Research on Tire Monitoring Data Prediction Based on LSTM-Informer Model

Tire health is an important factor to ensure vehicle driving safety and performance, and there is a complex temporal correlation and nonlinear relationship between the dynamic change of tire monitoring data and the degree of damage. In order to solve the problem that the existing methods are difficult to take into account both short-term dynamic changes and long-term trend modeling, this paper proposes a hybrid time series prediction method based on Long Short-Term Memory Network (LSTM) and Informer model for multivariate joint prediction of critical states of tires. The model uses LSTM to capture short-term fluctuation characteristics, and uses Informer to efficiently model the impact of long-term dependence and abnormal changes on damage accumulation. The experimental results show that compared with the traditional LSTM, Transformer and CNN models, the proposed LSTM-Informer model has significant advantages in prediction accuracy and robustness. The research results provide technical support for the development of intelligent tire pressure monitoring system (TPMS), and provide a scientific basis for real-time assessment and prediction and early warning of tire damage.