Professional Documents
Culture Documents
Fuzzy Gear Shifting Control Optimisation To Improve Vehicle Performance, Fuel Consumption and Engine Emissions
Fuzzy Gear Shifting Control Optimisation To Improve Vehicle Performance, Fuel Consumption and Engine Emissions
Fuzzy Gear Shifting Control Optimisation To Improve Vehicle Performance, Fuel Consumption and Engine Emissions
Research Article
Jony Javorski Eckert1 , Fabio M. Santiciolli1, Rodrigo Y. Yamashita1, Fernanda C. Corrêa2, Ludmila C.A.
Silva1, Franco G. Dedini1
1Integrated Systems Laboratory (LabSIn), Integrated Systems Department (DSI), University of Campinas - UNICAMP, 200 Mendeleyev Street,
Campinas 13083970, SP, Brazil
2Electrical Engineering Department, Federal Technological University of Paraná - UTFPR, Monteiro Lobato Avenue, Jardim Carvalho, Ponta
Abstract: This study presents a multiobjective optimisation applied to the gear shifting fuzzy control of a vehicle equipped with
an automated manual transmission (AMT), aiming to improve acceleration performance and reduce engine fuel consumption
and emissions. An Adaptive-Weight Genetic Algorithm was employed to find optimum fuzzy membership functions, according to
the input and output ranges, and also optimum control rules with their respective weights. The vehicle behaviour is represented
by longitudinal dynamics simulations developed in Simulink/Matlab™ interface, associated with the ADVISOR™ fuel converter
block, that provides the engine emissions and fuel consumption. These simulations were based on the FTP-75 emissions test
procedure, that considers cold and hot phases of the driving cycle evaluating the engine transient operation as a function of the
catalyst efficiency during the warm-up period. The optimum fuzzy control with the best trade-off among the optimisation criteria
presented 19.72% fuel saving associated with 12.90% hydrocarbon, 29.20% carbon monoxide and 17.02% nitrogen oxides
emissions reduction and an acceleration performance improvement when compared to a standard gear shifting procedure for a
manual controlled gearbox. Moreover, the optimised fuzzy gear shifting control, improves the relationship between fuel
consumption and emissions significantly, when compared to another optimum AMT control based on speed limits only.
IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669 2659
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
r = 0.98rg(1 + 0.01kv(V)) (7)
2 Ro3 − Ri3
T cl = μclFnN 2 (8)
3 Ro − Ri2
T tNtNdηtd ax
Fx = − ((Ie + It)(NtNd)2 + IdNd2 + Iw) 2 (10)
r r
Fig. 3 Tire slip coefficient in the function of the traction force Ft
However, the tires present a maximum transmissible force in
the contact, and it limits the vehicle acceleration performance. The The speed V [m/s] is defined by numeric integration of ax by
maximum tire traction force Fmax [N] is defined by the (11) ODE45 provided by the Simulink™ database. The engine speed ωe
according to the tire-ground friction coefficient μ, the vehicle [rad/s] is then calculated by the (14), according to the transmission
wheelbase L [m], gravity centre height h [m], the longitudinal ratio (NtNd), tire dynamic radius r [m] and the tire slipping
distance between the rear axle and the gravity center c [m] and the coefficient e that estimates the difference between the tire
vehicle weight W [kg]. tangential speed and the vehicle displacement speed, as a function
of the traction force Ft as shown in the Fig. 3.
Wc Mh ax
Fmax = μ − (11)
2L 2L V NdNt
ωe = (14)
r(1 − e)
The effective traction force Ft [N] will be the lower value
between the available Fx and maximum Fmax traction forces, 2.1.1 Vehicle parameters.: Table 1 shows the parameters for the
therefore Fx = Fmaxi f Fx ≥ Fmax or Ft = Fxi f Fx < Fmax . The vehicle simulated vehicle based on the 1.0L engine frontal traction
longitudinal acceleration ax [m/s2] is calculated by Brazilian Chevrolet Celta™.
