An efficiency control strategy of dual-motor multi-gear drive algorithm

Abstract

The Dual-motor multi-gear coupling powertrain (DMCP) has the potential to improve transmission system efficiency and driving comfort, but its complex structure and multiple working modes present challenges. The switching between different modes is easy to cause longitudinal biggish vehicle jerk. To address these issues,this paper introduces the Deep Deterministic Policy Gradient (DDPG) algorithm in the design of an Energy Management Strategy (EMS) that minimises total drive power consumption. And the number of working modes is divided and simplified. The process of switching dual motor and single motor to single motor is introduced in detail. The simulation results using AMESim and MATLAB show that the energy management strategy can effectively improve the economy, achieve no power interruption during mode switching, shift impact is less than 8m/s3, and output torque is remains stable.

Keywords:

• Dual-motor
• multi-gear
• energy management
• no power interruption
• mode-shift

1. Introduction

Most pure electric vehicles (Xing et al., 2022) currently use single motor single speed ratio drive system, which has the advantages of simple structure, low noise, and low cost (Gerssen-Gondelach & Faaij, 2012), but there is still a problem of low efficiency of the drive train. Multi-speed ratio drive train is an effective way to improve the efficiency of the drive train (Bottiglione et al., 2014). Conventional mechanical automatic transmissions (AMT) experience power interruptions during gear shifts, resulting in poor ride comfort and sudden changes in vehicle travel (Diao et al., 2022). As a result, traffic flow prediction is considered a key issue in Intelligent Transportation Systems (ITS), and Graphical Convolutional Neural Networks (GCN) has been effectively used for traffic prediction due to their excellent performance in spatial correlation modeling (Diao et al., 2022; Hu et al., 2022). A dual motor multi-speed coupled drive system is a new AMT system (Shilei, 2018; Zhiyao, 2017; Zhu et al., 2013) that can solve the power interruption problem. Often ride comfort or economy of the vehicle contradict each other (Miro-Padovani et al., 2015), to improve the economy comfort, an energy management strategy torque control strategy is essential (Jiang et al., 2019). The research on energy management of multiple power sources by domestic and foreign scholars mainly focuses on the engine-motor hybrid system, which treats the motor as an auxiliary power source and is primarily used to improve the efficiency of the engine (Wang et al., 2017), so it does not apply to the dual-motor pure electric vehicle; the energy management strategy mainly adopts the rule-based logic threshold strategy (Shuo et al., 2014), which has fewer working modes and single management rules and cannot be applied to the complex drive combination of the dual-motor pure electric vehicle.

With the development of artificial intelligence technology, reinforcement learning-based energy management strategy are able to achieve near-global optimal solutions without any prior knowledge by training an agent.The agent is trained to interact with environment (vehicle) state and improve the action (power allocating) by proper rewards (energy consumption), which makes the optimization of energy management strategy practical in real-time. Sorniotti et al. (2012) proposed a two-speed transmission without power interruption and investigated its control strategy to obtain better power and economic performance. Liang et al. (2018) proposed a new clutchless dual-input automatic transmission in which the motor compensates the power in time during shifting, improving driving comfort while reducing shift shock compared to conventional automatic transmissions. This improves driving comfort and reduces shift shock compared to traditional automatic transmissions. Yang et al. (2016) proposed a real-time torque distribution strategy for pure electric vehicles with multiple power sources, and the simulation showed that the real-time torque distribution strategy was close to the dynamic planning results. Hu et al. (2017) proposed a set of single-motor to dual-motor torque drive switching strategies based on the dynamics of the components during the dual-motor mode switching process to minimise the shift shock. Hu et al. (2015) proposed a torque-compensated shifting strategy based on the analysis of torque fluctuation before and after shifting of two-speed automatic transmission. Zhang et al. (2016) used a planetary gear dual-motor coupled configuration and proposed a neural network-based online torque distribution scheme. Li et al. (2018) used a dual-motor planetary gear coupling, determined the operating intervals of four operating modes based on the principle of minimum energy consumption, and developed a dual-motor speed-torque distribution strategy table. Muduli et al. (2021) presents a new open-end winding induction motor (OEWIM)-based dual-motor differential four-wheel drive (D4WD) for the electric vehicle (EV). Constant speed operation through cruise control is achieved using direct torque control (DTC) algorithm. Kai Wen et al. (2022)proposed an innovative design and verification method for a dual-motor power-coupled drive system (DMPCS), covering the whole process of configuration design, parameter optimization and system verification of a pure electric tractor drive system. Zhao et al. (2019) proposed a new systematic extraction method for energy management strategies by designing explicit shift schedule lines using a nonlinear support vector machine (SVM) and a new effective factor adjustment method.However, the demand torque distribution in a dual-motor system is still an important issue. Dynamic Programming(DP) algorithm can be used to find the optimal control policy for the energy management problem. However, DP cannot explore the global optimal strategy without complete prior knowledge of the future driving cycle. The algorithm usually is used as a benchmark. In recent years, artificial intelligence technologies, which are represented by deep learning and deep reinforcement learning, have gained tremendous focus to solve energy coupling optimization problems.

