Network Reconfiguration And Performance Optimization Of Train Communication Ethernet

Due to the increasingly massive and complex data carried by train communication network(TCN),the Train Communication Ethernet(TCN Ethernet)has become a crucial and promising direction of TCN for its high bandwidth and strong compatibility.However,because of the rapidly enlarging scale and the complex functions of the communication system,the shortcomings of TCN Ethernet are becoming increasingly prominent,such as the weakness in traffic scheduling and the poor management in topology.It is very easy to cause performance degradation or failure such as abnormal traffic transmission and communication link interruption,leads to errors in control signals and even system dysfunction,and ultimately endangers the safe and reliable operation of the train.The network reconfiguration optimization technology is a novel design method based on fault diagnosis and performance analysis,which abstracts and decomposes specific network resources and functions,and reasonably allocates and designs functional units according to the required optimization objectives.Considering these advantages,this novel technology benefits networks with faults in positive recovery,performance optimization,and suppression of the impacts induced by the fault.Therefore,in order to ensure the efficient and safe operation of the train and meet the higher requirements of the TCN Ethernet in overcoming faults and optimizing performances,the network reconfiguration optimization technology is worth further research as an intelligent fault-tolerant control method with fault self recovery and performance optimization.Aiming at the fault recovery and performance optimization,the network reconfiguration and optimization strategy is proposed for the TCN Ethernet,which focuses on the self adjustability in network sources and the self recovery of topology routing planning.The strategy proposed in this dissertation includes three aspects,that is,the network resource pre-scheduling reconfiguration,the sub-network resource dynamic rescheduling,and the topology routing reconfiguration.The work and corresponding contributions in this dissertation are mainly presented as follows.1.To optimize the synergic pre-scheduling allocation among several kernel devices in the TCN Ethernet,the pre-scheduling regonfiguration strategy based on Bacterial Foraging Optimization with Self-adaptive Chemotactic strategy(SCBFO)is proposed.First,a control model of real-time resources within TCN Ethernet systems is established based on time-triggered TCN Ethernet architecture.Then the period and the initial phase are selected as key factors in the pre-secheduling reconfiguration control,and corresponding constrants and optimal targets are founded.Finally,the SCBFO pre-scheduling regonfiguration strategy is accomplished,with which the reconfigured results,the searching speed,and the searching sability are all taken into account.2.Since the fast dynamic-scheduling self-tuning capability is required by Ethernet Consist Network(ECN)subnets,a dynamic-scheduling reconfiguration strategy based on Multi-Objective Fuzzy Particle Swarm Optimization(MOFPSO)is proposed.According to the independent characteristics of ECN subnets,a structure analysis method is first established for star subnets where the network switch plays a kernel role.Then,according to the switching transmission foundation of ECN subnets,the scheduling control of real-time traffic is modeled and sorted in time domain,and the link transmission is divided into variable time windows.Based on these works,the constraints and optimal objectives of the dynamic-scheduling reconfiguration are formed.Finally dynamic scheduling reconfiguration strategy based on MOFPSO is formed,which can quickly complete the dynamic-scheduling reconfiguration optimization of the scheduling table in ECN subnets under abnormal traffic conditions.3.In order to realise the optimization of topology routing planning under faults,a novel strategy of network topology routing reconfiguration based on Differential Evolution hybrid Tabu algorithm(TDE)is proposed.Based on the real-world network structure of the TCN Ethernet,a sparsity model is established as a mathematical description for the network topology,which includes node state matrix,port connectivity matrix and directed link matrix.Comprehensive evaluation indexes for the network topology routing performances are designed,including link load balancing,forwarding delay and delay jitter.Finally,the network topology routing reconfiguration strategy based on TDE is proposed to quickly and effectively deal with the network topology mutation caused by link failures.4.To verify effectiveness of above proposed theory in practical applications,a network reconfiguration optimization experimental platform based on TCN Ethernet is established.Combined the proposed reconfiguration optimization strategy of the TCN Ethernet with the actual metro train Based on the Train Real-time Data Protocol(TRDP),the experimental platform of the TCN Ethernet is built up,and means of real-time monitor and control of network resources and transmission links are designed.Through simulation experiments and experimental tests,the pre-scheduling reconfiguration of network resources,the dynamic-scheduling reconfiguration of subnet resources and the topology routing reconfiguration strategy are all proved effective,which indicates that the network reconfiguration optimization strategy is able to provide a novel optimization scheme in the research of TCN Ethernet fault-tolerant control.