Fuzzy Logic ontrol o lutch or Hybrid Vehicle Vu Trieu Minh Mechanosystem - Department o Mechatronics Tallinn University o Technology trieu.vu@ttu.ee Abstract This paper provides a design o an automatic clutch controller or hybrid electrical vehicle (HEV using uzzy logic. The use o uzzy logic can reduce the diiculty o mathematical modeling o complex systems since uzzy logic can deal with uncertain and imprecise data and problems which may have several solutions rather than one. Fuzzy logic algorithms or the automatic clutch controller are developed to achieve a smooth and ast engaging transition. omprehensive simulations or the whole hybrid electrical vehicle are conducted in Matlab 9a. An experimental test or a real damping clutch is also carried out. Results show that the active regulation o the clutch slipping ration can considerably reduce the vehicle vibration in resonance requencies. The new system can handle the clutch engagement with low jerk and high comort. I. INTRODUTION This paper is a modiied version o a paper submitted and printing in Journal o Systems and ontrol Engineering []. The idea or the automatic clutch controller o hybrid electrical vehicles is that a dry-plate riction clutch in manual transmission always provides higher transmission eiciency (97% than a wet clutch (torque converter in automatic transmission (86%. I an automatic controller or a dry-plate riction clutch can be successully installed and the clutch pedal can be eliminated rom the vehicle, the driver can treat the new system like a normal automatic transmission. Since the dry riction clutch is always the most eicient transmission available and much cheaper than the torque converter in automatic transmission vehicles, this paper develops an automatic controller or this simple dry riction clutch. Some other obvious advantages o the new system including the reduction o noise and vibration are also investigated. In the parallel hybrid vehicle, the primary power source, an internal combustion engine (IE and the secondary power source, an electrical motor (EM are independently installed so that both can separately or together propel the vehicle. Typically the control o the transitional engagement between EM and IE are based on the heuristic knowledge on the characteristics o the IE and EM []. For comort and saety reasons, several control approaches or smothering the engagement o clutches have been developed including back stepping control [3], optimal control [4] and model predictive control [5] and [6]. In this paper, a new real-time uzzy logic scheme to control the automatic dry riction clutch is developed to control the engagement o the clutch. The motivation o using uzzy logic control in this study is the ability o an intelligent controller based on uncertain and imprecise inormation. In automotive industry, the successul applications o uzzy logic to control anti-lock braking system (ABS can be seen. The control close-loop time or this ABS is about 5 milliseconds. Within this time interval, the microcontroller can collect all sensor data, process and compute the ABS algorithms, drive the bypass valves or the brake luid, and conduct the brake activities successully. In the next section o this paper, a typical parallel hybrid vehicle and clutch engagement model is developed, and then, uzzy logic control algorithms are ormulated. omprehensive simulations or the hybrid vehicle are then conducted in Matlab 9a to illustrate the perormance o the new controller. Experimental results applied or a real damping clutch to veriy and handle the vehicle resonance vibration requencies are presented. And inally, conclusions and recommendations rom the study are drawn. II. LUTH ENGAGEMENT MODELING Figure shows the coniguration o a very common parallel HEV, which consists o a conventional IE and two EM and EM. An automatically controllable clutch separates the drivetrain into two parts: part - IE with EM, and part - EM and the rest o the transmission. EM serves as a starter and a generator. This rear wheel is equipped with a standard automated gearbox without a torque converter. Fig.. oniguration o parallel hybrid powertrain During the pure electrical drive at low speeds (less than 5km/h, the clutch is open and a series hybrid coniguration is achieved. In this operating range, the EM propels the vehicle. The transmission rom series mode to parallel mode takes place at high speeds (more than 5km/h by closing the clutch. The EM activates the IE to run the vehicle while the EM turns o. EM then acts as a generator to recharge the battery. During some critical operations, depending on the demand o the driver or during some essential heavy loads or other emergency cases i needed, both EM and EM can be automatically turned on to assist the IE to propel the vehicle. lutch is one o the most important components in a vehicle drivetrain. lutch allows the connection and transmission o the driving orce rom the IE or EM and EM to the wheels. The clutch system or this model consists o two dry riction disks connected in the ends o two rotating shats. One shat is attached to the IE or EM while the other shat is connected to the gearbox and the dierential gearbox to propel the vehicle. There are two separate modes o the clutch connections: locked together and rotating at the same angular speed (engaged, or decoupled and rotating in at dierent angular speed (disengaged. Handling transmission between the two 6