To improve the efficiency and performance of automotive drive systems, reduce energy consumption and emissions, enhance driving comfort, promote the development of the automotive industry, and increase the competitiveness of automobiles.In this paper, the optimal output strategy of DMCP is investigated to achieve the lowest energy consumption at steady-state drive by using the dual-motor multi-speed coupled drive system (DMCP) shown in Figure 1. There is no power interruption during mode switching. By introducing the structure and drive mode of the DMCP system, the research focuses on the division of the lowest drive power consumption mode distribution, the real-time drive mode selection method and the dual-motor torque distribution strategy, and the power interruption-free switching strategy during the dual-motor mode switching.

Figure 1. Dual motor multi-speed coupling drive system.

The remainder of this paper is organised as follows. Chapter 2 introduces the structure of the two-motor multi-speed coupled drive system, Chapter 3 shows the real-time torque distribution strategy, Chapter 4 introduces the no-power interruption mode switching strategy, Chapter 5 uses the AMESim platform to build the physical model of the system, and MATLAB/Simulink is used to build the control strategy and perform the joint simulation. Chapter 5 is a summary of the paper

1.1. contributions

The contribution of this article is as follows, which addresses the issues of complex system structure and conflicting work modes in a dual-motor multi-gear drive system: (1) This article designs a real-time energy management strategy and divides and simplifies the working modes to minimise the total driving power while ensuring the vehicle dynamics. (2) A dual-motor power interruption mode switching control strategy is proposed to solve the problem of difficult switching between multiple driving modes.

2. Dual motor multi-speed coupling drive system structure

As shown in Figure 1, the structure has two power sources M1 (motor 1) and M2 (motor 2). M1 is connected to the first shaft of the gearbox, and the gears of M1-1 (1st gear of motor 1) and M1-2 (2nd gear of motor 1) are shifted through synchroniser 1; M2 is connected to the second shaft of the gearbox, and the gears of M2-1 (1st gear of motor 2) and M2-2 (2nd gear of motor 2) are shifted through synchroniser 2; M1 and M2 are torque coupled through the output shaft of the gearbox and then output to the outside. By controlling the position of synchroniser 1 and synchroniser 2 engagement sleeve and the opening and closing of M1 and M2, DMCP can realise eight operating modes, which are M1-1 driving mode (SM1-1), M1-2 driving mode (SM1-2), M2-1 driving mode (SM2-1), M1-1 and M2-1 coupled driving mode (DM1-1), M1-2 and M2-1 coupled drive mode (DM2-1), M1-2 and M2-2 coupled drive mode (DM2-2), M1-1 and M2-2 coupled drive mode (DM1-2), and the states of the motor and synchroniser in each mode are shown in Table 1.

Table 1. Operating status table of motor and synchroniser in each mode.

Mode	Motor 1	Motor 2	Synchronizer 1	Synchronizer 2
SM1-1	•	⊖	$\to$	$\oslash$
SM1-2	•	⊖	$\leftarrow$	$\oslash$
SM2-1	⊖	•	$\oslash$	$\to$
SM2-2	⊖	•	$\oslash$	$\leftarrow$
DM1-1	•	•	$\to$	$\to$
DM2-1	•	•	$\leftarrow$	$\to$
DM2-2	•	•	$\leftarrow$	$\leftarrow$
DM1-2	•	•	$\to$	$\leftarrow$

Note: “•” means the corresponding motor is in starting state, “⊖” means the corresponding motor is in stopping state, “$\to$” means the corresponding synchroniser's bonding sleeve and right side bonding, “$\leftarrow$” means the corresponding synchroniser's bonding sleeve left side bonding, “$\oslash$” means the corresponding synchroniser's bonding sleeve empty sleeve.

