基于 ReliefF-RBF的路面不平度识别算法研究

路面不平度对道路车辆行驶安全性及车辆动力学响应具有重要影响。通过将路面不平度识别与先进悬架控制结合，有望能进一步提升乘员舒适性和车辆的操纵稳定性。现有基于数据驱动的路面分类方法难以高效处理时变参数与车速，现有基于模型的路面识别算法需要已知精确车辆模型，在实际应用中面临车辆物理参数难以获得的问题。提出一种融合模型和数据驱动的路面分类算法，采用基于模型的方法反算等效路面轮廓，结合数据预处理方法，对车辆响应和反算等效路面轮廓数据进行滤波；对等效路面轮廓和响应信息进行时域频域特征计算，采用ReliefF算法进行关键特征提取，构建基于径向基函数神经网络的路面分类器，进行路面分级识别；通过仿真试验和实车试验验证了不同车辆参数和车速下所提出的算法鲁棒性。

Research on Road Roughness Recognition Algorithm Based on ReliefF-RBF

Road surface unevenness significantly affects both the driving safety of road vehicles and their dynamic responses. However, the existing data-driven methods for road surface classification struggle to efficiently handle time-varying parameters and vehicle speeds. Meanwhile, the existing model-based road surface recognition algorithms require known and accurate vehicle models, facing the challenge of acquiring vehicle physical parameters in real-world applications. This paper proposes a novel pavement classification algorithm that begins by back-calculating the equivalent pavement profile, followed by data pre-processing.Subsequently, it computes time and frequency domain features for the equivalent pavement profile and response information, and key features are extracted using the ReliefF algorithm. A radial basis function neural network is used to construct a classifier for pavement grading and recognition. Finally, the robustness of the proposed algorithm is verified through simulation tests and real-vehicle tests with different vehicle parameters and speeds.