modes is a major modelling challenge. As the system loses one degree o reedom in the locking up process, the transmitted torque through this process becomes a discontinuity. The magnitude o the torque alls rom the maximum value to a lowest value necessary to keep the two disks o the clutch spinning at the same speed. The two dynamic models are used to simulate the locked mode (engaged and the decoupled mode (disengaged. A switching mechanism is developed to recognize the precise moments o the transitions between the two modes and activate the switches accordingly. Figure shows a clutch system, where M is the driving torque (input; F N is the normal orce between riction disks; J and J are moments o inertia; k β and k β are the damping coeicients in two shats; µ K and µ S are kinetic and static coeicients o riction; ω and ω are angular velocities o the two shats; ri and r O are the inner and outer radii o the clutch riction suraces; R is the clutch equivalent net radius; M is the torque transmitted through the clutch; And M R is the torque required or maintaining the locked position. J M ( J k Locked β J kβ ω M = M = (5 J + J The clutch remains locked until the magnitude o the riction Static torque ( M exceeds the static riction capacity ( M : Static M = R F Nµ S (6 3 A switching diagram or the clutch activities is illustrated in Figure. 3. Fig. 3. Friction mode transmissions Fig.. lutch system The dynamic equations or the locked mode are derived as ollows: J & ω = M kβω M ( J & ω = M ωk β The ull torque capacity in a clutch is a unction o its area A, riction orce ( F, and the corresponding radii ( ( r, r and O r I : T rf = A A A, or F µ T = r r θ, or ro π N π ( r ri O r I 3 3 ( ro ri T = R F Nµ with R = 3 ( ro ri When the clutch is slipping, the model uses the kinetic riction coeicient ( µ : K T = sgn( ω ω T = sgn( ω ω R F µ 3 Slipping N K When the clutch is locked or ω = ω = ω, the system acts as single unit and equation ( can be combined into a single equation or the locked mode: ( J + J & ω = M ( k + k ω, (4 or β β ( (3 In this clutch model, the engine speed rapidly changes and leads to the rapid changes in accelerations and jerks o the vehicle. The engine speed undegoes a rapid change o acceleration as it synchronizes with the drivetrain via the clutch engagement rate. For a low jerk property, it would be better to have the longest possible engagement time and to avoid any sudden step input to the clutch. However, in reality, this is not practical as excessive slipping leads to overheating o the clutch resulting a short operating lie. Ideally, any engagement should not last any more than 3 to 4 seconds. III. FUZZY LOGI ONTROL An automatic controller or the clutch must be designed to perorm two operating modes: shiting connections and changing gears. Shiting connection is the mode o the clutch engagement or disengagement depending on the driver intention via the driver action on the engine throttle. Thereore, the controller will determine the shiting connection mode via the position and the rate o the throttle operation. The changing gear mode is selected based on the current torque load and the vehicle velocity matching with the engine speed. Based on the requirements o the two clutch operating modes, uzzy logic control rules are selected with the use o slip-eedback regulator. These rules base on the dierence between the input and the output speeds o the clutch to create the eedback loop to control the engagement pressure via its slip gains. The clutch slip is simply the speed dierence between the input and the output shats and inally reaching zero when the clutch is locked. The clutch engagement pressure will be regulated proportionally on the slip rate with special notions on this clutch lockup and the idling engine speed. A set o uzzy logic algorithms is developed to control the automatic clutch, which can understand the driver intention. For example, an aggressive pressure on the accelerator pedal or a rapid throttle opening rate can be understood that the 7