The relevant parameters of the whole vehicle are shown in Table 2.

Table 2. Overall vehicle parameters.

Parameters	Numerical value
Overall vehicle mass m(kg)	1780
Wheel rolling radius r(m)	0.317
Rolling resistance coefficient f	0.015
Windward area A(m)	2.1
Air resistance coefficient Cd	0.3
Car rotating mass conversion factor δ	2.1
Maximum speed(Km/h)	180
Battery voltage(V)	372
Battery Capacity(Ah)	72
Battery power(KWh)	26.78
M1/M2 peak power(KW)	23.5/61.5
M1/M2 peak torque(Nm)	60/170
M1/M2 rated speed(rpm)	3740/3455
M1/M2 maximum speed(rpm)	11500/9500
M1-1 /M1-2(i1_1/ i1_2)	1.667/1
M2-1/ M2-2(i2_1/ i2_2)	1.2558/0.7378
Main reduction ratio(i0)	7

Figure 2 shows the external characteristic curves for each operating mode corresponding to Table 1. When the driver demands high torque, DMCP can output the demanded high torque with dual motor coupling. As the torque distribution ratio of the two motors is variable, there is always a torque distribution ratio that minimises the total drive electric power; when the electric vehicle drives with low torque at low speed, DMCP can output the demanded torque with a single motor alone, thus reducing the backup power of the vehicle in this mode and improving the economic performance of the whole vehicle.

Figure 2. External characteristics of the drive system.

3. Real-time torque distribution strategy (PSC)

The two drive motors of DMCP, motor M1 is called economy motor, and its motor efficiency is shown in Figure 3. Motor M1 is mainly used to drive alone at low speed and small torque demand. Motor M2 is called a power motor, and its efficiency MAP diagram is shown in Figure 3(b), which is mainly used to compensate for the power of motor M1 to meet the power requirements of the whole vehicle design.

Figure 3. MAP diagram of economic motor and power motor. (a) Economic motor MAP diagram (b) Power motor MAP diagram.

To fully utilise the energy-saving potential of the DMCP, it is necessary to select the lowest power consumption drive method while satisfying the dynamics. (1) $T{\mathrm{\_}}_{req}=T_{M1}{i}_1+T_{M2}{i}_2$(1) $i\mathrm{\_}1$ is the first gear speed ratio of the first axis, $i1\mathrm{\_}2$ is the second gear speed ratio of the first axis, $i21$ is the first gear speed ratio of the second axis, and $i22$ is the second gear speed ratio of the second axis.Where $i_1$ represents $i1\mathrm{\_}1$ or $i1\mathrm{\_}2$; $i\mathrm{\_}2$ represents $i2\mathrm{\_}1$ or $i2\mathrm{\_}2$; $T\mathrm{\_}rep$ is the vehicle demand torque; $T\mathrm{\_}M1$ is the target torque of motor M1 and $T\mathrm{\_}M2$ is the target torque of motor M2. The demand torque $T\mathrm{\_}rep$ can be obtained from the driver model, and (2(2) $T_{M2}=\frac{T{\mathrm{\_}}_{req}-T_{M1}{i}_1}{i_2}$(2) ) can be obtained from (1(1) $T{\mathrm{\_}}_{req}=T_{M1}{i}_1+T_{M2}{i}_2$(1) ) as follows: (2) $T_{M2}=\frac{T{\mathrm{\_}}_{req}-T_{M1}{i}_1}{i_2}$(2) The speed of the motor can be obtained by converting the speed of the vehicle, (3) $\underset{\mathrm{M}1}{\overset{\bullet }{\theta }}=\frac{\overline{v}{i}_0{i}_1}{0.377\mathrm{r}},\underset{M2}{\overset{\bullet }{\theta }}=\frac{\overline{v}{i}_0{i}_2}{0.377\mathrm{r}}$(3) $\underset{\mathrm{M}1}{\overset{\bullet }{\theta }}$ and $\underset{M2}{\overset{\bullet }{\theta }}$ indicate the speed of M1 and M2, $v^{-}$ indicates the speed unit of Km/h, $i_0$ is the primary reduction ratio, and r denotes the wheel radius.

