In order to face the growing challenges from air pollution, dependence on fossil oil and greenhouse gas emissions, electric vehicles have attracted unprecedented amount of global attentions from the governments, academia, industry, public and environmental organizations. Electric racing becomes more popular under this background: a new championship called Formula E has been held for three years since 2014. An upcoming championship named Roborace in 2017 will be the first global championship for autonomous electric race cars, which will open a new page of racing. Electric racing is undoubtedly a good platform to draw the attentions of the public on electric vehicles and also for testing and improving the most advanced design and control technologies. Compared with conventional internal combustion engine vehicles, electric vehicles can have very flexible powertrain topologies, which can be 1-motor, 2-motor and 4-motor driving with different mounted positions, and the transmissions can be single-speed or multi-speed. From the design point of view, the selection of the powertrain layouts, the motors and transmissions can affect the dynamic performance of the electric vehicles directly. As for control, different steering, accelerating and braking operations will result different trajectories, velocity profiles and lap time. Thus, the advanced design and control technologies are always expected. Recognizing the limitations of the conventional powertrain design approaches, this work is dedicated to achieving further improvements by proposing innovative optimal design approaches of the electric powertrain, with an electric race car as the platform. In order to test the performance of a designed powertrain, a corresponding control strategy should be developed. However, there are various kinds of control approaches for the electric vehicles, and accordingly, different control strategies may result in different results with the same designed powertrain. Considering this, the optimal control of the electric race car is coupled into the optimal design problem. The final results includes both the optimal design and control solutions for different powertrain layouts. In order to represent the vehicle behavior for cornering, braking, acceleration and comfort performance studies for four wheel driving vehicles with independent suspensions more accurately, a 14-DOF vehicle model is necessary. In this work, a 14-DOF vehicle model together with a suspension model considering the details of toe angle, camber angle, anti-roll force and suspension forces, are developed based on Lagrangian dynamics. To accurately predict the behavior of a vehicle, it is also required to estimate the external forces acting on the vehicle as precisely as possible. An empirical tire model based on the well-known Magic formula equations is programmed to calculate the tire forces. In particular, the tire model developed in MATLAB supporting inputs of the standard '.tir' tire data file. In order to evaluate the effect of design parameters of the motor and the transmission to the lap time of the race car, the mass model of the motor and transmission mainly concerning the dependence of the mass and output torque of the powertrain on the design parameters are derived. A virtual driver model is also devised to track a given trajectory depicted in curvilinear coordinate system based on the proposed control logic, and the obtained results are served as the initial guess of the optimal powertrain design and control problem. Heavy computing workload is a common issue in large scale optimization and optimal control problem. In order to improve the computational efficiency, all of the mentioned models programmed supports matrices operations. The entire vehicle model is validated with a well-known vehicle dynamics simulator 'VI-CarRealTime' developed by VI-Grade. After a detailed reviewing of the numerical approaches for optimal control problems, a MATLAB software package for General DYNamic OPTimal control problems (abbreviated as GDYNOPT) based on direct collocation methods is developed. GDYNOPT is implemented different transcription methods including both the local collocation and global collocation approaches, differential methods including forward, central, complex step and analytical differential methods. Moreover, it has the features of automatic scaling based on linear scaling and a proposed average gradients scaling approach, sparsity and supporting parallel computation. The optimal powertrain design and control problems of the 1-motor driving, 2-motor and 4-motor driving topologies based on the developed entire vehicle model are formulated and solved with GDYNOPT based on an direct transcription method for the first time with reference to the existing literature. In addition, an innovative approach is proposed to smooth the control trajectories. The optimal powertrain design parameters, control arcs and the optimal racing lines are obtained and analyzed with different number of collocation nodes. The obtained optimal design parameters in this work can be used as the reference for the motor and transmission design of electric race cars, while the optimal control results can serve as the benchmark to develop and evaluate the closed loop control strategy. In addition, the obtained racing line and steering wheel angles can be used to train the race car driver in a car simulator. The methodology proposed in this work can also be applied in the design and control of the common type ICE vehicles, EVs and HEVs with different driving profiles and objective functions.

