Abstract: The total vehicle mass is a critical parameter in vehicle dynamics modeling and control system design, directly influencing the dynamic performance, energy efficiency, and safety control capabilities of electric vehicles. In distributed-drive electric vehicles, due to the rapid response and precise torque regulation of electric drive systems, the impact of total vehicle mass on control accuracy is particularly pronounced. However, vehicles frequently encounter complex operating conditions such as variable payloads during operation. Fixed-mass assumptions can easily lead to model mismatch, thereby compromising the stability and robustness of control systems. To enhance model adaptability and estimation accuracy, this study develops an adaptive Kalman filter to achieve robust vehicle speed estimation, thereby providing reliable speed reference information for mass estimation. Subsequently, based on the vehicle longitudinal dynamics model, an online vehicle mass estimation method is constructed using a recursive least squares (RLS) algorithm with a forgetting factor. By approximating resistance terms, this method reduces dependence on unmeasurable parameters and improves practical feasibility. Simulation results demonstrate that the proposed method achieves high-precision, rapidly convergent mass estimation under typical acceleration scenarios, providing theoretical support for optimizing control strategies in distributed-drive systems.

Key words: distributed drive electric vehicle; recursive least squares method; vehicle mass estimation; adaptive speed estimation; vehicle dynamics

摘要： 整车质量是车辆动力学建模与控制系统设计中的关键参数，直接影响电动汽车的动力 性、经济性与安全控制性能。在分布式驱动电动汽车中，由于电驱系统响应快速、扭矩调节 精细，整车质量对控制精度的影响更为显著。然而，车辆在运行过程中常面临载荷变化频繁 等复杂工况，固定质量假设易导致模型失配，进而影响控制系统的稳定性与鲁棒性。为提升 模型适应性与估计精度，文章通过构建自适应卡尔曼滤波器完成对车速的鲁棒估计，为质量 估计提供可靠车速参考信息；进而基于车辆纵向动力学模型，采用带遗忘因子的递推最小二 乘（RLS）算法构建整车质量在线估计方法。该方法通过阻力项近似处理，降低了对不可测 参数的依赖，提升了实际可行性。仿真验证结果表明，该方法在典型加速工况下可实现高精 度、快速收敛的质量估计，为分布式驱动系统优化控制策略提供了理论支撑。

关键词: 分布式驱动电动汽车；递推最小二乘法；整车质量估计；车速自适应估计；车辆动力学