Total drive motor power consumption $P_e$: (4) ${\mathrm{P}}_{\mathrm{e}}=\frac{T_{\mathrm{M}1}{{\theta }^{\bullet }}_{\mathrm{M}1}}{{\eta }_{\mathrm{M}1}}+\frac{T_{\mathrm{M}2}{{\theta }^{\bullet }}_{\mathrm{M}2}}{{\eta }_{\mathrm{M}2}}$(4) where ${\eta }_{M1}$ and ${\eta }_{M2}$ denote the efficiency obtained by interpolating M1 and M2 under $T_{M1}$, $\underset{\mathrm{M}1}{\overset{\bullet }{\theta }}$ and $T_{M2}$, $\underset{\mathrm{M}2}{\overset{\bullet }{\theta }}$ conditions.

Calculate the drive power consumption in each of the eight modes. (5) $\begin{array}{rl}min& P_e(T{\mathrm{\_}}_{req},T_{M1},i_1,i_2)\end{array}$(5) (6) $\begin{array}{rl}\mathrm{s}.\mathrm{t}.& 0\le \underset{M1}{\overset{\bullet }{\theta }}\le \underset{M1}{\overset{\bullet }{\theta }},max\end{array}$(6) (7) $\begin{array}{rl}& \begin{array}{rl}& \mathrm{i}\mathrm{f},\mathrm{T}\mathrm{\_}\mathrm{r}\mathrm{e}\mathrm{q}/{\mathrm{i}}_1>T_{M1},max\underset{M1}{\overset{\bullet }{\theta }}\\ & T_{M1}=T_{M1},max\underset{M1}{\overset{\bullet }{\theta }};\\ & \mathrm{i}\mathrm{f},\mathrm{T}\mathrm{\_}\mathrm{r}\mathrm{e}\mathrm{q}/{\mathrm{i}}_1<=T_{M1},max\underset{M1}{\overset{\bullet }{\theta }}\\ & T_{M1}=\mathrm{T}\mathrm{\_}\mathrm{r}\mathrm{e}\mathrm{q}/{\mathrm{i}}_1\end{array}\end{array}$(7) (8) $\begin{array}{rl}& 0\le T_{M1}\underset{M1}{\overset{\bullet }{\theta }}\le T_{M1},max\end{array}$(8) (9) $\begin{array}{rl}& 0\le \underset{M2}{\overset{\bullet }{\theta }}\le \underset{M2}{\overset{\bullet }{\theta }},max\end{array}$(9) (10) $\begin{array}{rl}& 0\le T_{M2}\underset{M2}{\overset{\bullet }{\theta }}\le T_{M2},max\underset{M2}{\overset{\bullet }{\theta }}\end{array}$(10) Where $\underset{M1}{\overset{\bullet }{\theta }},max$ and $\underset{M2}{\overset{\bullet }{\theta }},max$ denote the maximum speed of M1 and M2; $T_{M1},max\underset{M1}{\overset{\bullet }{\theta }}$, and $T_{M2},max\underset{M2}{\overset{\bullet }{\theta }}$ denote the maximum output torque of M1 and M2 at $\underset{M1}{\overset{\bullet }{\theta }}$ and $\underset{M2}{\overset{\bullet }{\theta }}$.

