With the increasing development of autonomous driving technology, adaptive cruise control (ACC) and lane keeping systems gradually replace direct human control of vehicles. However, differences in drivers’ experience, personality and driving habits lead them to exhibit different characteristics in vehicle following behaviour. Existing longitudinal control strategies are mostly based on standardised approaches, which fail to meet the individualised needs. For this reason, this study proposes a personalised adaptive cruise control system that better adapts to different drivers’ driving styles by analysing and modelling human driving behaviour. The driver classification module classifies the driving data (e.g., conservative, aggressive, stable) and downloads the corresponding IRL model to control the vehicle, and the dataset of the classification model is tested and updated with the collected data. The experimental results show that compared with IDM-based ACC, this system improves the speed and distance gap by 15.1% and 21.4%, respectively, which verifies the effectiveness of the method, and the classified data can more accurately meet the needs of different drivers.