16 2. CO-DESIGN OPTIMIZATION FOR CYBER-PHYSICAL VEHICLE SYSTEM
For the feedback term, a linear proportional-integral (PI) controller is adopted to damp the
torsional oscillation:
T
fb
D
K
P
C K
I
Z
dt
e
0
(2.26)
e
0
D T
m;tgt
2T
hs
=i
0
i
g
; (2.27)
where the feedback gains K
P
and K
I
are tuning parameters of the PI controller, and e
0
is the
tracking error.
(3) Controller for conservative driving style: Since the conservative drivers usually care more
about energy efficiency and smooth driving feel by carefully operating the brake and acceleration
pedals, the low-level plant controller adopts the same combined feed-forward and feed-back
architecture as the moderate one to ensure vehicle drivability.
2.4 DRIVING-STYLE-BASED PERFORMANCE
EXPLORATION AND PARAMETER OPTIMIZATION
2.4.1 DESIGN SPACE EXPLORATION
Based on the system constrains formulated, namely the requirements for vehicle speed, grade-
ability, and brake stability shown in Equations (2.3)–(2.6), the boundaries of the related physical
plant parameters can be calculated, and the design space is then achieved.
2.4.2 PERFORMANCE EXPLORATION METHODOLOGY
In order to carry out multi-objective optimization under different driving styles, the impacts of
related parameters on the performance indicators should be explored. To do so, the following
exploration algorithm is proposed (see Table 2.3).
As shown in Table 2.3, assuming that, within the Parameter Library , there are several
parameters, namely P
1
, P
2
, : : :, P
i
, C
1
, C
2
, : : :, C
j
, deciding one Performance. P
1
; P
2
; : : : ; P
i
represent parameters of the physical plant, while C
1
; C
2
; : : : ; C
j
indicate controller variables.
Under pre-defined driving event E with valid design space, the selected vehicle Performance
is simulated in the Simulink environment stepping each parameter with a suitably small step.
After simulation-based global exploration, the Best Performance K with its corresponding value
selections of the parameters can be attained.
2.4.3 DRIVING-STYLE-ORIENTED MULTI-OBJECTIVE
OPTIMIZATION
(1) Aggressive-driving-style based optimization: is driving style requires to maximize vehicle
dynamic performance first and foremost. However, a good performance in terms of energy effi-
ciency is also expected to be guaranteed. erefore, the trade-off between dynamic performance
2.4. DRIVING-STYLE-BASED PERFORMANCE EXPLORATION 17
Table 2.3: Algorithm for performance exploration
Algorithm 1: Performance Exploration
Input: Parameter Library{P
1
, …, P
i
, C
1
,…,C
j
} ⊆ ξ, Event E
Output: Best Performance Point K
function Global Exploration (ξ, E)
Performance ←{}; Paras ←{};
while p
1
∈P
1
do
while p
2
∈P
2
do
⋮
while p
i
∈P
i
do
while c
1
∈C
1
do
while c
2
∈C
2
do
⋮
while c
j
∈C
j
do
Performance ← Simulation (E, P
1
,..,P
i
,C
1
,…,C
j
)
end while
Paras ← Performance (C
j
);
⋮
end while
Paras ← Performance (C
1
,C
2
,…,C
j
);
end while
Paras ← Performance (P
i
, C
1
,C
2
,…,C
j
);
⋮
end while
Paras ← Performance(P
1
, P
2
,…,P
i
, C
1
,C
2
,…,C
j
);
K ← Best Performance Point (Paras);
Return K, Paras
end function
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.