A Speed Control for the Reduction of the Shift Shocks in Electric Vehicles with a Two-Speed AMT

Abstract

In the present paper, a speed control algorithm with fast response characteristics is proposed to reduce the shift shock of medium/large-sized electric vehicles equipped with a two-speed AMT. Shift shocks, which are closely related with to the vehicles' ride comfort, occur due to the difference between the speed of the motor shaft and the load shaft when the gear is engaged. The proposed speed control method for shift shock reduction can quickly synchronize speeds occurring due to differences in the gear ratios during speed shifts in AMT systems by speed command feed-forward compensation and a state feedback controller. As a result, efficient shift results without any shift shock can be obtained. The proposed speed control method was applied to a 9 m- long medium- sized electric bus to demonstrate the validity through a simulated analysis and experiments.

I. INTRODUCTION

Due to the ongoing global depletion of fossil fuels and regulations on vehicle exhaust gases, environmentally friendly vehicles and related components are being actively studied and developed in the automobile industry. Environmentally friendly vehicles can be largely divided into Hybrid Electric Vehicles (HEVs) and Electric Vehicles (EVs). First, HEVs refer to vehicles driven using two or more types of power sources. Although the combinations of power sources are diverse, systems that drive vehicles using internal combustion engines and electric motors are the most common. EVs generally refer to those vehicles that are driven using purely electric energy. EVs are driven by changing chemical energy from storage batteries into electric energy and driving electric motors with electric energy. Therefore, EVs are advantageous in that no exhaust gas or environmental pollution is generated during operation and noise output is minimal. EVs are largely divided into Battery Electric Vehicles (BEVs) and Fuel Cell Electric Vehicles (FCEVs) depending on the type of storage batteries [1].

As shown in Fig. 1, EVs are generally composed of electric motors, inverters for controlling the electric motors, batteries that supply power to the inverters, and decelerators to deliver power to the vehicle wheels. Decelerators, which are one of the core components of EVs, are devices that change the motor outputs to fit the running performance characteristics of vehicles and deliver these outputs to the wheels. In general, small/medium-sized EVs, such as NEVs or golf cars, use only decelerators with a single gear [2]. This is because the performance of the electric motors applied to small/medium-sized EVs is similar to the low speed high torque and the high speed low torque driving output performance required by the vehicles. Therefore, small/medium-sized EVs have multi-speed transmissions, as with internal combustion engine vehicles. However, as the development of EVs has expanded to include medium/large-sized vehicles, such as buses, the application of multi-speed transmissions to EVs is being expanded, as with existing engine vehicles [3]-[5].

Fig 1.Block diagram of major components of electric buses.

Vehicle transmissions can be largely divided into Manual Transmissions (MTs), Auto Transmissions (ATs), Continuously Variable automatic Transmissions (CVTs), and Automatic Manual Transmissions (AMT). Because MTs are cumbersome, due to the fact that drivers should shift them firsthand, most vehicles are applied with ATs. However, in general ATs are expensive and have problems with fuel efficiency and acceleration performance because their power transmission efficiency is low. The ATs of existing vehicles cannot be simply applied to EVs. On the other hand, although CVTs have many advantages, they are not suitable for the transmission of the large amounts of power required by medium/large-sized vehicles, such as electric buses, because they are applied with belts [6], [7]. Therefore, attention to the development of AMTs is increasing, since AMTs have the convenience of ATs while maintaining high fuel efficiency and acceleration performance, which are the advantages of existing MTs. Accordingly, studies have been conducted on the application of two-speed AMTs to electric buses [8], [9]. Two-speed transmissions for EVs have been studied in relation to their diverse mechanical structures, as well as their energy efficiency and driving performance [10]. However, unlike existing single decelerators, when two-speed transmissions are applied, shift shocks occur during speed shifts.

Therefore, in the present study, a speed control algorithm is proposed that has speed shift sequences, while considering fast shifts from the driving modes to the speed shift modes and fast response characteristics. The performance of the algorithm is verified through simulations and experiments.

 

II. TWO-SPEED TRANSMISSION SYSTEMS FOR ELECTRIC VEHICLES

A. Two-Speed AMT Systems for Electric Buses

Fig. 2 shows comparisons of the speed torque characteristics and speed output power characteristics of electric motors and internal combustion engines. As can be seen from Fig. 2, unlike the output performance of engines, the output performance of motors is similar to the required load characteristics of vehicles, since the torques of the motors are high at low speeds and low at high speeds. This enables driving in wide speed ranges. Therefore, whereas internal combustion engine vehicles are generally applied with multi-speed transmissions, EVs equipped with an electric motor can be sufficiently driven when equipped with two-speed transmissions.

