Train Driver State Detection System Based on PCA and SVM

Train drivers play an important role in the process of train running and their driving state directly determines whether the train arrives safely and in time. Therefore, ensuring the good mental state of train drivers is related to the safety of train and passengers. Long-distance driving, short rest and high-intensity working environment may lead to a series of behaviors such as inattention, dozing or fatigue, which may cause serious safety accidents during the train driver driving. In order to realize the real-time monitoring of the train driver’s state and reduce the safety accidents caused by the train driver’s own driving behaviors, this paper designs and implements the train driver’s state detection system based on PCA and SVM. First, the system processes the video captured by the train camera to extract the images of the train driver’s head posture. Second, use these images as PCA technology training samples for feature extraction, and classify the driver’s head postures through SVM technology. Third, the system recognizes the current driving state of the train driver and reminds the illegal operation behavior, ensuring the correctness of the driver’s behavior to the greatest extent. Experimental results show that the driver state detection accuracy rate reaches 86.6667%.