Taking a hybrid electric vehicle using double-row planetary gear power coupling mechanism as a research object, this study proposes a coordinated control algorithm of “torque distribution, engine torque monitoring, and motor torque compensation” in an attempt to realize coordinated control for driving mode switching. Characteristic analysis of the power coupling mechanism was carried out, and the control strategy model in MATLAB/Simulink was built. Subsequently, the analysis of mode switching from the electric mode into joint driving mode was simulated. In addition, a multibody dynamics model of the power coupling mechanism was established and the simulation analysis during mode switching process was carried out. The results show that the proposed coordinated control strategy serves to effectively reduce the torque fluctuation and the impact degree during the mode switching process and improve the ride comfort of the vehicle. In the meantime, the time-domain and frequency-domain characteristics of gear meshing force and bearing restraint force indicate that the mode switching process of the dynamic coupling mechanism is quite stable and this control strategy contributes to improving the characteristics such as vibration and noise.
Hybrid electric vehicles have become a research focus due to their low emissions and outstanding advantages in energy conservation, which has made them the best choice for lowering vehicle energy consumption [
In order to solve this problem, some scholars have conducted relevant researches in this regard. Wang et al. proposed a motor torque compensation method based on PID strategy to control the mode switching process of a parallel hybrid electric vehicle [
Among all of the mode switching processes of hybrid electric vehicle, the engine starting process is the most unstable. First of all, there are many resistances in the engine starting process. These resistances will obviously slow down the engine starting speed. Then, temperature is the most important factor considering the external influence. When the engine is just started, the temperature is low and the viscosity of the lubricating oil is large, so the friction resistance between the components is large. At the same time, when the temperature is low, the atomization of fuel is poor, which leads to the deterioration of combustion process and the slow starting process of engine. When the working condition of the engine is stable, the output torque will be stable. It can be seen that the dynamic response of the engine is slow due to internal and external factors. In the mode switching process, the response difference between power sources will cause the fluctuation of the vehicle output torque, resulting in poor ride comfort. Therefore, it is necessary to develop a control strategy to compensate the engine to improve the vehicle ride comfort during mode switching process. The control technology of the motor is mature, which can realize the high-performance closed-loop control of the motor. Therefore, using the motor to compensate the engine is a simple and effective method.
In this paper, a hybrid electric vehicle using double-row planetary gear power coupling mechanism is selected as the research object. A dynamic coordinated control strategy of “torque distribution + engine torque monitoring + motor torque compensation” is proposed. The control strategy model was built in MATLAB/Simulink, and the vehicle model established in LMS.AMESim was used for the joint simulation. Furthermore, the rigid-flexible coupled dynamics model of the double-row planetary gear power coupling mechanism was established. We carried out the simulation analysis of the dynamics model during mode switching process in order to verify the proposed control strategy.
As shown in Figure
Double-row planetary gear power coupling system.
The main components and parameters of the hybrid electric vehicle.
Component | Parameter | Value |
---|---|---|
Engine | Power (kW) | 73 |
Peak torque (Nm) | 142 | |
MG1 | Power (kW) | 42 |
MG2 | Power (kW) | 60 |
Peak torque (Nm) | 207 | |
Battery | Maximum power (kW) | 27 |
Battery capacity (Ah) | 6.5 | |
Coupling mechanism | Front row characteristic parameter | 2.6 |
Rear row characteristic parameter | 2.636 |
Depending on the vehicle demand, the driving system flexibly combines the engine and the motors. When the vehicle runs in low power, it operates in electric mode while the engine is shut down to reduce emissions. When the vehicle runs in a normal state, the system keeps the engine in an optimal working condition, and in the meantime the battery can be charged by the motor MG1. As a result, the system can greatly improve the vehicle energy efficiency and the fuel economy [
In order to analyze the mechanism, a diagram is used to indicate its power transmission, as shown in Figure
Power transmission of coupling mechanism.
The power coupling mechanism is analyzed using the isolation method. The engine and front row planet carrier are considered together and the dynamics equation can be expressed in accordance to
Analyzing the motor MG1 and front row sun gear, the following equation can be obtained:
Analyzing motor MG2 and rear row sun gear, the following equation can be obtained:
The dynamics equation of the composite ring gear can be expressed in accordance to
The transmission characteristics of planetary gears can be expressed in accordance to
When the power coupling mechanism is in different working modes, the mathematical model will evolve into different forms.