The speed requirement of real-time calculation of the torque allocation strategy is considered, while the iterative optimization-seeking algorithm is avoided. Therefore, in this paper, the global optimization is performed by using the equal-grid-spacing point-fetching method, which does not require derivative and iterative operations. By gridding $T_{M1}$, the feasible solution matrix of 100 torques is generated in $0\le T_{M1}\underset{M1}{\overset{\bullet }{\theta }}\le T_{M1},max$ with equal spacing, and the drive power consumption satisfying the conditions in each drive mode is calculated by Equations (1(1) $T{\mathrm{\_}}_{req}=T_{M1}{i}_1+T_{M2}{i}_2$(1) ) to (5(5) $\begin{array}{rl}min& P_e(T{\mathrm{\_}}_{req},T_{M1},i_1,i_2)\end{array}$(5) ). For the dual-motor coupled drive at low speed with high torque and high speed with high torque, the power consumption of the four drive modes with $T_{M1}$ is given in Figure 4. There is a clear difference between the different modes. The energy consumption is lower in DM1-1 mode at low-speed high torque and DM2-2 mode at high-speed high torque. Figure 5 gives the effect of $T_{M1}$ on power consumption at DM1-1 and DM2-2 modes, with the lowest power consumption at the ideal operating point. Following Figure 2, the energy consumption of each mode is compared to obtain Figure 6.

Figure 4. Power consumption in different modesr. (a) Low-speed high-torque coupling case (b) High-speed high-torque coupling case.

Figure 5. Power consumption at different torques. (a) Power consumption in DM1-1 mode (b) Power consumption in DM2-2 mode.

Figure 6. Drive mode division. (a) 8 types of drive mode division (b) 6 types of drive mode division.

From Figure 6(a), it can be seen that DM2-1 and DM1-2 drive modes account for a small percentage. Referring to Figure 4 and related data analysis, the difference in power consumption between DM2-1 and DM1-2 modes is 0.5%–2% compared with the next lowest mode, so discarding these two modes can reduce the frequency of dual motor mode switching and the complexity of the control strategy, while having little impact on the economy of the whole vehicle. Therefore, DMCP adopts the drive mode shown in Figure 6(b). The PSC strategy calculates the energy consumption magnitude of the six modes of drive in real time and selects the drive mode with the lowest energy consumption; in the dual-motor operating mode, the PSC can determine the values of the gears and torques of the two motors; in the single-motor operating mode, the PSC can select the working motor and its corresponding gear; since the energy consumption magnitude of each motor can be calculated in real-time, it is not necessary to Since the energy consumption of each motor can be calculated in real-time, there is no need to calibrate the shift curve. To avoid frequent mode switching, a penalty function for mode switching is designed and added to the torque distribution control strategy.

4. No power interruption mode switching strategy

Unlike conventional AMTs (automatic transmissions), the DMCP's two motors are independently controlled and can output torque separately or coupled to each other, and can provide torque compensation during mode switching. As a result, the DMCP can switch without power interruption and keep the transmission output torque smoothly changing. As shown in Figure 6, the mode switching is divided into four types of switching processes: two-motor mode switching to two-motor mode, two-motor mode switching to single-motor mode, single-motor mode switching to two-motor mode, and single-motor mode switching to single-motor mode, and there are 48 different mode switching processes. This paper selects the representative DM1-1 switching to DM2-2 and SM1-1 switching to SM2-1 to study. (11) $\begin{array}{rl}& {\mathrm{T}}_{\mathrm{i}\mathrm{n}}={\mathrm{T}}_{\mathrm{M}1}{\mathrm{i}}_1-\left(\mathrm{J}+{\mathrm{i}}_1\right)\underset{M1}{\overset{\bullet }{\theta }}+{\mathrm{T}}_{\mathrm{M}2}{\mathrm{i}}_2-\mathrm{J}\underset{M2}{\overset{\bullet }{\theta }}\end{array}$(11) (12) $\begin{array}{rl}& \mathrm{m}\mathrm{a}=\frac{{\mathrm{T}}_{\mathrm{i}\mathrm{n}}{\mathrm{i}}_0}{\mathrm{r}}-\frac{1}{2}\rho \mathrm{A}{\mathrm{C}}_{\mathrm{d}}{\mathrm{v}}^2+\mathrm{m}\mathrm{g}f+\delta \mathrm{m}\frac{d\mathrm{v}}{d\mathrm{t}}\end{array}$(12) (13) $\begin{array}{rl}& \mathrm{J}=\frac{{\mathrm{\partial }}^2\mathrm{V}}{\mathrm{\partial }{\mathrm{t}}^2}=\frac{\mathrm{\partial }\mathrm{a}}{\mathrm{\partial }\mathrm{t}}\end{array}$(13) ${\mathrm{T}}_{\mathrm{i}\mathrm{n}}$ represents the input torque of the main reducer, different mode δ is regarded as constant, J is the equivalent rotational inertia on the three axes, and is regarded as continuous, ignoring the effect of ramp resistance.