Ft − Rx − DA
ax = (12) 2.1.2 ADVISOR™ engine model.: Once the engine operation
M point (T e and ωe) is defined, the fuel consumption (FC) [l] and
emissions (hydrocarbon (HC) [g/km] NOx [g/km] and carbon
However, the traction force Ft is a function of the acceleration
monoxide (CO) [g/km]) are calculated by the ADVISOR™ fuel
ax in both limiting criteria (available traction force and tire
converter block as in [10], according to the maps shown in the Fig.
transmissible force), therefore, the available (Fx) and maximum
4. ADVISOR™ is a free vehicular simulation tool developed by
(Fmax) forces need to be recalculated in an iterative process among [28], that contains engine specific fuel consumption and emissions
the (10)–(12). After the convergence, the engine effective torque T e maps from tests performed according to FTP standards for transient
[Nm] is defined by regimes.
In the simulations, the engine starts considering the ambient
Ftr ax temperature of 20° C, and according to the operation regime, the
Te = + ((Ie + It)(NtNd)2 + IdNd2 + Iw) (13)
NtNdηtd r ADVISOR™ fuel converter block estimates the generated
combustion heat and its transfer by the tailpipe, changing the
catalyst efficiency that is multiplied by the emissions maps outputs.
2660 IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
algorithm to find, a point T 6 that defines where the input torque
will be considered maximum High.
The third input is the engine temperature, that affects directly
the tailpipe emission during the cold start section of the FTP-75
cycle. This membership function is divided into three triangular
stages: cold, heating, and hot, with a range between the ambient
temperature (20° C) and the engine ideal running temperature (95°
C), controlled by the cooling system.
The output membership function presents unitary range, and it
is composed of three stages: downshift (trapezoidal) that aims to
increase the transmission ratio, keep gear (triangular) resulting in
maintaining the current gear, and upshift (trapezoidal) that aims at
the decrease of the transmission ratio. Therefore, the fuzzy output
has value 0 ≤ FZ ≤ 1, that is used to control the AMT to define the
1 ≤ Gear ≤ 5 to be used according to the current gear Gi. The gear-
shifting process follows the rules shown by the (15), as a function
of the inferior Li and superior Ls limits, that are also defined as
design variables in the optimisation process. However, some rules
override the FZ command, as the gearbox neutral position
(Gear = 0) that acts when the speed is null (V = 0) and there is not
a throttle signal input. The engine maximum (ωe = 6500 rpm) and
minimum (ωe = 900 rpm) speeds acts as constraints. Fig. 5 shows
the membership functions and the points that can be alternated by
the optimisation process
T = [T 1 T 2 T 3 T 4 T 5 T 6] (17)
The rules with their respective weights are also optimised. The
control is composed of 27 rules FR(n) (1 ≤ n ≤ 27) that represent all
Fig. 4 Engine maps for fuel consumption and emissions [10, 28] combinations among the three inputs, and for each rule, the
(a) Specific fuel consumption, (b) Specific CO emissions, (c) Specific NOx emissions, optimisation algorithm selects one of the three possible outputs:
(d) Specific HC emissions downshift (FR(n) = 1), keep gear (FR(n) = 2) or upshift (FR(n) = 3),
and defines the rules weights 0 ≤ RW(n) ≤ 1. The Table 2 shows the
2.2 Optimisable fuzzy logic gear shift control rules and weights.
In the same way, as the membership function design variables,
The AMT is controlled by Mamdani fuzzy (Matlab™ toolbox) the rules and weights are also combined in vectors.
inference system, and the fuzzy control is optimised using a genetic
algorithm. The optimisation process occurs off-line, and the FR = [FR1 FR2 FR3 … FR25 FR26 FR27] (20)
optimum fuzzy control with the best trade-off among the
optimisation criteria is used to operate in real-time the AMT RW = [RW 1 RW 2 RW 3 … RW 25 RW 26 RW 27] (21)
system. The fuzzy AMT control has three membership functions
inputs. The first one is the engine speed, with the range between The defuzzification method is also optimised among the five
the maximum allowed (6500 rpm) and the idle speed (900 rpm) of options provided by the fuzzy mamdani toolbox. The
the engine, and it is divided into three triangular stages: low, defuzzification method is represented by the design variable
normal, and high. 1 ≤ DF ≤ 5 where centroid (DF = 1), bisector (DF = 2), mom
The second input is the wheels required traction torque T rp (5),
(DF = 3), lom (DF = 4) and som (DF = 5).