Fig. 2.Comparison between the performance of engines and that of motors.

Fig. 3 shows a speed torque performance curve of vehicles equipped with two-speed transmissions. As can be seen from Fig. 3, if two-speed transmissions are applied to electric buses, outputs can be obtained in wider ranges of speed and torque than those obtained with single decelerators.

Fig. 3.Performance of vehicle applied with a two-speed transmission.

Fig. 4 shows the two-speed AMT system 3D model applied in the present study. The two-speed AMT system consists of a Gear Shift Unit (GSU), a Transmission Control Unit (TCU), a Motor Control Unit (MCU), and a transmission that enables automatic transmissions based on a manual transmission. The GSU is in charge of the transmission gear connections and disconnections. The gear connection and disconnection operations are performed through hydraulic pressure and solenoid valves.

Fig. 4.Block diagram of a two-speed AMT system.

The solenoid valve operations are controlled by the TCU and the TCU is equipped with communication functions to check the states (temperature, pressure, flow rate, etc.) of the transmission and exchange information with the other control systems of the vehicle, as well as I/O to exchange digital signals. The MCU is in charge of motor control and shift control. It also calculates the optimum shift time points according to the driver’s will to shift speed and the vehicle driving information through the vehicle information (vehicle speed, throttle opening level, shift lever position). This is done to send shift commands to the TCU and to control the motor.

The two-speed AMT gear applied in the present study is a Simpson planetary gear type, while multi-plate clutches and a hydraulic control are applied to the shift control. The basic components of planetary gear types are sun gears, planetary gears, carriers, and ring gears. Unlike general gears that revolve around a fixed shaft at a certain rotation ratio, planetary gear types refer to gear systems based on structures where planetary gears around a sun gear are supported by a carrier to revolve around the sun gear. In this case, the sun gear rotates and the planetary gears rotate and revolve.

The revolutions of the planetary gears lead to movements of the carrier. Fig. 5 shows block diagrams of planetary gears according to the number of speeds in a two-speed AMT. In this structure, the power is applied to the sun gear, transmitted to the two sets (1st gear, 2nd gear) of planetary gears arranged on a single carrier, and outputted through the carrier.

Fig. 5.1st and 2nd planetary gear unit.

B. Mathematical Model of Two-Speed AMT Systems

To shift two-speed transmissions, a clutch configuration is necessary to cut power to the gear terminal currently connected and to apply power to the terminal to be shifted to. A multi-plate clutch was applied. It consists of multiple plates so that large torques can be transmitted as the power is distributed to multiple wear plates and delivers torques with the frictional force generated when pressure is applied. This multi-plate clutch controls torque transmissions through the carrier by controlling the revolutions of the planetary gears. The reduction gear ratios of the 1st and 2nd gears can be calculated from Table I and equation (1) [11].

TABLE IDATA OF REDUCTION GEARS

where, Zs is the number of teeth of the sun gear and ZR is the number of teeth of the ring gear.

As shown in Fig. 6, during the 1st gear and 2nd gear driving of two-speed AMT systems, the motor shaft is connected to the transmission shaft according to the structure of the planetary gears in the transmission. In addition, when the clutch is disconnected for shifts, the vehicle load is discharged. The motor driving system for electric buses equipped with two-speed AMTs can be divided into a motor, a transmission, and a vehicle load. The power is transmitted through the driving shaft. This system can be approximated as shown in Fig. 7. The vehicle load, seen from the side of the motor, is recognized as a load transmitted to the motor shaft through the driving shaft and transmission. This is shown by equation (2). Therefore, the equation of the motion of a vehicle load driving systems using motors can be expressed as shown by equation (3).

Fig. 6.Shift gears mechanism.

Fig. 7.Simplified models of powertrain.

where, TC refers to the load torque generated by the wheel set, ωL refers to the rotating speed of the wheel set, JL refers to the inertia moment of the wheel set, and BL refers to the coefficient of the rotational frictional force of the wheel set. In addition, TM refers to the load torque generated by the electric motor shaft, ωM refers to the rotational speed of the electric motor shaft, JM refers to the inertia moment of the electric motor shaft, and BM refers to the coefficient of the rotational frictional force of the electric motor shaft. When the travel ranges of the motors and the vehicles in vehicle driving systems are the same regardless of the gear ratios (N: 1), according to the number of speeds, equation (4) is valid. In addition, if gear losses and inertia are ignored, the magnitudes of the power at both sides of the transmission are the same. Therefore, to control the motor effectively, the vehicle load system can be converted into the torque and speed of the motor, as shown by equation (5).

