Robust Mass Estimation for Heavy-duty Vehicles under Multi-source Disturbances

Vehicle mass estimation (VME) is a critical input for control and safety systems in heavy-duty vehicles (HDVs), with direct implications for transportation efficiency and driving safety. In real-world road transport, however, torque interruptions during gear shifting, stochastic sensor signal anomalies, and insufficient persistent excitation make conventional real-time estimation methods highly dependent on specific operating conditions and signal quality, both of which are difficult to sustain in practice. Leveraging the fact that the total vehicle mass remains time-invariant during a single transportation mission, this study reformulates the real-time VME problem into a one-shot estimation framework. The proposed approach integrates dynamic and kinematic vehicle models to decouple vehicle mass from road grade, incorporates an online data-screening strategy to retain only quality-assured samples, and performs one-shot parameter estimation using the M-estimator Sample Consensus (MSAC) algorithm once a predefined number of samples is reached. This design enables accurate mass estimation under representative driving conditions while inherently mitigating the adverse effects of gear-shift disturbances, stochastic sensor anomalies, and insufficient excitation. Sensitivity analysis demonstrates that a sampling frequency of 10 Hz combined with a dataset of 500 high-quality samples achieves maximum relative estimation errors within 5% under all tested loading conditions, while configurations of 10 Hz/400-point and 20 Hz/500-point offer alternative trade-offs between accuracy and data collection time. Simulation and real-vehicle experiments further demonstrate that the proposed method significantly outperforms conventional estimation methods across multiple cargo loading conditions and all tested scenarios. The proposed framework offers a practical and scalable solution for high-precision VME in real-world HDV operations, with potential applications in cargo monitoring, adaptive control, and active safety systems. • Innovation: One-shot framework eliminates PE dependency in HDV mass estimation. • Robustness: Suppresses gear-shift transients and stochastic sensor signal anomalies. • Applicability: Enables robust mass estimation under representative operating conditions. • Configuration: 10 Hz/500-point setup achieves ≤ 5% error under representative conditions. • Validation: Outperforms recursive benchmarks in simulation and real-vehicle tests.