Performance Testing And Networking Optimization Of Industrial Ethernet

In recent years, industrial Ethernet technologies have made great advances in the management layer, control layer and field layer within a network. The integration of multi-layer functions and operation systems across the whole network has become the research and industry focus in the long run. In general, transmission delay and data response time should comply with the predefined upper limit in the field-level industrial applications. Considering the inherent uncertainty of Ethernet, most researchers adopt Carrier Sense Multiple Access/Collision Detect (CSMA/CD) as medium access mechanism. Current researches on the real-time performance analysis of industrial Ethernet are mostly based on the theoretical analysis or ordinary Ethernet-based equipment performance testing, which cannot reflect the operational status of different industrial fields in practice.Considering the current research absence in the industrial Ethernet real-time performance study, we design series of industrial Ethernet test platform solutions and propose a PROFINET-based industrial control network system in this thesis. Additionally, we complete the system integration, component installation and fault debug. The analysis of proposed testing methods for delay and IO Cycle are provided in detail.This thesis analyzes the technical solutions based on the optimized real-time channels and VLAN priority identification mechanisms in PROFIENT protocols. We establish the mathematical model of the delay time, both in terms of the real-time data transmission and the mixed transmission with real-time and delay-tolerant data. Furthermore, we design the testing program of the delay time and IO Cycle. By using the linear and tree topologies to test the delay effect caused by network topology and the length of data packet. The testing results are analyzed to configure the IO Cycle combing with the theoretical calculations and practical results.Based on the practical testing results, the network configuration principles of industrial Ethernet is proposed in this thesis. Considering the two-level tree topology industrial Ethernet, we establish the mathematical model of sub-network division problem by assigning different weight parameters, in order to minimize the total data traffic of backbone network and the transmission load difference between sub-networks. In addition, we improve the fitness function and selection operator of the genetic algorithms and propose another genetic simulated annealing algorithm to address the problem. The simulation results show that the total data traffic of backbone network can be decreased by 48%, and each sub-network load is balanced. The frequency of real-time data frames transmitted through the switches are reduced, which effectively improves the real-time performance of industrial Ethernet.