The key of coordinated control for mode switching is to control the real-time output torque of power sources, and then the total torque does not fluctuate greatly. The premise of this method is that the control system can obtain the real-time output torque of the engine. There are two ways to realize this function, one method is to use torque sensor to measure directly, and the other is to use algorithm to estimate engine torque indirectly. This paper uses BP neural network optimized by genetic algorithm to estimate engine torque.
BP neural network is composed of three parts: input layer, intermediate layer, and output layer. The intermediate layer can be expanded to multilayers. All neurons in the adjacent layers are connected, but there is no connection between neurons in the same layer. The main advantage of BP neural network is its strong nonlinear mapping ability. The structure of BP network model is as shown in Figure
Structure of BP network model.
Genetic algorithm simulates the evolution theory and gene genetics principle in the development of nature and uses the continuous evolution to search the global optimal solution. The new species are more adaptable to the environment with each evolution, and the optimal individual in the last population is the optimal approximate solution.
Data processing is to transform the original data into that which can be recognized by neural network algorithm, and it is an important step of algorithm operation. Data processing can improve the data quality and plays an important role in improving the accuracy and performance of network training. The experimental data used in this paper are shown in Table
Experimental data of engine.
n | 10 | 20 | 30 | 40 | 50 | 60 | 70 | 80 | 90 | 100 |
---|---|---|---|---|---|---|---|---|---|---|
800 | 25.4 | 61.2 | 83.7 | 101.4 | 103.3 | 104.8 | 105.3 | 105.8 | 106.5 | 107.4 |
1000 | 22.8 | 54.9 | 79.5 | 100.8 | 104.6 | 105.1 | 105.4 | 106.2 | 107.8 | 108.5 |
1200 | 20.2 | 49.6 | 76.2 | 99.7 | 107.3 | 108.6 | 109.1 | 109.9 | 110.2 | 111.6 |
1400 | 16.3 | 45.5 | 72.4 | 98.4 | 108.4 | 110.9 | 111.5 | 112.5 | 113.8 | 114.3 |
1600 | 41.4 | 68.2 | 97.6 | 110.6 | 113.4 | 114.2 | 114.9 | 115.6 | 117.1 | |
1800 | 37.2 | 64.1 | 95.4 | 113.1 | 115.8 | 116.7 | 117.1 | 118.4 | 120.2 | |
2000 | 34.3 | 59.8 | 93.1 | 115.6 | 118.1 | 119.2 | 120.1 | 121.6 | 122.9 | |
2200 | 31.1 | 55.7 | 90.8 | 117.5 | 122.8 | 123.8 | 124.7 | 125.3 | 126.8 | |
2400 | 28.9 | 51.6 | 88.4 | 119.8 | 125.5 | 127.4 | 129.8 | 130.6 | 131.8 | |
2600 | 23.5 | 47.9 | 87.6 | 120.8 | 127.2 | 130.8 | 132.6 | 133.5 | 134.6 | |
2800 | 21.6 | 43.3 | 85.4 | 120.5 | 129.7 | 133.3 | 135.2 | 135.9 | 136.1 | |
3000 | 19.6 | 40.1 | 82.1 | 120 | 130.3 | 135.1 | 136.6 | 137 | 137.5 | |
3200 | 37.5 | 78.9 | 117.9 | 131.1 | 136.4 | 137.5 | 138.1 | 138.4 | ||
3400 | 34.9 | 76.5 | 115.6 | 132.2 | 137.9 | 138.2 | 138.9 | 139.6 | ||
3600 | 31.6 | 72.8 | 113.5 | 133.3 | 138.3 | 138.9 | 139.8 | 140.8 | ||
3800 | 28.2 | 69.2 | 112.6 | 134 | 139 | 139.2 | 140 | 141.9 | ||
4000 | 25.5 | 65.8 | 111.3 | 133.6 | 137.9 | 138.7 | 140.2 | 142.8 | ||
4200 | 62.5 | 110.6 | 132.6 | 136.5 | 140.1 | 141 | 141.6 | |||
4400 | 59.9 | 109.8 | 131.7 | 135.2 | 138.9 | 139.5 | 140.2 | |||
4600 | 57.7 | 108.1 | 129.6 | 134.7 | 137.6 | 138.4 | 139 | |||
4800 | 56.1 | 107.8 | 126.9 | 132.6 | 136 | 137.1 | 137.8 | |||
5000 | 55.2 | 106.7 | 125.5 | 130.5 | 132.2 | 135 | 136.5 | |||
5200 | 105.9 | 124 | 128.6 | 129.5 | 134.6 | 135.5 |
The data processing of neural network is the data normalization that is to scale the original data according to specific rules, so that the processed data falls in a specific range. The calculation method is as follows:
The structure and parameters of BP neural network are shown in Table
Structure and parameters of algorithm.