driver needs a high pressure ramp rate on the clutch engagement or a short time. While a gradual pressure on the accelerator pedal will lead to a long and smooth clutch engagement period. A ast release o the accelerator pedal leads a ast drop o engine speed and ast reduction o clutch pressure. The clutch is disengaged with negative pressure gain and the engine is running in idling condition. The gear shiting operations are also activated accordingly based on the output torque load and the vehicle speed. A set o uzzy logic rules is built with variables or throttle in positions: losed (< %, Narrow (-5%, Normal ( 5-85% and Wide (85-% and in rates: Low (<5%/s, Normal (5-85%/s, and High (>85%/s, and variables or engine speed: Very Low (<8rpm or Idling speed, Low (8-rpm, Normal (-35rpm, and High (>35rpm, and or engine speed rate: Dropping Quickly (<- 3rpm/s, Dropping Slightly (-3-rpm/s, Stable (rpm/s, Rising Slightly (-3rpm/s, and Rising Quickly (>3rpm/s. A uzzy clutch controller is designed where the clutch pressure is a unction o the slip, slip gain, and uzzy logic rules. Variables or the slip gain o this controller are: Negative (<-., Slightly Negative (<-., Zero (=, Low (.5, Normal (.75, and High (.. The uzzy logic rules or controlling slip gain are as ollows: Throttle Rules: I Throttle is Narrow, or Throttle Rate is Low, then Slip Gain is Low, I Throttle is Normal, or Throttle Rate is Normal, then Slip Gain is Normal, I Throttle is Wide, or Throttle Rate is High, then Slip Gain is High, Stalling Rules: I Engine Speed is Dropping Rapidly or is Very Low then Slip Gain is Negative, I Engine Speed is Dropping Slowly then Slip Gain is Slightly Negative, I Engine Speed is Stable then Slip Gain is Low, I Engine Speed is Rising Slowly then Slip Gain is Normal, I Engine Speed is Rising Rapidly then Slip Gain is High, Driver Intention hange and Engine Speed Rules: Low, then Slip Gain is Slightly Negative, High, then Slip Gain is High, High, then Slip Gain is High, Low, then Slip Gain is Low, High, then Slip Gain is Normal, Low, then Slip Gain is Low, Stationary Rule: I Throttle is closed then Slip Gain is Zero. Figure 4 shows a uzzy logic controller developed in Matlab R9s. This uzzy logic controller compiles all data o the throttle position, throttle rate, engine speed, and engine speed rate with the above uzzy logic rules and then, calculate an engagement gain. The clutch pressure now is a product o slip signal rom the speed sensors in the two riction clutch disks and the gain provided rom this uzzy logic controller. Fig. 4. Fuzzy riction clutch controller The uzzy logic surace graphics or the above slip gain rules are shown in Figure 5-7. Fig. 5. Fuzzy logic surace throttle against throttle rate Fig. 6. Fuzzy logic surace throttle against engine speed Fig. 7. Fuzzy logic surace throttle against engine speed rate IV. LUTH ENGAGEMENT SIMULATION A comprehensive HEV model is developed using Matlab Simulink R9a including all IE, EM, EM, battery, controllable riction clutch, uzzy logic controller, and vehicle dynamic parts. A system control centre is developed to synchronize the activities o all components as shown in Figure 8. 8
6 4 Friction lutch Torque onverter Slip % Vehicle Speed km/h - 3 4 5 6 7 8 9 5 5 3 4 5 6 7 8 9 Fig.. lutch slip and vehicle speed Friction llutch Torque onverter Fig. 8. omprehensive simulation Overall simulations are irstly tested or the hybrid vehicle with an automatic transmission gearbox (Simpson 4 speeds having the gear ratios o Gear :.393; Gear :.45; Gear 3:.; and Gear 4:.677, and compared with the perormance o a torque converter or a rapid ull throttle opening in 3 seconds and then, slow throttle closing. Figure 9 shows that the automatic riction clutch provides aster and smoother gear shiting and leads to a higher vehicle speed due to the higher eiciency o the torque transmission (97% to 86%. But when we release the accelerator pedal and close the throttle, the automatic riction clutch disengaged instantaneously and completely, the vehicle speed is reducing slower than that o the torque converter. That can be used to save more energy or the hybrid vehicle to recharge the battery. Gears Transmission Vehicle Speed (km/h.5.5 5 Gears in Friction lutch Gears in Torque onverter 4 6 8 4 6 8 Friction lutch Torque onverter 4 6 8 4 6 8 Fig. 9. Automatic gear shiting and vehicle speed In igure, hybrid vehicle was tested by maintaining a ully open throttle or seconds. It can be seen that the slip rates or the automatic riction clutch are higher in the starting period due to the torque converter providing a better luid damping. However, when reaching stability, the slip o the torque converter is considerably higher than that o the riction clutch (7% to 3% due to the higher eiciency o dry riction clutch. That leads to a higher vehicle speed or this automatic dry riction clutch over the torque converter. A test result or the perormance o uzzy logic controller is shown in igure where the throttle positions were varied with dierent rates. The throttle is aggressively opened rom -% in 3 seconds, then maintained or 5 seconds, then gradually closed in seconds, then re-opened in seconds, then again maintained or another seconds, and inally closed at a ast rate. The clutch orces are regulated accordingly based on the gain determined rom the uzzy logic rules. Fig.. Perormance o automatic clutch controller Simulations with uzzy logic algorithms show that the clutch controllers can understand well the driver intention. An aggressive pressure on the accelerator pedal or a rapid throttle opening rate leads to a high pressure ramp rate on the clutch engagement or a short time. While a gradual pressure on the accelerator pedal initiates a long and