In order to face the growing challenges from air pollution, dependence on fossil oil and greenhouse gas emissions, electric vehicles have attracted unprecedented amount of global attentions from the governments, academia, industry, public and environmental organizations. Electric racing becomes more popular under this background: a new championship called Formula E has been held for three years since 2014. An upcoming championship named Roborace in 2017 will be the first global championship for autonomous electric race cars, which will open a new page of racing. Electric racing is undoubtedly a good platform to draw the attentions of the public on electric vehicles and also for testing and improving the most advanced design and control technologies. Compared with conventional internal combustion engine vehicles, electric vehicles can have very flexible powertrain topologies, which can be 1-motor, 2-motor and 4-motor driving with different mounted positions, and the transmissions can be single-speed or multi-speed. From the design point of view, the selection of the powertrain layouts, the motors and transmissions can affect the dynamic performance of the electric vehicles directly. As for control, different steering, accelerating and braking operations will result different trajectories, velocity profiles and lap time. Thus, the advanced design and control technologies are always expected. Recognizing the limitations of the conventional powertrain design approaches, this work is dedicated to achieving further improvements by proposing innovative optimal design approaches of the electric powertrain, with an electric race car as the platform. In order to test the performance of a designed powertrain, a corresponding control strategy should be developed. However, there are various kinds of control approaches for the electric vehicles, and accordingly, different control strategies may result in different results with the same designed powertrain. Considering this, the optimal control of the electric race car is coupled into the optimal design problem. The final results includes both the optimal design and control solutions for different powertrain layouts. In order to represent the vehicle behavior for cornering, braking, acceleration and comfort performance studies for four wheel driving vehicles with independent suspensions more accurately, a 14-DOF vehicle model is necessary. In this work, a 14-DOF vehicle model together with a suspension model considering the details of toe angle, camber angle, anti-roll force and suspension forces, are developed based on Lagrangian dynamics. To accurately predict the behavior of a vehicle, it is also required to estimate the external forces acting on the vehicle as precisely as possible. An empirical tire model based on the well-known Magic formula equations is programmed to calculate the tire forces. In particular, the tire model developed in MATLAB supporting inputs of the standard '.tir' tire data file. In order to evaluate the effect of design parameters of the motor and the transmission to the lap time of the race car, the mass model of the motor and transmission mainly concerning the dependence of the mass and output torque of the powertrain on the design parameters are derived. A virtual driver model is also devised to track a given trajectory depicted in curvilinear coordinate system based on the proposed control logic, and the obtained results are served as the initial guess of the optimal powertrain design and control problem. Heavy computing workload is a common issue in large scale optimization and optimal control problem. In order to improve the computational efficiency, all of the mentioned models programmed supports matrices operations. The entire vehicle model is validated with a well-known vehicle dynamics simulator 'VI-CarRealTime' developed by VI-Grade. After a detailed reviewing of the numerical approaches for optimal control problems, a MATLAB software package for General DYNamic OPTimal control problems (abbreviated as GDYNOPT) based on direct collocation methods is developed. GDYNOPT is implemented different transcription methods including both the local collocation and global collocation approaches, differential methods including forward, central, complex step and analytical differential methods. Moreover, it has the features of automatic scaling based on linear scaling and a proposed average gradients scaling approach, sparsity and supporting parallel computation. The optimal powertrain design and control problems of the 1-motor driving, 2-motor and 4-motor driving topologies based on the developed entire vehicle model are formulated and solved with GDYNOPT based on an direct transcription method for the first time with reference to the existing literature. In addition, an innovative approach is proposed to smooth the control trajectories. The optimal powertrain design parameters, control arcs and the optimal racing lines are obtained and analyzed with different number of collocation nodes. The obtained optimal design parameters in this work can be used as the reference for the motor and transmission design of electric race cars, while the optimal control results can serve as the benchmark to develop and evaluate the closed loop control strategy. In addition, the obtained racing line and steering wheel angles can be used to train the race car driver in a car simulator. The methodology proposed in this work can also be applied in the design and control of the common type ICE vehicles, EVs and HEVs with different driving profiles and objective functions.

