Optimized Sigmoid-Based Complete Ensemble Empirical Mode Decomposition for Energy Management in Hybrid Electric Vehicles

Abstract

This article proposes a sigmoid-based particle swarm optimization (PSO) for a complete ensemble empirical mode decomposition (SPSO-CEEMD) applied to the energy management system of a hybrid electric vehicle. The low-frequency power demand, to be supplied by the lithium-ion battery (LIB) and internal combustion engine (ICE), is calculated by the CEEMD, while sigmoid functions define the ICE reference, avoiding discontinuities in the control strategy and limiting the response frequency in the implementation of power, velocity and angle control loops for the ICE butterfly valve actuator. High-frequency demand is handled by the ultracapacitor (UC), which controls the dc-link voltage. The sigmoid functions are optimized to reduce the ICE fuel consumption and the LIB aging, considering ICE emissions as constraints in the PSO. To make the UC available in next peak demands, its terminal voltage restoration is relaxed by a phase-lag compensator (PLC) tuned to actuate only after power delivers, which reduces the influence in the LIB dynamic. Experimental and numerical results under the HWYCOL, SC03, and a Brazilian real-world drive cycles show that SPSO-CEEMD reduces the total operational cost, LIB stress and aging compared to state-of-the-art strategies. Despite larger UC voltage restoration error with the PLC, LIB power dynamic is not significantly affected, increasing its lifetime by 2.74% and 10.96% compared to traditional PI and low-pass filter strategies, respectively. Moreover, the total operational cost is reduced by 18.28% and 47.54% in relation to the exclusive operation strategy and interval type-2 fuzzy logic control adapted from the literature.