that considers only the powertrain traction torque. This
membership function range is defined from null until 1500 Nm,
that is the maximum tire-ground transmissible torque. This input is 2.3 Problem formulation
also divided into three stages: low (triangular), normal (triangular) The optimisation is based on three criteria that provide trade-off
and high that is trapezoidal because the traction torque T rp usually solutions among minimum fuel consumption and emissions in the
does not reach the limit (1500 Nm), allowing the optimisation same way as improving vehicle acceleration performance. The first
IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669 2661
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
a function of FTP-75 V cs, and the simulated V s speeds, discretised
in s intervals of 0.1 s [10], and the mean values V¯ c and V̄ of the
simulated and standard profiles.
∑ V cs − V¯ c V s − V̄
2
R= (23)
∑ V cs − V¯ c ∑ V s − V̄
2 2
2662 IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3 convergence, the fuzzy AMT control configurations, presented in
f kmax − f k(X)
Ft(X) = ∑ f kmax − f kmin
+ Pp(X) (29) final Pareto frontier, are considered as optimum compromised
k=1 solutions. Among these solutions, the most promising ones are
analysed in the next section.
2.4.1 Selection, crossover and mutation.: After classification
according to the Ft(X) value, two solutions that are denominated 3 Results and discussion
member 1 (M1) and member 2 (M2) are randomly selected from the
population by the roulette wheel technique, that increases the The algorithm convergence results in the 3D pareto frontier of non-
selection probability SP(X) for the higher Ft(X) among the dominated solutions among the three convergence criteria as shows
population members (1 ≤ X ≤ PM) as shows (30). the Fig. 6. The best-compromised solution is the one that presents
the highest fitness value max (Ft) among the Pareto frontier
Ft(X) solutions. Among the optimum solutions presented in Fig. 6, four
SP(X) = PM (30) solutions are selected: The fuel economy solution (min f 1), the low
∑X = 1 Ft(X)
emissions solution (min f 2), the best performance solution
(min f 3), and the compromised solution (max Ft).
The crossover operator combines the design variables (or sets of
fuzzy design variables) presented in the chromosomes of X(M1) Analysing the optimised solution, it was possible to observe
that the performance improvement is directly associated with fuel
and X(M2) to generate new solutions. The combination is
consumption increase (Fig. 6d). The best performance (min f 3) and
performed by defining the crossover operator 0 ≤ Cr(v) ≤ 1 using
fuel economy (min f 1) solutions present a high value for the EF
the Matlab™ function rand. If Cr(v) ≤ 0.5, the crossover keeps the
(Figs. 6a and b), and the low emissions min f 2 and best
design variable from M1 (vM1), on the other hand if Cr(v) > 0.5 the
compromised (max Ft) solutions present close behaviour, both
design variable will be provided by M2 (vM2). located in an intermediary consumption/performance regime (Fig.
The fuzzy membership function points will be combined with 6d) that decreases the engine tailpipe emissions.
their respective vectors (R, T, C, G, FR and RW) to avoid To provide a standard result to be compared to the optimised
unstable chromosomes that do not respect the constraints C. The solutions, an alternative gear shift control based on speed limits
crossover chromosome XCr is simulated and the results are (same used as acceptable performance limit constraint) is defined
included in the population if the acceptable performance limit was as a standard solution. The Table 4 shows the four selected
satisfied. optimum solutions reached results, and Table 5 exposes their
To insert new design variables values into population, the respective chromosomes.
mutation operator randomly changes some of the values of the The vehicle behaviour using the standard gear shifting strategy
crossover chromosome XCr. Differently of the crossover operator, is shown in Fig. 7, where it is possible to observe the resulting gear
the mutation process is applied for each design variable shifting and speed profiles, engine speed and temperature during
(1 ≤ j ≤ 80), including the ones stored in vectors (R, T, C, G, FR the 2500 s of the FTP-75 driving cycle. This standard solution is
and RW). The algorithm defines random value of the mutation fully dominated by the optimised configurations, and it is only used
operator 0 ≤ Mt( j) ≤ 1. If Mt( j) < 0.5, the mutation operator to provide a comparison with the optimised ones.
keeps the value of the crossover chromosome XMt( j) = XCr( j), on the
other hand, if Mt( j) ≥ 0.5, the design variables will be mutated 3.1 Fuel economy solution
according to the mut values shown in the Table 3.