Algorithm | Parameters | |
---|---|---|
BP neural network | Number of neurons in input layer | 2 |
Number of neurons in the output layer | 1 | |
Number of intermediate layers | 1 | |
Number of neurons in the intermediate layer | 9 | |
Network structure | 2 × 9 × 1 | |
Input-intermediate layer transfer function | Tansig | |
Intermediate-output layer transfer function | Logsig | |
Training algorithm | Trainlm | |
Maximum training times | 1000 | |
Learning rate | 0.1 | |
MSE | 0.001 | |
Genetic algorithm | Initial population size | 50 |
Code binary digit | 10 | |
Genetic algebra | 100 | |
Crossover probability | 0.7 | |
Mutation probability | 0.01 |
Genetic algorithm error.
Using genetic algorithm is to find the optimal weight and threshold value and assign these values to BP neural network for training. Figure
BP neural network error.
Using the nonlinear network obtained by the above algorithm, the output torque estimation of engine is carried out, and the map of engine torque estimation is obtained, as shown in Figure
MAP of engine torque estimation.
In essence, the driving torque fluctuation and longitudinal impact during mode switching are attributed to the difference in the dynamic response of the engine and motors. Therefore, the process of “engine starting or stopping” necessitates coordinated control. On this basis, this paper proposed a dynamic coordinated control strategy of “torque distribution + engine torque monitoring + motor torque compensation” to improve the ride comfort during the mode switching process. First, the engine torque and the motor torque are distributed according to the torque demand (as shown in (
Subsequently, the switching process from electric mode to the joint driving mode is taken as an example. The lever method is used to analyze the working condition of the system at this stage, as shown in Figure
Lever method of speed.
Power transmission of engine starting process.
The flow chart of the dynamic coordination control during this mode switching process is shown in Figure
Flow chart of dynamic coordinated control.
When the vehicle operates in the pure electric mode, it is driven by the motor MG2, the engine, and motor MG1 which are stopped, so
When the engine is started based on the torque distribution strategy, the start flag of the dynamic control strategy
Therefore, the actual torque demand of the motor MG2 can be expressed as
The motor MG2 keeps torque compensation during the process of the engine starting up. When the actual torque of the engine reaches 95% of the necessary torque, the torque compensation can be considered to be ended. In this case, the stop flag
In the joint driving mode, when the engine is working, its output power is divided into two parts through the front row planetary gear: one part drives the vehicle through the ring gear, and the other part drives the motor MG1 to generate electricity, and the output torque can be expressed as
There are subjective and objective evaluation methods for ride comfort evaluation. Subjective evaluation method varies from person to person due to personal feelings. Therefore, the objective evaluation method can be used to make more accurate judgments. In this case, the concept of impact degree can be introduced as the evaluation index of the mode switching process. The impact degree is defined as the changing rate of the longitudinal acceleration of the vehicle [
In addition, the characteristics of vibration and noise of the power coupling mechanism are closely related to the meshing transmission of the planetary gears [
With the LMS.AMESim, a vehicle dynamics model was established. The models of vehicle energy distribution strategy and coordinated control strategy for the mode switching were established in MATLAB/Simulink; thus, joint simulation was carried out. The results are shown in Figures
NEDC simulation result.
Output torque of power sources. (a) Output torque of each power source before control. (b) Output torque of each power source after control.
