Research On Intelligent Generation Method Of Train Dispatching Command Based On Early Warning Text Information

With the development of China’s economy and transportation industry,in recent years,high-speed railways have continued to grow and develop.High-speed railway traffic dispatching commands have played an extremely important role in ensuring the safety,highspeed,and on-time operation of trains.In the traffic dispatching command,writing and issuing train dispatching orders is one of the most time-consuming and energy-intensive tasks.At present,train dispatching orders in China are mainly made manually by dispatchers,and are communicated to relevant departments or personnel through computer dispatching order software,fax or telephone.The drafting of dispatching orders takes up a lot of time for dispatchers.Especially in the emergency handling process under abnormal conditions,the dispatcher needs to receive and process a large amount of information,comprehensively and efficiently grasp the system operating status and make correct decisions in time,and issue dispatching orders.In this case,problems with scheduling commands often occur.In view of the above problems,the paper studies the intelligent generation method of dispatching commands,mainly from the following three aspects:(1)The railway dispatching mechanism,dispatcher and dispatching order are analyzed systematically.After a detailed analysis of the functions and work contents of each professional dispatching room and professional dispatcher,the main characteristics,main types,release processes,and release scenarios of dispatch orders are introduced,and dispatch orders are analyzed from both qualitative and quantitative perspectives.The release of the status quo and the reasons for the problems laid the foundation for the proposal of intelligent dispatch commands.(2)Research on the intelligent generation method of scheduling commands.In response to the emergency treatment under abnormal conditions,a seq2 seq model based on a long-term and short-term memory network(LSTM)is constructed.By inputting emergency event text information,the dispatching command output is obtained.And use the dispatch command correction module to modify and optimize the five aspects of the generated dispatch command,including: mileage information,line direction information,station information,speed value information,and train number information,by constructing different correction strategies for different emergency events.Scheduling command correction module to ensure the correctness of the content of the scheduling command.(3)Finally,the actual data of Guangzhou Railway Group was used to verify the model.The results show that the seq2 seq model based on LSTM can effectively use early warning text information to generate scheduling commands with an accuracy of 90.1%.Using ROUGE criteria to evaluate the final results,the results show that: Rouge-1 is 0.859,Rouge-2 is 0.783,Rouge-L is 0.812;The modified scheduling command of the module is 19.0%,18.5%,and37.9% higher on the Rouge-1,Rouge-2,and Rouge-L indicators than the unmodified scheduling command,which verifies the effectiveness of the method.