Although the motor outputs are connected to the transmission outputs according to the structure of the planetary gears in the transmission while the vehicle is driving with the 1st or 2nd gear, as shown in Fig. 6, the vehicle load is disconnected when the clutch has been disconnected for shifts.

Therefore, TM while is shown by equation (5) during driving, it is shown by equation (6) while the transmission gear is being shifted because only IM and BM exist during gear shifting.

C. Two-Speed AMT Automatic Shift Sequence

As mentioned earlier, if two-speed transmissions are applied to electric buses, outputs can be obtained in wider speed and torque ranges than with single decelerators. However, unlike existing single decelerators, when two-speed transmissions are applied, shift shocks occur during shifting. That is, shift shocks occur because the speed of the motor shaft and the speed of the load shaft attached to the clutch are not identical during gear connections. Although the shocks occurring due to speed differences are absorbed by using torque converters in the case of internal combustion engines, applying torque converters is not reasonable in the case of EVs because efficiency is an important element for EVs. Therefore, shift shocks should be reduced through synchronization of the speed of the motor or the engine. In addition, shifting should be conducted quickly to maintain the acceleration of the vehicle and for the driver’s comfort. Since shifting quality is directly related to the ride comfort of vehicles, reducing shift shocks and shifting time is the very important [12], [13].

A key to fast shifting is fast synchronization of the speed of the electric motor following changes in gear ratios, and if the speed is not sufficiently synchronized, shift shocks will occur [14]. The shift sequence in existing manual transmissions is as follows.

The vehicle is driven by acceleration through the acceleration pedal at normal times, and the speed is shifted through differences in the rev counts of the gears while relying on the driver’s operation using the clutch pedal when the speed is changed. Thereafter, the vehicle is driven again through the acceleration pedal, reflecting the driver’s will. However, in the case of the two-speed AMTs of EVs, drivers control only the acceleration pedal and the speed is automatically shifted by the driving signals from the MCU according to the vehicle conditions.

Fig. 8 shows automatic gear shift sequences, where (a) shows the case of a driver continuously pressing the acceleration pedal to accelerate the vehicle and (b) shows a case where the vehicle is decelerated.

Fig. 8.Gear Shift Sequence.

1) Section A: This is a general driving section where the driving motor generates torques reflecting the driver’s will following the operation of the accelerator. As a result, the vehicle speed is accelerated or decelerated according to the increases or decreases in the torque.

Section B: When a point where gear shifting is possible has been reached, the MCU controls the motor torque to ‘0’ to cut the power transmission and request a disconnection from the gear at the currently connected terminal. When the signals indicate that the gear has been disconnected, the MCU begins motor speed control to synchronize the speed determined according to the gear level and the vehicle speed. When the desired motor speed has been reached, the MCU requests the connection of the gear at the desired level.

2) Section B: This is a section where the motor speed is maintained until the signals indicate that the gear has been connected. If the speed is not smoothly synchronized, shift shocks will occur. When the signals indicating the completion of gear connection have been received, this section is converted into Section C.

3) Section C: This is a point where the gear connection has been completed. As with Section A, the MCU obtains torque values from the driver’s accelerator operation to control the motor.

D. Speed Controller for Speed Synchronization

The speed synchronization methods to reduce shift shocks are largely divided into those that use torque control and those that use speed control.

The control methods using torque perform accelerating operations in the area of 1 upper limit during up-shifts, and perform decelerating operations through braking operations in the area of 2 upper limit during down-shifts. However, shift shocks are expected because accurate speed synchronization in a short time is not easy, since the gears are shifted through torque control [15]. Therefore, speed control is applied as a method for speed synchronization in the shifting sections. Although existing internal combustion engine vehicles equipped with existing AMT have been applied with speed control, since this speed control is a simple PI control of the amount of throttle valve opening, it cannot be applied to EVs as is [16], [17]. In addition, although speed control through speed command feed-forward compensation was proposed in the case of EVs applied with an AMT, the controller design is difficult because there is no concrete mention of gains from the feed-forward compensation, and single loop controllers, where the driving voltage is the speed controller outputs, were applied instead of general dual loop controllers [18]. Although the speed differences in normal states are important, shift shocks are not much different unless the speed errors are larger than a certain extent. On the other hand, fast responses of the speed controls are more important because drivers’ emotions are important.

