With the rapid advancement of intelligent connected vehicle (ICV) technology, vehicle safety and reliability are facing increasingly severe challenges. Anomaly behavior recognition, as a key approach to ensuring normal vehicle operation, has become a major research focus. This study investigates the application of large models in anomaly behavior recognition for ICVs. By integrating multimodal data and deep learning frameworks, the proposed method improves both detection accuracy and real-time performance. In the Introduction, we analyze the types of anomalies currently faced by ICVs and their potential risks. Subsequently, in the Methodology section, we propose a large-model-based anomaly recognition framework that employs pre-trained models to process vehicle sensor data and communication information, enabling efficient and adaptive anomaly detection. The Conclusion summarizes the advantages of the approach and outlines future research directions. This study aims to provide new insights and references for enhancing the safety protection of intelligent connected vehicles.
Zeng et al. (Mon,) studied this question.