A Study on Fuel Efficiency Optimization Control Algorithm for Fuel Cell Hybrid Electric Vehicles Using Hierarchical Nonlinear Model Predictive Control

Abstract

A Study on Fuel Efficiency Optimization Control Algorithm for Fuel Cell Hybrid Electric Vehicles Using Hierarchical Nonlinear Model Predictive Control Lee, Jooin Dept. of electrical engineering Graduate School of Hanyang University This paper proposes an Energy Management System (EMS) for optimizing the fuel efficiency of Fuel Cell Hybrid Electric Vehicles (FCHEVs) using Hierarchical Nonlinear Model Predictive Control (HNMPC). Each layer of the HNMPC elaborates urban traffic conditions and the operational characteristics of the target vehicle. The Upper-level of HNMPC considers future traffic information and the dynamic characteristics of the fuel cell and battery to generate a Reference State of Charge (SOC) Profile for a long-time prediction horizon. The Reference SOC Profile represents the battery's available operational range to the target vehicle as it travels along the driving route. In this paper, a Nonlinear Model Predictive Control (NMPC) approach is proposed for generating the Reference SOC Profile by incorporating average speed information and road grade information. The Lower-level of HNMPC considers surrounding traffic information and the Reference SOC Profile to generate control inputs for vehicle speed and hybrid power pack power distribution ratios for a short-time prediction horizon. The Lower-level of HNMPC must take the surrounding traffic environment into account, such as traffic lights and leading vehicles, to determine speed control inputs. Additionally, it needs to consider the Reference SOC Profile to calculate power distribution control inputs. In particular, on the Lower-level, unlike the Upper-level, computation time becomes highly critical, since it is crucially focused on performing iterative optimization during a short control cycle. Therefore, this thesis proposes a NMPC method that utilizes the Intelligent Driver Model (IDM) to generate optimal control inputs within a short computation time. To validate the performance of the proposed HNMPC based EMS, we employed a simulation environment that reflects real driving environments. Simulation scenarios were designed based on real driving information measured from the roads of Seoul, Korea. Microscopic traffic simulation software was utilized to replicate the traffic environment of the real roads, and fuel efficiency simulation software was employed to model the fuel efficiency characteristics of the target vehicle. Through various simulations that incorporate real driving conditions, it was confirmed that the proposed HNMPC generates control inputs that minimize fuel consumption within a short computation time and effectively stabilizes the target vehicle. Furthermore, the performance of the proposed HNMPC is verified using the Rapid Control Prototyping (RCP) platform.