$◯1$ When the TCU (gearbox controller) receives the signal for mode switching, it first determines the current mode based on the position of the synchroniser bonding sleeve and the operation of the two motors and then determines the target mode based on the PSC strategy.

$◯2$ M1 unloads the torque to 0, while M2 compensates by lifting the torque to keep the torque of the transmission output shaft stable.

$◯3$ Synchronizer 1 is released and the binding sleeve is returned to the middle position.

$◯4$ M2 is output separately and M1 regulates the speed to meet the allowable range of the target gear combined with the speed difference.

$◯5$ Synchronizer 1 engages the target gear and M2 unloads torque to 0, while M1 up-torques in accordance with M2's unload, keeping the output torque stable.

$◯6$ Synchronizer 2 is released and the binding sleeve is returned to the middle position.

$◯7$ M1 is output separately and M1 regulates the speed to meet the allowable range of the target gear combined with the speed difference.

$◯8$ Synchronizer 2 engages the target gear, M1 unloads torque to the M1 torque assigned by PSC, M2 increases torque to the assigned target torque, mode switching is completed, and the target torque is steadily output. The specific DM1-1 switching to DM2-2 mode control process is referred to in flow Figure 7.

Figure 7. DM1-1 mode switching DM2-2 mode flow.

The process of SM1-1 mode switching SM2-1 mode is similar to the above mode switching, which is divided into 4 processes with specific reference to Figure 8. Before M1 unloads the torque, M2 is set to the target gear; while M1 unloads the torque, M2 increases the torque to make the output torque stable; when is 0, synchroniser 1 is disengaged and the mode switching is completed.

Figure 8. SM1-1 mode switch SM2-1 mode flow chart.

5. Simulation and analysis of results

In this paper, the AMESim platform is used to establish the physical model of the system, MATLAB/Simulink is used to establish the control strategy and conduct joint simulation, and the simulation model is shown in Figure 9. Figure 9(a) shows the simulation model of the two-motor multi-shift system, including the battery, motor, synchroniser, and driver models; the MATLAB control model shown in Figure 9(b) includes the complete vehicle controller (VCU) for the interface, the transmission controller (TCU) for the powerless interrupt mode switching strategy, the motor controller (MCU) for the motor speed and torque, and the real-time mode selection and torque PSC for real-time mode selection and torque distribution.

Figure 9. AMESim and MATLAB joint simulation model. (a) AMESim Dual Motor Multi-Shift System Physical Model (b) Simulink control model for dual motor multi-speed system.

5.1. Economic simulation

The DMCP system can select different drive modes and torque distribution ratios in real time according to the driver's demand, which has greater energy-saving potential than the conventional single-motor drive system (SMTSP). In order to verify the energy-saving effect of the PSC strategy on the DMCP system, the state of charge (SOC) at the beginning of the battery is set to 90%, and NEDC, UDDS, and 10–15 cycle operating conditions are simulated respectively, and the speed following cases are shown in Figure 10(a–c), respectively. The economic motor and power motor are selected to be stacked instead of SMTSP single motor for economic comparison, and the simulation working condition SOC changes are shown in Figure 10(d). The range improvement at NEDC, UDDS, and 10–15 cycle conditions are all over 10%. Figure 11 shows the actual operation of M1 and M2 in the acceleration section starting at 180s in UDDS in Figure 10(b). From the figure, it can be seen that M1, as the economic motor, has more operating points than M2, indicating that the real-time PSC strategy is able to select the more efficient SM1 drive mode as the current operating mode; both motors operate in the high-efficiency zone, with M1 mostly above 80% and M2 more than half of them operating in the 90% zone.

Figure 10. Cycle working condition and SOC variation graph. (a) NEDC cycle work following situation (b) UDDS cycle work following situation (c) 10–15 Circulating conditions follow (d) Comparison of the change of SOC in single cycle working condition.

Figure 11. UDDS acceleration section M1, M2 actual operation. (a) M1 Workpoint Distribution (b) M2 work point distribution.

