An Uncertain Environment for Energy Management in a Fuel Cell/Ultra Capacitor Hybrid Electric Vehicle Using Neural Network

This study explores the realm of energy management in fuel cell/ultra-capacitor hybrid electric vehicles (FCHEVs) in the presence of uncertainties. Resilience, alongside fuel efficiency and the slow dynamics of the fuel cell (FC) system, is considered a crucial performance criterion. Unlike previous research that mainly aimed at achieving optimal performance in deterministic settings, this study considers uncertainties affecting power generation, conversion, and demand. Errors can arise from various factors, including variations in operational conditions, estimation, and modeling. Neglecting such uncertainties in real-world scenarios can lead to increased operational costs, reduced overall system efficiency, performance failures, and even impractical outcomes due to violations of constraints. To address this, a Neural Network (NN) based energy management system (EMS) is proposed to ensure optimal yet resilient performance when input parameters are unknown. The suggested NN-based approach shields the system's performance from challenges related to viability and optimality by incorporating uncertainty into the cost function and constraints. Introducing conservatism level parameters, attributed to the NN approach's reputation for conservatism, enables more nuanced decision-making and lowers performance costs.