Fig. 9 shows a general speed control method that performs dual loop control. Each dual loop controller has a speed controller on its outside and a current controller in its inside.

Fig. 9.Block diagram of general speed control method.

The speed controller generally creates torque commands or current commands through PI control. When a torque command/current command has been created, the current is controlled through the current controller. To complete the clutching operation of the AMT without any shift shock, speed responses should be given within a determined time without any overshoot. In addition, satisfactory performance cannot be obtained using existing general dual loop control methods. To solve this problem, a speed controller with speed command feed-forward compensation and speed feedback processes is proposed.

Fig. 10 shows a block diagram of the proposed method. The proposed method compensates for the sum of the feed-forward compensation value, as calculated using a given speed command and the motor inertia, while the motor frictional force is determined according to the actual speeds to the PI speed controller output. There is no problem for the amplification of noise since it is applied to the method of feed-forward compensating for the value multiplied by the inertia to differential of the speed command. Therefore, the final torque command T* is given as shown by equation (7).

where:

Fig. 10.Block Diagram of proposed speed control method.

Equations (8), (9), and (10) refer to the electric motor speed commands and the actual electric motor speed. From the torque commands, the current commands are given with the following equations.

Meanwhile, if it is assumed that the current of the output torque of the electric motor is well controlled by the internal loop, the output torque of the electric motor follows the torque commands. If it is assumed that electric motor output torques follow the torque commands, the dynamic characteristics of the electric motor errors will be given from equations (2), (8), (9), and (10), as equation (13). Meanwhile, the speed responses can quickly follow the speed commands without any overshoot.

The transfer function of the proposed method may be obtained from Path1 and Path2. The closed loop transfer functions of Path1 and Path2 are as shown by equation (14) and (15). Therefore, the transfer function of the proposed method is sum of Path1 and Path2. This is the same as equation (16).

However, actual systems must take into account the parameters of the load. The parameters for high-performance control systems, which are the inertia and friction forces, must be applied to compensate. In particular, the friction force easily changes its characteristics due to changes in the surrounding environment such as temperature or humidity.

First, if it exists that the error of the J for the actual load system and the J* for the aware value and the error rate is referred to as ‘K’, a bode plot for equation (16) is shown in Fig. 11. As a result, it can be seen that there is a problem at low speeds when the vehicle is nearly stopped. However, it is not a problem to apply the proposed control method, because there is no relation between the pattern of the gear shift and low speeds near ‘0’. Therefore, unlike general dual loop controllers, the response characteristics of the proposed method follow the speed command in most frequency bands.

Fig. 11.Bode plot of speed control.

In addition, the transfer function of the block in Fig. 10 is shown in equation (17). At this time, if it exists that the error of the B for the actual load system and the B* for the aware value, the transfer function of block is the same as equation (18).

Meanwhile, if there is a change in the speed command for shifting, the steady-state error disappears as equation (19). However, if it exists that the error of the B for the actual load system and the B* for the aware value, the steady-state error exist. Therefore, it is important to accurately estimate the frictional force.

The frictional force that must be considered when the shifting of a two-speed AMT is a viscous friction. The frictional force of the transmission may increase the estimate error according to the load condition. In addition, estimating a confidence value is very difficult, due to the surrounding environmental factors such as the motor connection type, backlash of the gear, temperature and humidity [19][20][21]. In particular, continuous operation of a vehicle increases the oil temperature of the transmission. Therefore, as shown in Fig. 12, the viscous friction force is reduced. In addition, if a vehicle left in the cold weather is operated, the viscous friction force will be high due to the low temperature of the transmission oil.

Fig. 12.Oil Temp. vs Friction Torque.

Fig. 13 shows simulation results of the response speed of the proposed speed controller in accordance with the error of a prediction the viscous friction force. If the real friction is lower than the predicted friction, a large overshoot occurs. In addition, if the real friction is higher, the response is delayed.

Fig. 13.Speed Response for friction force error.

Therefore, in the present study, this was compensated with a more reliable frictional force by using the AMT oil temperature information in real time. The compensation method is the same as equation (20).