Comparison of impact degree before and after control.
The dynamic model was simulated using the New European Driving Cycle (NEDC) [
Figure
Comparison before and after coordinated control.
Before control | After control | Improvement percentage | |
---|---|---|---|
Maximum torque deviation (Nm) | 264.11 | 48.3 | 81.71 |
Maximum impact degree (m/s3) | 9.28 | 0.94 | 89.87 |
In summary, after the coordinated control strategy was added to the vehicle, the maximum torque deviation and the maximum impact degree were greatly reduced; the vehicle ride comfort is improved.
In order to study the vibration and noise at the switching moment, the solid model of the whole system was established. Then, it was imported into ADAMS for constructing a multirigid dynamic model. Thereafter, ANSYS was performed to flex the input shaft of the motors and engine. The rigid-flexible coupled dynamics model was established finally, as shown in Figure
The rigid-flexible coupled dynamics model.
In order to study the dynamic characteristics of the power coupling mechanism under this control strategy, the analysis of mode switching was performed. The simulation load is the torque converted from whole vehicle to the ring gear and power sources data obtained by the joint simulation. The results are as follows.
Figure
Angular velocity of shafts. (a) Angular velocity of the shaft of motor MG1. (b) Angular velocity of the engine shaft.
The vibration and noise of the power coupling mechanism are closely associated with the meshing of the planetary gears. Therefore, the meshing force between the gears is selected as the research object. In the analysis of the meshing force characteristics of the planetary gears, the meshing frequency can be expressed in accordance to the following equation [
The meshing frequency of rear row can be solved with (
Figure
The time-frequency characteristics of the meshing force. (a) Sun gear and planet gear of front row. (b) Ring gear and planet gear of front row. (c) Sun gear and planet gear of rear row. (d) Ring gear and planet gear of rear row.
The bearing restraint force acts as an excitation source to induce the power coupling mechanism to generate vibration, which has important significance in terms of analysis of the vibration and noise characteristics of the power coupling mechanism. In this paper, the rotating pair is set in each bearing center to simulate the bearing, and the time-domain and frequency-domain characteristics of the bearing restraint force were observed, as shown in Figure
The time-domain and frequency-domain characteristics of the bearing restraint force. (a) Input shaft bearing of engine. (b) Input shaft bearing of MG1. (c) Input shaft bearing of MG2. (d) Bearing of composite ring gear.
According to Figure During the mode switching, the engine is driven by the motor MG1 to obtain the idle speed and starts to output torque. The time-domain characteristics of bearing restraint force of motor MG1 and engine change smoothly during this process. In addition, the frequency-domain characteristics remain stable without obvious peak values. The bearing restraint force of the input shaft of motor MG2 has peaks at its meshing frequency The composite ring gear couples the torque of the engine and the motors. During the switching process, it must undertake the impact of the engine. As a result, its dynamic characteristics are more complicated, but it only has peak values at the frequencies
In summary, under this control strategy, the power coupling mechanism works smoothly and the time-domain and frequency-domain characteristics are sound during the mode switching process, which contributes to improving the vibration and noise of the vehicle.
In order to improve the ride comfort of a hybrid electric vehicle using double-row planetary gear power coupling mechanism during mode switching process, a dynamic coordinated control strategy of “torque distribution + engine torque monitoring + motor torque compensation” was proposed. The mathematical model and control strategy model were established; some simulation analyses were carried out in order to verify the proposed coordinated control strategy. The main conclusions can be summarized as follows: When switching from electric mode to joint driving mode is applied, the dynamic coordinated control strategy proposed in this study can effectively reduce the driving torque fluctuation and vehicle impact and improve the vehicle ride comfort. It is verified that the proposed dynamic coordinated control strategy is effective. The simulation analysis of the time-frequency characteristics of the gear meshing force and bearing restraint force indicates that the gear meshing force and bearing restraint force in the power coupling mechanism change smoothly and there is no obvious impact under the proposed control strategy, which is conducive to improve the vibration and noise characteristics of vehicles.
The data used to support the findings of this study are available from the corresponding author upon request.
The authors declare that there are no conflicts of interest regarding the publication of this paper.
This work was financially supported by the National Natural Science Foundation of China (51575238).