Research And Improvement On Real-Time Performance Of Train Ethernet

With the development of trains in the direction of intelligence,high speed and comfort,the train structure is more and more complex,and the application types of train network are becoming more and more abundant.In addition to the traditional control,surveillance,and fault diagnosis services,the passenger communication data in the train communication network is also increasing.The traditional train communication network has low bandwidth and a single network structure,which cannot meet the transmission requirements of large data volume.Industrial Ethernet has the advantages of large bandwidth,good compatibility,flexible networking,etc.It has been widely used in industrial fields.So it is also a future development direction of the train communication network.This paper is based on national science and technology support projects,and study the application of industrial Ethernet in train communication network,the main research contents and innovations of this paper are as follows:(1)Based on the real-time and reliability problems of industrial Ethernet in train network,this paper mainly analyzes the factors affecting the real-time performance of industrial Ethernet and the improvement methods.It proposes to improve the real-time performance of Ethernet through two aspects:optimized switch scheduling algorithm and time-division multiplexing communication mechanism.(2)In the aspect of switch scheduling,a scheduling algorithm combining with PQ and improved DWRR is proposed,which is also one of the innovations of this paper.In this algorithm,high-priority train control data is preferentially transmitted,and other service data is scheduled by an improved DWRR algorithm,ensuring the real-time performance of train control data and the fairness of other service data.Finally this paper uses OPNET platform to build the simulation model of train communication network and verifies that this algorithm is effective.(3)In time-division multiplexed networks,the whole network clock needs precise synchronization.In order to improve the synchronization accuracy,a single neuron PI control method based on Kalman filter is proposed.This is also an innovation point of this paper.First,Kalman filter is used to improve the accuracy of the deviation through correcting the clock deviation obtained by the clock synchronization process.Then single neuron PI control is applied to realize continuous indirect compensation of clock deviation and improve clock synchronization accuracy by adjusting the frequency of the slave clock.Finally,the algorithm is verified on the MATLAB platform.The results show that the clock synchronization accuracy is improved by this algorithm.