 

III. SIMULATION AND EXPERIMENTAL RESULTS

In the present study, to show the validity of the proposed method, a simulated analysis was conducted using Matlab/Simulink and automatic shift tests were conducted through driving a medium-sized 9 m electric bus. In the simulated analysis, the speed controller control period was set to 10 m/s, which is identical to actual systems. In addition, the gains of the current controller were set to be sufficiently large so that ideal current control characteristics can be obtained. Furthermore, to examine the characteristics of the speed responses to speed commands, simulated analyses were conducted using the shift patterns applied to actual electric buses.

A. Simulation Results

Fig. 14 shows the results of simulated analyses of the speed response characteristics obtained when a general dual-loop control speed control method and the proposed method were applied when an electric bus was accelerating and decelerating.

Fig. 14.Compare of speed response.

In the ‘Torque Control’ section, the accelerator values are applied as torque values; and in the ‘Speed Control’ section, the speed that fits the gear to be shifted to and the vehicle speed are used as command values. The speed commands are made using Ramp inputs to reduce shocks and noises during shifting. Figs. 14 (a) and (b) show characteristics graphs of the gear-up shift section during acceleration and the gear-down shift section during deceleration, respectively.

Upon reviewing the response characteristics of general PI speed controllers, it can be seen that the response characteristics are not good because the controllers cannot follow Ramp commands and overshoots occur at points where normal states are reached. Although shift shocks may occur due to shift time delays in such cases, when the proposed method is used, excellent speed response characteristics can be obtained. Thus, shift shocks can be reduced thanks to the fast speed synchronization.

TABLE IISIMULATION CONDITIONS

B. Experimental Results

Fig. 15 shows the outer appearances of a manufactured all-in-one module comprising a manufactured two-speed AMT, a driving inverter, and a driving motor, along with a medium-sized bus equipped with this module. Table III shows the specification of the medium- sized electric bus.

Fig. 15.Integrated traction system and electric bus.

TABLE IIIVEHICLE SPECIFICATION

Acceleration driving was conducted on actual roads. To check the control performance, the vehicle speed, vibration, and CAN data were obtained using MCU data and Data Acquisition System (DAS; Model: CL-7016) equipment of Imc Co., as shown in Fig. 17. Fig. 16 shows the vehicle speed and vibration values measured using the DAS and the inverter state information received from the CAN data. The inverter state information includes the torque command values, motor speeds, shift commands, and shift states. The sampling times of the GPS, inverter state information, and vibration values are 200 m/s, 100 m/s, and 10 m/s, respectively.

Fig. 16.Torque reference, motor RPM, speed, transmission state and vibration value of electric bus.

Fig. 17.Integrated traction system and electric bus.

Fig. 16(a) shows experimental waveforms when a general PI speed controller was applied, and Fig. 16(b) shows the resultant waveform when a general PI speed controller was applied. The shift operation was performed when a shift command was given at the same time as the proposed compensation speed controller. Fig. 16(c) shows the resultant waveform when the proposed compensation speed controller was applied. The experimental results are shown to be different from simulation results in terms of the gear shifting point. This is due to the inertia loads and frictional forces of the transmission due to the integrated system architecture.

A comparison between the results of a general controller and those of the proposed controller are given in Table IV.

TABLE IVDIFFERENCE OF RESULTS

First, based on a comparison between the shift times in Fig. 16(a) and Fig. 16(c), it can be seen that the shifting time was shortened when the proposed speed controller was applied compared to when a general PI speed controller was applied, since the shifting was shortened from 2.87s to 1.89s. In addition, if a general PI speed controller is applied and the gear is shifted in the same shifting time as that in Fig. 16(c), shift shocks occur, as shown in Fig. 16(b), due to the fact that speed differences occur. It can be seen that shift shocks occur when the two-speed gear is connected and the shift shocks are approximately two times the shock value in Fig. 16(c), which is -0.00045g. Therefore, it can be seen that the speed response characteristics of the proposed compensation speed controller are better than those of existing PI speed controllers.

 

IV. CONCLUSIONS

In the present study, a new compensation speed control algorithm was proposed for shift sequences. It is suitable for AMT systems and fast speed synchronization during shifting to reduce the shift shocks in vehicle systems equipped with two-speed AMT. This algorithm feedback compensates for the feed-forward compensation calculated using speed commands given to general PI speed controller outputs as well as the motor inertia and motor frictional force according to actual speeds. Since the proposed compensation speed control algorithm can obtain better speed response characteristics than general PI speed controllers, it can quickly synchronize speeds during shifting. The foregoing conclusions were verified through simulations, where the proposed shift sequence and speed controller were applied to a medium-sized electric bus. In addition, the validity of the proposed method was verified through actual driving.