Energy flow analysis and control strategy optimization for a hybrid wheeled all-terrain vehicle under complex driving scenarios

The energy management of Hybrid Wheeled All-Terrain Vehicles (HWATVs) operating in unstructured environments faces significant challenges due to stochastic power demands, the scarcity of real off-road data, and the absence of representative standard driving cycles. To address these intertwined constraints, this study establishes a physics-informed, data-driven optimization framework that bridges full-spectrum energy flow testing with adaptive control. Specifically, a high-fidelity powertrain test bench is first established using the Design of Experiments (DOE) method to precisely quantify the multi-physical energy loss characteristics of the engine, motors, and battery. Using the measured loss maps as physical anchors, four typical off-road patterns are then directly extracted from 23,559 s of real-vehicle data via Principal Component Analysis (PCA) and K-means clustering, and stochastic validation cycles are synthesized via Markov chains. For each identified pattern, the Grey Wolf Optimizer (GWO) is adopted to calibrate the mode-switching thresholds and SOC bounds. These calibrated parameter sets are subsequently deployed through an online driving condition recognition mechanism that enables real-time adaptive switching. Hardware-in-the-Loop (HIL) validation under both typical and stochastic driving cycles demonstrates that, compared with traditional rule-based strategies, the proposed approach exhibits improved robustness in stochastic environments, reducing fuel consumption by 0.99%–7.01% and decreasing battery charge/discharge cycling frequency by 2.14%–15.30%. The energy-flow post-analysis reveals that the savings originate from the coupling of high-efficiency engine operation with minimized series-path conversion losses, providing a practical foundation for special-purpose off-road vehicles in GPS-denied environments.