIMULINK code were adjusted so
that the output in a skyhook control mode (α = 1) matched the original skyhook
controller as closely as possible.

75
7.3 Experimental Results
Figure 7.7 shows the swept sine acceleration transmissibility of the groundhook system
for two gains. As expected, the resonance peak is relatively large for both cases. Figure
7.8 shows the time response for the groundhook system for H=50. The damper current
often saturates at the maximum level of 1 Amp, indicating that the controller is
attempting to control the system. There are, however, relatively low levels of jerk for the
extreme current discontinuities.

Groundhook Controller, α=0


3

2.5 H=25
H=50
Acceleration Transmissibility

1.5

0.5

0 0 1
10 10
Frequency, Hz

Figure 7.7. Acceleration Transmissibility of Groundhook Control.

76
Groundhook Time Response, α=0, H=50
10

Acceleration, m/s2
5

-5

-10
0 0.5 1 1.5 2 2.5 3 3.5 4

1.5
Damper Current, Amps
1

0.5

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Time, Sec.

Figure 7.8. Time Response of the Groundhook Controller.

Figures 7.9 and 7.10 show the acceleration transmissibilities for the hybrid systems.
Increasing the gain, H, shifts the transmissibility in frequency, whereas increasing α
(moving towards a skyhook control) decreases the level of transmissibility. Comparing
the hybrid control to a pure skyhook system shows that the skyhook system has much
better high-frequency isolation.

Hybrid Controller, H = 32
1.4

α=0.25
1.2 α=0.50
α=0.75
Acceleration Transmissibility

0.8

0.6

0.4

0.2

0 0 1
10 10
Frequency, Hz

Figure 7.9. Hybrid Transmissibility for H=32.

77
Hybrid Controller, H = 50
1.4

α=0.25
1.2 α=0.50
α=0.75

Acceleration Transmissibility
1

0.8

0.6

0.4

0.2

0 0 1
10 10
Frequency, Hz

Figure 7.10. Hybrid Transmissibility for H=50.

Finally, Figs. 7.11 through 7.13 show the acceleration of the seat in the time domain for
various levels of α and H. Increasing α increases the amount of jerk, which makes sense
since for the pure groundhook, control there is very little jerk, while the pure skyhook
control tends to have a great deal of jerk. Further, increasing the gain H also increases
the amount of jerk.
Hybrid Controller, α=0.25, H=50
10

4
Acceleration, m/s2

-2

-4

-6

-8

-10
0 0.5 1 1.5 2 2.5 3 3.5 4
Time, Sec.

Figure 7.11. Acceleration Time Response for α=0.25 and H=50.

78
Hybrid Controller, α=0.75, H=32
10

Acceleration, m/s2
2

-2

-4

-6

-8

-10
0 0.5 1 1.5 2 2.5 3 3.5 4
Time, Sec.

Figure 7.12. Acceleration Time Response for α=0.75 and H=32.

Hybrid Controller, α=0.75, H=50


10

4
Acceleration, m/s2

-2

-4

-6

-8

-10
0 0.5 1 1.5 2 2.5 3 3.5 4
Time, Sec.

Figure 7.13. Acceleration Time Response for α=0.75 and H=50.

79

You might also like