Optimal powertrain design and control of an electric race car

YU, HUILONG

Abstract

In order to face the growing challenges from air pollution, dependence on fossil oil and greenhouse gas emissions, electric vehicles have attracted unprecedented amount of global attentions from the governments, academia, industry, public and environmental organizations. Electric racing becomes more popular under this background: a new championship called Formula E has been held for three years since 2014. An upcoming championship named Roborace in 2017 will be the first global championship for autonomous electric race cars, which will open a new page of racing. Electric racing is undoubtedly a good platform to draw the attentions of the public on electric vehicles and also for testing and improving the most advanced design and control technologies. Compared with conventional internal combustion engine vehicles, electric vehicles can have very flexible powertrain topologies, which can be 1-motor, 2-motor and 4-motor driving with different mounted positions, and the transmissions can be single-speed or multi-speed. From the design point of view, the selection of the powertrain layouts, the motors and transmissions can affect the dynamic performance of the electric vehicles directly. As for control, different steering, accelerating and braking operations will result different trajectories, velocity profiles and lap time. Thus, the advanced design and control technologies are always expected. Recognizing the limitations of the conventional powertrain design approaches, this work is dedicated to achieving further improvements by proposing innovative optimal design approaches of the electric powertrain, with an electric race car as the platform. In order to test the performance of a designed powertrain, a corresponding control strategy should be developed. However, there are various kinds of control approaches for the electric vehicles, and accordingly, different control strategies may result in different results with the same designed powertrain. Considering this, the optimal control of the electric race car is coupled into the optimal design problem. The final results includes both the optimal design and control solutions for different powertrain layouts. In order to represent the vehicle behavior for cornering, braking, acceleration and comfort performance studies for four wheel driving vehicles with independent suspensions more accurately, a 14-DOF vehicle model is necessary. In this work, a 14-DOF vehicle model together with a suspension model considering the details of toe angle, camber angle, anti-roll force and suspension forces, are developed based on Lagrangian dynamics. To accurately predict the behavior of a vehicle, it is also required to estimate the external forces acting on the vehicle as precisely as possible. An empirical tire model based on the well-known Magic formula equations is programmed to calculate the tire forces. In particular, the tire model developed in MATLAB supporting inputs of the standard '.tir' tire data file. In order to evaluate the effect of design parameters of the motor and the transmission to the lap time of the race car, the mass model of the motor and transmission mainly concerning the dependence of the mass and output torque of the powertrain on the design parameters are derived. A virtual driver model is also devised to track a given trajectory depicted in curvilinear coordinate system based on the proposed control logic, and the obtained results are served as the initial guess of the optimal powertrain design and control problem. Heavy computing workload is a common issue in large scale optimization and optimal control problem. In order to improve the computational efficiency, all of the mentioned models programmed supports matrices operations. The entire vehicle model is validated with a well-known vehicle dynamics simulator 'VI-CarRealTime' developed by VI-Grade. After a detailed reviewing of the numerical approaches for optimal control problems, a MATLAB software package for General DYNamic OPTimal control problems (abbreviated as GDYNOPT) based on direct collocation methods is developed. GDYNOPT is implemented different transcription methods including both the local collocation and global collocation approaches, differential methods including forward, central, complex step and analytical differential methods. Moreover, it has the features of automatic scaling based on linear scaling and a proposed average gradients scaling approach, sparsity and supporting parallel computation. The optimal powertrain design and control problems of the 1-motor driving, 2-motor and 4-motor driving topologies based on the developed entire vehicle model are formulated and solved with GDYNOPT based on an direct transcription method for the first time with reference to the existing literature. In addition, an innovative approach is proposed to smooth the control trajectories. The optimal powertrain design parameters, control arcs and the optimal racing lines are obtained and analyzed with different number of collocation nodes. The obtained optimal design parameters in this work can be used as the reference for the motor and transmission design of electric race cars, while the optimal control results can serve as the benchmark to develop and evaluate the closed loop control strategy. In addition, the obtained racing line and steering wheel angles can be used to train the race car driver in a car simulator. The methodology proposed in this work can also be applied in the design and control of the common type ICE vehicles, EVs and HEVs with different driving profiles and objective functions.
COLOSIMO, BIANCA MARIA
GIGLIO, MARCO
CHELI, FEDERICO
18-lug-2017