The fuel economy solution (min f 1) presents the lowest fuel
If the mutated chromosome XCr meets the constraints C (27), it
consumption to perform the FTP-75 driving cycle, presenting
is simulated and the results are included in the population if the
24.49% fuel saving with 8.99% performance improvement as
performance constraint is satisfied.
compared to the standard solution. The tailpipe emissions present
values close to the standard solution (-13.19% NOx, +10.34% CO
2.4.2 Population limit and convergence criterion.: The and −2.15% HC), but the resultant EF is higher than the min f 2
crossover and mutation operators increase the population size PS. and max Ft as shown in Table 4. The high EF of the fuel economy
When the population reaches a limit value of Plim = 100, the solution as compared to the low emissions (min f 2) and the best
algorithm eliminates the higher Pareto ranking solutions (worst compromised (max Ft) solutions, is resultant of the engine warm-
results) from the database. To avoid eliminating the whole up period increase as shows the Fig. 8.
population, when it becomes composed only of the Pareto frontier The fuel economy solution keeps the engine operating at low
(rank=1), and reaches Plim size, this limit increases as speed. In almost all cycle this strategy provides an upshift before
Plim = Plim + 50. the engine reaches 3000 rpm, as shows the Fig. 9, reaching the
The algorithm convergence is determined by the repeatability of fourth and fifth gears in all acceleration sections, saving fuel
the Pareto frontier of non-dominated solutions for 50 generations, because it is a higher efficiency region of the specific fuel
which indicates the stagnation of the evolution process. After consumption map (Fig. 4a). Furthermore, running the engine at
IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669 2663
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 6 Engine maps for fuel consumption and gaseous emissions
(a) f 1 × f 2 × f 3, (b) f 1 × f 2, (c) f 2 × f 3, (d) f 1 × f 3
Table 4 Reached results of the selected solutions Table 5 Chromosomes of the selected solutions
Results Solutions X Solutions
Standard min f 1 min f 2 min f 3 max Ft min f 1 min f 2 min f 3 max Ft
FC [l] 1.47 1.11 1.17 1.64 1.18 R [4235 1992 [2481 1355 [2481 1355 [2481 1355
EF — 2.11 0.16 2.21 0.25 4029 5019 4638 5962 4638 5606 4638 5962
1−R 0.00089 0.00081 0.00015 0.00009 0.00013 4994] 3329] 2831] 3329]
distance, km 17.59 17.60 17.73 17.76 17.73 T [459 220 616 [367 241 454 [221 140 454 [344 241 454
999 727 1229] 680 515 1257] 680 474 1257] 680 515 1326]
NOx, g/km 0.235 0.204 0.192 0.360 0.195
C [29.61 29.41 [64.79 24.07 [29.61 29.41 [24.71 20.73
CO, g/km 1.911 1.977 1.397 1.680 1.353
49.30 69.35 47.07 84.58 49.30 69.35 63.30 92.17
HC, g/km 0.279 0.273 0.238 0.262 0.243 67.18] 67.50] 67.18] 66.16]
Ft — 2.72 3.85 3.02 3.86 G [ 0.3 0.31 0.09 [0.30 0.31 0.12 [0.31 0.35 0.20 [0.31 0.35 0.20
0.54 0.92 0.54 0.54 0.82 0.75 0.43 0.57 0.40 0.43 0.57 0.40
0.59 ] 0.85] 0.75 ] 0.75]
low speed provides less ignition per time interval, saving fuel, but FZ bisector bisector mom bisector
prolonging the engine warm-up period (where the majority of the Li 0.0785 0.1028 0.0668 0.0785
engine emissions are generated), increasing the tailpipe emissions
because of the longer transient regime period, until the catalyst Ls 0.5520 0.5520 0.8027 0.5520
reaches maximum efficiency.
