Dynamic Modeling and Parameter Identification of a Plug-in Hybrid Electric Vehicle
MetadataShow full item record
In recent times, mechanical systems in an automobile are largely controlled by embedded systems, called micro-controllers. These automobiles, installed with micro-controllers, run complex embedded code to improve the efficiency and performance of the targeted mechanical systems. Developing and testing these control algorithms using the concept of model based design (MBD) is a cost-efficient and time-saving approach. MBD employs vehicle system models throughout the design process and offers superior understanding of the system behaviour than a traditional hardware prototype based testing. Consequently, accurate system identification constitutes an important aspect in MBD. The main focus of this thesis is to develop a validated vehicle dynamics model of a Toyota Prius Plug-in hybrid vehicle. This model plays a crucial role in achieving better fuel economy by assisting in the development process of various controller designs such as energy management system, co-operative adaptive cruise control system, and trip planning module. In this work, initially a longitudinal vehicle dynamics model was developed in MapleSim that utilizes acausal modeling techniques and symbolic code generation to create models that are capable of real-time simulation. Here, the motion in longitudinal direction was given importance as it is the crucial degree of freedom (DOF) for determining the fuel consumption. Besides, the generic and full-fledged vehicle dynamics model in Simulink-based Automotive Simulation Models (ASM) software was also modified to create a validated model of the Prius. This software specifically facilitates the implementation of the model for virtual data collection using a driving simulator. Both vehicle models were verified by studying their simulation results at every stage of the development process. Once the vehicle models were fully functional, the accurate and reliable parameters that control the vehicle motion were estimated. For this purpose, experimental data was acquired from the on-road and rolling dynamometer testing of the Prius. During these tests, the vehicle was instrumented with a vehicle measurement system (VMS), global-positioning system (GPS), and inertial measurement unit (IMU) to collect synchronized vehicle dynamics data. Parameters were identified by choosing a local optimization algorithm that minimizes the difference between simulated and experimental results. Homotopy, a global optimization technique was also investigated to check the influence of optimization algorithms on the suspension parameters. This method of parameter estimation from on-road data is highly flexible and economical. Comparison with the parameters obtained from 4-Post testing, a standardized test method, shows that the proposed methods can estimate parameters with an accuracy of 90%. Moreover, the longitudinal and lateral dynamics exhibited by the developed vehicle models are in accordance with the experimental data from on-road testing. The full vehicle simulations suggest that these validated models can be successfully used to evaluate the performance of controllers in real time.
Cite this version of the work
Sindhura Buggaveeti (2017). Dynamic Modeling and Parameter Identification of a Plug-in Hybrid Electric Vehicle. UWSpace. http://hdl.handle.net/10012/12398