A driving-cycle predictive control approach for energy consumption optimization of hybrid electric vehicle

Abstract

This paper proposes an adaptive control strategy of fuel consumption optimization for hybrid electric vehicles (HEVs). The strategy combines a moving-horizon-based nonlinear autoregressive (NAR) algorithm, a backpropagation (BP) neural network algorithm, and an equivalent consumption minimization strategy (ECMS) method to reduce energy consumption. The moving-horizon-based NAR algorithm is applied to predict the short future driving cycle. The BP neural network algorithm is employed to recognize the driving cycle types, which provides the basis for the adaptive ECMS. Based on the abovementioned approach, the power split of the fuel and electric system is determined in advance, and the optimal control of energy efficiency is achieved. A driving experiment platform is established, taking a synthetic driving cycle composed of several standard driving cycles as the target cycle, and the control strategy is tested by the driver’s real operation. The results indicate that, compared with the basic ECMS, the A-ECMS with moving-horizon-based driving cycle prediction and recognition has better SOC (state of charge) retention and reduces the fuel consumption of the engine by 3.31%, the equivalent fuel consumption of the electric system by 0.9 L/100 km and the total energy consumption by 1 L/100 km. Adaptive ECMS based on driving cycle prediction and recognition is an effective method for the energy management of HEVs.

Review of Energy Management Methods for Fuel Cell Vehicles: From the Perspective of Driving Cycle Information

“…Previous studies have shown that driving pattern can greatly influence the effectiveness of EMMs [56][57][58]. For example, in the urban driving pattern, vehicles start and stop frequently, while in the highway driving pattern, vehicles often drive at a constant speed.…”