Dynamics and Model-Predictive Anti-Jerk Control of Connected Electric Vehicles
MetadataShow full item record
Electric Vehicles (EVs) develop high torque at low speeds, resulting in a high rate of acceleration. However, the rapid rise in torque of an electric motor creates undesired torsional oscillations, with vehicle jerk arising as a result of wheel slip or flexibility in the half-shaft. These torsional oscillations in the halfshaft lead to longitudinal oscillations in the wheels, thus reducing comfort and drivability. In this research, we have designed an anti-jerk longitudinal dynamics controller that damps out driveline oscillations and improves the drivability of EVs with central-drivetrain architecture. The anti-jerk longitudinal dynamics controller has been implemented for both traction and cruise control applications. We have used a model predictive control (MPC) approach to design the controller since it allows us to deal with multiple objectives in an optimal sense. The major scope of this research involves modeling, parameter identification, design and validation of the longitudinal dynamics controller. The real-time implementation has been demonstrated using hardware-in-the-loop experiments utilizing fast MPC solvers. The MapleSim software, which utilizes symbolic computation and optimized-code generation techniques to create models that are capable of real-time simulation, has been used to develop the longitudinal dynamics plant model. Road tests have been conducted on our test vehicle, a Toyota Rav4 electric vehicle (Rav4EV), to identify the parameters for the longitudinal dynamics model. Experimental data measured using a vehicle measurement system (VMS), global-positioning system (GPS), and inertial measurement unit (IMU) was used for parameter identification. Optimization algorithms have been used to identify the model parameters. A control-oriented model of the EV, which includes a flexible halfshaft and effect of wheel-slip transients, has been developed with the aim of controlling driveline oscillations. The MPC-based anti-jerk traction controller regulates the motor torque corresponding to the accelerator pedal position, to serve the dual objectives of traction and anti-jerk control. The performance of this controllers has been compared to that of other controllers in the literature. Since most traction controllers are on-off controllers and are only activated when wheel slip exceeds a desired limit, they are not effective in anti-jerk control. The MPC-based anti-jerk controller is able to serve multiple objectives related to anti-jerk as well as traction, and is therefore superior to other controllers. A unified design combining the upper and lower level MPC-based cruise controller has also been formulated to meet the anti-jerk objective during cruise control. The cruise controller has been designed such that it is adaptive to changes in road friction conditions. The efficacy of both traction and cruise controllers has been demonstrated through model-in-the-loop simulation, and the real-time capability has been demonstrated through hardware-in-the-loop experiments.
Cite this version of the work
Mohit Batra (2018). Dynamics and Model-Predictive Anti-Jerk Control of Connected Electric Vehicles. UWSpace. http://hdl.handle.net/10012/12997