smooth clutch engagement period, A a ast release o the accelerator pedal results to a ast drop o in the engine speed and the ast reduction o the clutch pressure. Finally, the clutch is disengaged with negative pressure gain and the engine is running in idling speed. V. RESULTS OF RESONANE VIBRATION REDUTION Vibration reduction on the clutch transmission can be achieved by increasing the system damping and/or adjusting the clutch pressure. This is a description o the experimental results rom a real vehicle tested with an automatic dry riction clutch in a Honda ivic o.6l gasoline engine and 5-speed automation manual transer (AMT, where the clutch pressures are controlled and adjusted via a microprocessors and PID controller as shown in igure.. The critical vibration is ound at the engine speed o 86 rpm or at requency o 3 Hz when passing through the respective load ranges (Figure. At this critical requency, the clutch pressure is regulated (slightly released in order to vary the clutch slip rom -5%. A considerable reduction o the critical vibration is observed when the clutch slip is released at a value o.5% as seen in igure 3. Figure 4 indicates that the completed clutch engagement is achieved at approximately.7 seconds resulting in the drive shat torque being suiciently smooth with no oscillations or sudden drops. It is noticed that the controller is very sensitive to variations o the clutch riction coeicient and the clutch pressure. Any reduction o vibrations in the 9
clutch pressure will lead to the reduction in vibrations o the transmission torque resulting in the clutch ailing to lock-up in steady-state. Vibration Measurement 5-5 Vehicle Vibration Measurement 5 5 5 3 35 4 45 5 Time Millisecond (ms compensated on the clutch lie when increasing the slip value. However, the increase in uel consumption due to this application has not been determined yet. Spectrum Y(.5 Single-Sided Amplitude Spectrum 5 5 5 3 35 4 45 5 Frequency (Hz Fig.. Amplitude Spectrum o Vehicle Noise Measurement Fig. 6. Perormance o clutch with vibration eliminators In critical high requencies, the clutch with these vibration eliminators can introduce higher levels on comort. For this special clutch, the elastic coupling o additional disks achieves an extensive elimination o the emergence o resonance. The main disadvantages o this clutch are higher expense and larger space. Thereore, this kind o clutch can be applied only or the rear-wheel driven HEVs due to the limited available space or the ront-wheel drive vehicles. The more complex clutch also leads to the higher maintenance cost. Fig. 3. Speed and slip in the clutch at critical requency Fig. 4. lutch engagement and transmission torque Another solution to deal with the critical vibrations is to raise the system damping or to use the vibration eliminators. In this experimental test, some vibration eliminators with special design or elimination o the critical vibration amplitudes are employed (Figure 5. In this simple design, disks were added with higher damping characteristics to absorb the vibrations. Fig. 5. lutch with vibration eliminators The vibration eliminators can reduce 8% o vibration amplitude or high requencies above 6 Hz (Figure 6. Using new linings o sintered metal, ceramics, or ibrereinorcement plastics, the negative eects can be VI. ONLUSIONS The uzzy logic control rules or an automatic riction clutch o hybrid vehicle have been developed and tested. The new system can control the slip and oer ast engagements in low jerk and high comort. Experimental results show that regulating the clutch pressure and increasing the clutch damping characteristics can achieve the vibration reduction in critical requencies. The paper has oered useul contributions to the development o HEVs. The control schemes can be used in electronic control units or real HEVs applications. AKNOWLEDGMENT This research was supported by the Department o Mechatronics, Faculty o Mechanical Engineering, Tallinn University o Technology. REFERENES [] Minh VT. lutch control and vibration reduction or a hybrid electric vehicle, Journal o Systems and ontrol Engineering, ISSN: 959-658, accepted. [] Schouten NJ, Salman MA and Kheir NA. Fuzzy logic control or parallel hybrid vehicles, IEEE Trans ontrol Systems Technology ; (3: 46 468. [3] Kim TS, Manzie and Sharma R. Model predictive control o velocity and torque split in a parallel hybrid vehicle, Proceedings o the IEEE international conerence on systems, San Antonio, TX, October 9, ISSN: 6-9X, pp.4 9. [4] Paganelli G, Delprat S, Guerra TM, et al. Equivalent consumption minimization strategy or parallel hybrid powertrains, Proceedings o the IEEE vehicular technology conerence, Taipei, Tailwan, May, ISSN: 55-5, pp.76 8. [5] Sciarretta A, Back M and Guzzella L. Optimal control o parallel hybrid electric vehicles, IEEE Trans ontrol Systems Technology 4; (3: 35 363. [6] Minh VT, Hashim FB and Awang M. Development o a real-time clutch transition strategy or a parallel hybrid electric vehicle. Journal o Systems and ontrol Engineering,, vol. 6(, pp. 88-3. 3