The second parameter that influences the engine emissions are
the maps shown in Figs. 4b–d, that provides different amounts of cycle section, decreasing the emissions of NOx and HC. On the
tailpipe gases, as a function of the operation region. This effect is other hand, the CO emissions do not decrease enough to
observed when the fuel economy and the standard solutions are compensate the lower catalyst efficiency at the warm-up period and
compared, even with the standard solution providing a quicker therefore, the fuel economy solution results in 10.34% more CO
warm-up period (that improves the catalyst efficiency at the cold than the standard one.
start phase), this solution generates more NOx and HC than the The fuel economy optimised fuzzy logic membership functions
fuel economy strategy (Table 4). Even with the lower catalyst are shown in Fig. 10, and the rules with their respective weights are
efficiency at the start-up period, the fuel economy strategy runs the presented in the Table 6.
engine at better NOx and HC operation points during the remaining
2664 IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 7 Vehicle behaviour using the standard solution
Fig. 8 Engine warm-up according to the gear shifting strategy Fig. 10 Membership functions of the fuel economy solution
(a) Engine speed input, (b) Required traction torque input, (c) Engine temperature
input, (d) AMT control output
Fig. 9 Vehicle behaviour using the fuel economy solution The min f 3 solution ensures the best performance by
postponing the upshifts, that decrease the vehicle speed due to the
Table 6 Minimum fuel consumption Fuzzy control rules and engine torque supply interruption by the clutch decoupling during
weights the gear shifting process. These performance losses caused by
Engine temperature consecutive gear shiftings are clearly observed in the fuel economy
Engine Traction Cold Heating Hot solution (Fig. 12a) that anticipates the upshift, quickly reaching the
speed torque Output-weight [FR(n) - RW(n) ] fourth and fifth gears, reducing the available traction torque, that
does not enable the vehicle to reach the target speed even with the
low low keep-0.97 down-0.98 down-0.30
use of the maximum engine torque. On the other hand, the best
normal up-0.74 keep-0.97 down-0.52 performance solution only upshifts until the third gear (Fig. 12c),
high down-0.19 keep-0.21 keep-0.64 reaching the engine maximum speed (6500 rpm) several times as
normal low up-0.69 keep-0.45 down-0.48 shows the Fig. 13, that allows high available traction torque, but
normal keep-0.74 up-0.33 up-0.83 increases the fuel consumption and emissions by operating the
high down-0.67 keep-0.86 up-0.64 engine at poor efficiency regions.
Even with the quicker engine temperature increase (Fig. 8), that
high low up-0.74 keep-0.67 up-0.53
provides better catalyst efficiency during the cold start phase of the
normal keep-0.71 keep-0.24 down-0.99 cycle, the best performance solution presents the worst EF value
high up-0.85 up-0.48 down-0.58 among the optimum solutions (Figs. 6b and c). These phenomena
occur because of the engine operation region at low torque and
high speed, that results in high CO, NOx and HC emissions even
3.2 Best performance solution with the catalyst efficiency improvement at the beginning of the
cycle. However, the best performance solution decreases in 12.02%
The best performance solution (min f 3) presents the speed profile
CO and 6.09% HC emissions as compared to the standard solution,
closer to the FTP-75 as compared to the other analysed solutions. but with 53.19% more NOx associated with 11.56% fuel
Figs. 11 and 12 present the gear shifting profile of the high-speed consumption increase. The best performance optimised fuzzy logic
stretch of the hot start phase of the FTP-75 cycle (2140 to 2190 s).
IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669 2665
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 13 Vehicle behaviour using the best performance solution
Fig. 12 Vehicle performance and gear shifting profile Fig. 14 Membership functions of the best performance solution
(a) Fuel economy min f 1, (b) Low emissions min f 2, (c) Best performance min f 3, (d) (a) Engine speed input, (b) Required traction torque input, (c) Engine temperature
Best compromisse max Ft input, (d) AMT control output
membership functions are shown in Fig. 14, and the rules and This behaviour is also used by best-compromised solution as
weights are presented in the Table 7. shows the Fig. 15b, but in this case, it does not postpone the third
to the fourth upshift until the engine speed limit to save fuel. The
gear shifting profile, engine speed, temperature and resultant
3.3 Low emissions solution
vehicle speed are shown in Fig. 16. The fuzzy rules and weights
The low emissions solution (min f 2) results in −18.30% NOx, are presented in the Table 8, and the membership functions shown
−26.90%CO and −6.09% HC as compared to the standard solution, in Fig. 17.
