Research On Key Technologies Of Train Driver's Behavior Monitoring Based On Image Recognition

Safe driving is the basic requirement of rail transit operation.At present,more and more cities have Metro.The subway operation forms a network,and the train driver team is growing.The risk of train driving on the main line will also rise accordingly,which puts forward higher requirements for the driver and driver.As a special front-line operator,the train driver undertakes the responsibilities of train driving,vehicle fault handling,emergency response,passenger service and other aspects,which plays a very important role in ensuring the safe and stable operation of the train,and improving the management ability of the crew management department of the operation unit has become the top priority.In general,the train driver is a single driver,which can not achieve timely and effective personnel mutual control management;at the same time,due to the space limitations of the cab,it is difficult to monitor the behavior of train drivers without interference with large-scale monitoring equipment.In order to improve the human-computer interaction between the train driver and the train operation control system interface,and reduce the workload of the train driver,this paper studies a train driver driving behavior monitoring system based on image recognition and its key technologies,which can effectively identify and judge the various states and behaviors of the driver when driving on the main line,and timely and quickly warn the illegal behaviors Driving safety certificate.The main contents include:According to the requirements of face recognition,fatigue monitoring and upper body recognition of train drivers,a driving behavior monitoring system is designed.Firstly,a face recognition algorithm is established and a monitoring software system is designed to realize the recognition of ORL,GT and faces95 facial data sets.Secondly,the fatigue detection algorithm based on PERCLOS value is studied,and the fatigue detection algorithm based on PERCLOS is constructed by using Kalman filter.Finally,a limb recognition algorithm based on deep learning is proposed Finally,through the monitoring management and video communication platform,monitoring,alarming and reminding functions are realized.To sum up,this paper constructs a system composed of face recognition and fatigue monitoring system,gesture recognition system,terminal management system platform,major alarm push system and video communication module,and the reliability and practicability of the system are verified by experiments.The train driver behavior monitoring system constructed in this paper can be applied to the driving monitoring of train drivers in metro operation units and the driving behavior management of train drivers in Trunk Railways such as EMUs and passenger and freight dedicated lines.The system has wide universality and strong practical application effect.