In order to face the growing challenges from air pollution, dependence on fossil oil and greenhouse gas emissions, electric vehicles have attracted unprecedented amount of global attentions from the governments, academia, industry, public and environmental organizations. Electric racing becomes more popular under this background: a new championship called Formula E has been held for three years since 2014. An upcoming championship named Roborace in 2017 will be the first global championship for autonomous electric race cars, which will open a new page of racing. Electric racing is undoubtedly a good platform to draw the attentions of the public on electric vehicles and also for testing and improving the most advanced design and control technologies. Compared with conventional internal combustion engine vehicles, electric vehicles can have very flexible powertrain topologies, which can be 1-motor, 2-motor and 4-motor driving with different mounted positions, and the transmissions can be single-speed or multi-speed. From the design point of view, the selection of the powertrain layouts, the motors and transmissions can affect the dynamic performance of the electric vehicles directly. As for control, different steering, accelerating and braking operations will result different trajectories, velocity profiles and lap time. Thus, the advanced design and control technologies are always expected. Recognizing the limitations of the conventional powertrain design approaches, this work is dedicated to achieving further improvements by proposing innovative optimal design approaches of the electric powertrain, with an electric race car as the platform. In order to test the performance of a designed powertrain, a corresponding control strategy should be developed. However, there are various kinds of control approaches for the electric vehicles, and accordingly, different control strategies may result in different results with the same designed powertrain. Considering this, the optimal control of the electric race car is coupled into the optimal design problem. The final results includes both the optimal design and control solutions for different powertrain layouts. In order to represent the vehicle behavior for cornering, braking, acceleration and comfort performance studies for four wheel driving vehicles with independent suspensions more accurately, a 14-DOF vehicle model is necessary. In this work, a 14-DOF vehicle model together with a suspension model considering the details of toe angle, camber angle, anti-roll force and suspension forces, are developed based on Lagrangian dynamics. To accurately predict the behavior of a vehicle, it is also required to estimate the external forces acting on the vehicle as precisely as possible. An empirical tire model based on the well-known Magic formula equations is programmed to calculate the tire forces. In particular, the tire model developed in MATLAB supporting inputs of the standard '.tir' tire data file. In order to evaluate the effect of design parameters of the motor and the transmission to the lap time of the race car, the mass model of the motor and transmission mainly concerning the dependence of the mass and output torque of the powertrain on the design parameters are derived. A virtual driver model is also devised to track a given trajectory depicted in curvilinear coordinate system based on the proposed control logic, and the obtained results are served as the initial guess of the optimal powertrain design and control problem. Heavy computing workload is a common issue in large scale optimization and optimal control problem. In order to improve the computational efficiency, all of the mentioned models programmed supports matrices operations. The entire vehicle model is validated with a well-known vehicle dynamics simulator 'VI-CarRealTime' developed by VI-Grade. After a detailed reviewing of the numerical approaches for optimal control problems, a MATLAB software package for General DYNamic OPTimal control problems (abbreviated as GDYNOPT) based on direct collocation methods is developed. GDYNOPT is implemented different transcription methods including both the local collocation and global collocation approaches, differential methods including forward, central, complex step and analytical differential methods. Moreover, it has the features of automatic scaling based on linear scaling and a proposed average gradients scaling approach, sparsity and supporting parallel computation. The optimal powertrain design and control problems of the 1-motor driving, 2-motor and 4-motor driving topologies based on the developed entire vehicle model are formulated and solved with GDYNOPT based on an direct transcription method for the first time with reference to the existing literature. In addition, an innovative approach is proposed to smooth the control trajectories. The optimal powertrain design parameters, control arcs and the optimal racing lines are obtained and analyzed with different number of collocation nodes. The obtained optimal design parameters in this work can be used as the reference for the motor and transmission design of electric race cars, while the optimal control results can serve as the benchmark to develop and evaluate the closed loop control strategy. In addition, the obtained racing line and steering wheel angles can be used to train the race car driver in a car simulator. The methodology proposed in this work can also be applied in the design and control of the common type ICE vehicles, EVs and HEVs with different driving profiles and objective functions.
Tesi di dottorato
File allegati
File Dimensione Formato  
thesis_HuilongYu.pdf

Open Access dal 28/06/2020

Descrizione: Thesis text
Dimensione 19.33 MB
Formato Adobe PDF
19.33 MB Adobe PDF Visualizza/Apri

I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10589/134573