with 20.41% fuel saving. This solution mixes some characteristics
of the previously analyzed gear shifting control strategies. To 3.4 Compromised solution
ensure low fuel consumption, the low emission strategy anticipates
The best-compromised solution (max Ft) is located in the best
the upshifts, presenting behaviour close to the fuel economy
region in the relation of performance versus fuel consumption (Fig.
solution (Figs. 12a and b), keeping the engine running at a good
6d) and close to the low emissions solution, that ensure the best
efficiency region in the stabilised and hot start phases of the cycle.
trade-off among the three optimisation criteria evaluated, saving
On the other hand, to avoid the low catalyst efficiency during the
19.72% fuel as compared to the standard solution, with -17.02%
engine warm-up period, the (min f 2) solution presents gear shifting
NOx, -29.20% CO and -12.90% HC emissions. This solution also
behaviour close to the best performance solution, in the cold start presents an expressive improvement in the vehicle performance
phase of the cycle as shows the Fig. 15a that presents the same reducing the 1 − R relation from 0.00089 (standard) to 0.00013
cycle stretch previously shown in the Fig. 12 (but for the cold start
(max Ft). The simulated vehicle dynamic behaviour and gear
phase), where is possible to compare the difference between the
shifting profile are presented by Fig. 18.
gear shifting profiles caused by the engine temperature input
The best compromised solution presents behaviour close to the
control.
low emissions solution as shown in the Figs. 6c and e and Table 4,
2666 IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 7 Best performance Fuzzy control rules Table 8 Minimum emissions consumption fuzzy control
Engine temperature rules and weights
Engine Traction Cold Heating Hot Engine temperature
speed torque Output-weight [FR(n) − RW(n)] Engine Traction Cold Heating Hot
low low down-0.41 up-0.49 down-0.02 speed torque Output-weight [FR(n) − RW(n)]
normal up-0.73 down-0.58 keep-0.87 low low keep-0.15 keep-0.58 keep-0.59
high down-0.85 down-0.01 up-0.06 normal up-0.57 down-0.90 down-0.88
normal low keep-0.21 up-0.95 up-0.80 high down-0.18 keep-0.64 down-0.11
normal keep-0.72 down-0.40 up-0.52 normal low keep-0.20 down-0.77 keep-0.87
high down-0.13 up-0.24 down-0.26 normal down-0.84 down-0.01 up-0.35
high low down-0.34 up-0.99 up-0.18 high down-0.65 keep-0.72 keep-0.40
normal up-0.94 keep-0.56 down-0.72 high low keep-0.65 down-0.30 down-0.72
high down-0.51 keep-0.63 keep-0.81 normal up-0.99 up-0.30 up-0.88
high up-0.88 up-0.31 up-0.96
Finally, it was observed that the best compromised and the low
emissions solutions (among others with close behaviour) were able
to fully dominate the standard gear shifting control results.
IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669 2667
© The Institution of Engineering and Technology 2019
17518652, 2019, 16, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-cta.2018.6272 by Universidade Federal Da Paraiba, Wiley Online Library on [02/11/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 9 Best compromised consumption fuzzy control rules
and weights
Engine temperature
Engine Traction Cold Heating Hot
speed torque Output-weight [FR(n) − RW(n)]
low low down-0.79 keep-0.33 keep-0.34
normal down-0.76 down-0.85 down-0.49
high keep-0.77 down-0.88 keep-0.51
normal low keep-0.88 down-0.30 keep-0.26
normal keep-0.06 down-0.29 up-0.22
high down-0.33 keep-0.18 down-0.46
high low up-0.84 down-0.59 keep-0.19
normal keep-0.84 up-0.76 up-0.02
high up-0.21 up-0.65 up-0.56
IET Control Theory Appl., 2019, Vol. 13 Iss. 16, pp. 2658-2669 2669
© The Institution of Engineering and Technology 2019