A Method for Real-Time Road Surface Identification of Intelligent Tire Systems Based on Random Convolutional Kernel Neural Network

Intelligent tires have received much attention from scholars in many fields, providing a novel and effective research tool for autonomous driving and advanced vehicle control. In this paper, we proposed a novel approach that integrates intelligent tires with machine learning to achieve real-time road surface recognition. A triaxial accelerometer is used as the core of the wireless transmission intelligent tire system, and the X-, Y-, and Z-axis acceleration data are used as the sample set for the neural network model training during the multispeed and multisurface driving process. By conducting the comparative analysis of other neural network models adopted in this paper and related research results, the method for intelligent tire road surface recognition based on an enhanced random convolutional kernel neural network model is proposed. Due to its capability to achieve real-time acquisition of pavement adhesion coefficients under various vehicle speeds and road surface conditions, the superiority and strong generalization ability of the MiniRocket model in learning temporal signal features have been validated. Moreover, the accuracy and robustness of identification method have also been enhanced. This research will contribute to advancements in vehicle dynamic control, and provide real-time accurate road surface information for the motion control of vehicles under complex working conditions. It aims to enhance the safety of autonomous vehicles.