The study on driver's driving behavior is one of the most important parts in the domain of vehicle active safety, but it is difficult to study it directly because of high uncertainty. An algorithm is developed to evaluate a driver's driving behavior based on cloud model theory. That the driver's driving behavior has the characteristics of randomness, impulsivity and autonomy brings very great difficulty to study the driver's driving behavior directly, and the cloud model provides three numerical characteristics (Expected value, Entropy and Hyper entropy), which may describe the driver's characteristics preferably. Based on those characteristics, a set of rules is established which main refer to the cloud similarity between the different moving vehicle attitudes, so through the real-time data mining and it can accurately distinguish the driving behavior. Finally this paper constructs a data acquisition system by using an InvenSense's 6-Axis inertial measurement unit (IMU) as the center of this system, and an experiment on a real road shows that the algorithm makes the recognition results more close to the actual, efficiently identifies and warns drivers of bad driving behavior.
Keywords: driving behavior; cloud model; vehicle active safety; dynamic vehicle attitude; state monitoring