Research on an Intelligent Optimization Energy Management Strategy Based on Long-Term Multi-Dimensional Feature Recognition for Off-Road Vehicles

Abstract

Off-road vehicles, due to their large mass, complex road conditions, and drastic power demand variations, often face frequent engine start-stop and torque fluctuation issues under low state-of-charge conditions with traditional high-power engine configurations, leading to
degraded NVH (Noise, Vibration, and Harshness) performance and severely impacting the driving experience. To address the NVH issues of
range-extended electric off-road vehicles under low battery storage conditions, this paper proposes an intelligent optimization energy management strategy based on multi-dimensional long-cycle feature recognition. The strategy first introduces multi-dimensional long-cycle features
including congestion condition factors, driving behavior factors, mountain-road driving factors, and charging habit factors, constructing a
comprehensive correction factor system to perform multi-dimensional feature-based adaptive adjustment of the reference SOC curve, thereby
reducing unnecessary engine start-stop events at the source. Secondly, to address the challenges of multi-dimensional computation software
being difficult to implement in mass production and with excessive computational load on vehicle controllers, the multi-dimensional intelligent optimization algorithm is simplified for engineering application, enabling autonomous learning capability and real-time operation on
production controllers. Simulation results show that, compared with rule-based control strategies, the proposed strategy reduces overall fuel
consumption by approximately 4% while significantly reducing engine start-stop frequency, effectively improving vehicle NVH performance.