关键词: 车辆工程；罐车侧翻；车-路协同；多源信息融合；两阶段预警方法

Abstract: To address the issues of untimely, inaccurate, and unreliable rollover warnings for tanker trucks, a two-stage warning method based on vehicle-road collaborative perception was proposed, comprising pre-incident warning and in-process warning. In the pre-incident warning, tire pressure monitoring and vehicle-road communication were used to perceive tire and road conditions in real time, which expanded the scope and types of hazard identification and provided timely warnings in case of potential risks. In the in-process warning, the rollover threshold was dynamically estimated, and a long short-term memory network was employed, integrating with roll angle, lateral acceleration and yaw rate to predict the roll state, enabling accurate and reliable rollover warnings. To verify the performance of the proposed method, real vehicle tests were conducted in a closed testing ground. The research results demonstrate that the feasibility of pre-incident warning is verified under various working conditions, and the in-process warning achieves an accuracy of 98.65% during steering and lane-changing maneuvers. The proposed method covers the entire anti-rollover process and effectively enhances the driving safety of tanker trucks.

Key words: vehicle engineering; tanker rollover; vehicle-road cooperation; multi-source information fusion; two-stage warning method