关键词: 制动意图, 可拓学, 制动风格, 熵权法, 人-车协同

Abstract:

An extension theory-based identification method incorporating braking style correction was proposed to address the challenge of accurate braking intention recognition under complex operating conditions. The method integrated brake pedal force, pedal travel, and deceleration information to construct a human-vehicle cooperative extension analysis domain. The entropy weight method was employed to dynamically assign feature weights, and a braking style coefficient was introduced into the correlation function to correct individual differences. The multi-condition simulations show that accurate recognition is achieved in congested, red-light, and emergency braking scenarios; compared with not considering braking style, identification accuracy is significantly improved, with an average increase of 28.04% under aggressive style. The braking style coefficient exhibits significant scenario dependence: the correction coefficient range reaches 20.72% under congested conditions, where individual style divergence is most pronounced; during anticipatory stops, a consistent aggressive tendency is observed; and under emergency braking, all styles tend toward standardization.

Key words: braking intention, extenics, braking style, entropy weight method, human-vehicle cooperation