The University Scholars Program

The proposed research is focused on applying new implicit learning-based optimal control methods to the energy management of a power-split hybrid electric vehicle, a nonlinear and dynamic system that is a subject to uncertainty and generic disturbances. The main goal of the proposed research include the development and experimental verification of using new adaptive and implicit methods that yield asymptotic tracking for an uncertain nonlinear system based on the dynamic model of a power-split powertrain, the constraints on the states and the control inputs. With new implicit learning-based optimal control methods the challenge to solve a Hamilton-Jacobi equation, along with the lack of mathematical tools needed to compensate for the generic disturbances, it is now more likely that an approximate optimal solution can be reached. Through collaboration with Dr. Dixon and his research group, we will investigate how reinforcement learning based dynamic programming methods can be applied to uncertain nonlinear dynamic models. Since current methods are unable to analytically provide a solution to the Hamilton-Jacobi equation, adaptive control methods will be used to asymptotically approximate the solution. This method, as described above, can be applied in an optimal control framework to asymptotically approach the solution the optimal energy management system despite uncertainties and nonlinearities. In addition to working in Dr. Dixon’s research group, I will perform a literature review and work with graduate students in Dr. Dixon’s research group to develop control methods. I will then implement developed controllers to demonstrate the application of adaptive control methods and to yield an optimal solution to the energy-management of a Hybrid Electric Vehicle. Real world applications involve applying this method to better understand the use of this control method on energy systems of hybrid vehicles.