5.2. Shift process simulation

PSC determines whether to perform mode switching according to current vehicle speed and demand torque. When mode switching is performed, PSC sends mode switching commands to TCU, TCU sends speed and torque commands to MCU, and shift commands to the synchroniser. Figure 12 shows the simulation result of switching from DM1-1 mode to DM2-2 mode. At 12.97 seconds, TCU receives the current DM2-2 mode command and determines that it is currently in SM1-1 mode by the two synchroniser positions; MCU controls M1 to unload torque to 0, while M2 performs torque compensation; at 13.25 seconds, M1 unloads torque and synchroniser 1 completes disengagement, and M1 regulates speed according to Figure 7. 13.36 seconds M1 speed regulation is completed, synchroniser 1 is connected to 2nd gear; 13.78 seconds M1 is driven separately in 2nd gear, M1 increases torque, M2 torque unloading is completed; 13.85 seconds synchroniser 2 is disengaged, M2 speed regulation is completed according to Figure 7; 14.18 seconds synchroniser 2 is connected to 2nd gear, mode switching is completed; 14.46 seconds M1 and M2 torque reach the target value. In Figure 12(c), M1 performs torque compensation, and when the maximum torque of M1 cannot meet the compensation demand, M1 compensates with the maximum torque; in Figure 12(d), there is no torque interruption during the whole mode switching process. Figure 13 shows the simulation result of switching from SM1-1 mode to SM2-1 mode. At 14.18s, M2 is tuned according to Figure 8; at 14.31s synchroniser 2 is connected to 1st gear, M1 starts to unload torque to 0, M2 increases torque; at 14.68s synchroniser 1 finishes disengaging, M2 torque reaches the target value; 14.72s mode switching is completed. Figure 13(b-c) show that the M1 unloading torque process of mode switching process M2 can compensate for the demand torque to make the main reducer input torque stable and realise the mode switching without power interruption. Figure 12 mode switching process is complicated involving three times torque coordination and two times motor speed regulation, which takes about 1.21 seconds, and the output torque fluctuation section only takes 0.2 seconds; Figure 13 mode switching takes about 0.54 seconds, and there is no fluctuation of output torque during the whole mode switching process. The shocking degree during mode switching shown in Figures 12(e) and 13(e) is controlled within 8 m/s, which is less than the recommended German shock degree of 10 m/s3, and the shift shock is small.

Figure 12. DM1-1 mode switching DM2-2 mode simulation results. (a) Synchronizer 1 binding sleeve position (b) Synchronizer 2 binding sleeve position (c) M1, M2 torque following situation (d) Output torque during mode switching (e) Impact degree during mode switching.

Figure 13. SM1-1 mode switching SM2-1 mode simulation results. (a) Synchronizer 1 binding sleeve position (b) Synchronizer 2 binding sleeve position (c) M1, M2 torque following situation (d) Output torque during mode switching (e) Change in impact degree during mode switching.

6. Conclusion

In view of the existence of multiple drive modes in the DMCP system, the drive modes are divided and simplified for the structural characteristics of the system. With the aim of reducing the total drive power consumption, real-time mode selection and torque distribution strategies are designed to achieve the optimal economy of DMCP at the steady-state output. To address the problem of switching between multiple drive modes, the process of switching from DM1-1 mode to DM2-2 mode and SM1-1 mode to SM2-1 mode is focused on in conjunction with the characteristics of torque compensation for dual motor drive systems, and a specific mode switching process is developed. The simulation model of dual-motor multi-speed coupling drive is built by AMESim/MATLAB, and the effectiveness of real-time torque distribution strategy and no power interruption mode switching strategy is verified by simulation analysis; the economy is significantly improved compared with that of single-motor drive, and the output torque fluctuation during the two modes switching is slight, and the absolute value of the real vehicle impact is less than 8m/s3.

After that, we will continue to complete the total drive power consumption of all drive modes of the DMCP system, optimise the torque distribution strategy, and achieve further optimization of the economy of DMCP. Continue to study the optimization of shift impact during upshifting and downshifting when driving mode switching.And we will compare the proposed method with existing schemes from different dimensions to obtain a multi-dimensional optimised allocation strategy.

Disclosure statement

No potential conflict of interest was reported by the author(s).