Research On Collaborative Scheduling And Optimization In Reliable Real-time EPA Network

At present, in the process automation system, such as the chemical and metallurgical control system, the applications of the Industrial Ethernet have become very popular. However, the application of EPA is limited in the communication systems of factory automation system, such as robotics, numerical control systems, and so on. Because the theoretical research of networked control system is focus on the stability of control system based on the controlled object of process automation in academic. And the research focuses on the methods to improve real-time performance or reliability respectively in engineering. As a result, it is extremely valuable to propose the real-time and reliable requirements to achieve the stable of Networked control system(NCS) in factory automation system. According to these requirements, the methods should also be researched to achieve the real-time stable communication.The MADB(Maximum Allowable Delay Bound) and MATI(Maximum Allowable Trasnfer Interval) in NCS are set as the real-time performance and reliability of Industrial Ethernet. After surveyed major research issues in industrial Ethernet real-time and reliability, the research is conducted based on the EPA, which is the first international standard in the area of industrial automation in China. Combined with the fault detection and recovery methods described in the DRP (Distributed Redundancy Protocol), the collaborative scheduling and optimization of reliable real-time communication in EPA network is presented. The achievements of research obtained include:1. According to the EPA communication schedule mechanism, the MADB is set as the real-time parameter. The EPA delay model is established, and the uncertain factors in this model is analysed. In order to meet the system reliable communication requirements, the DRP fault recovery methods are introduced into the EPA network. The EPA delay model is updated after the disturbance of DRP data introduced. The need for DRP and the EPA collaborative scheduling is verified by the comparison of EPA delay models before and after the introduction of disturbance of DRP data.2. Respectively, according to DRP fault detection mechanism of the system, the self-healing time model is established, and it is proved that the fault recovery prolongs because of the high loading EPA real-time data, as well as the decrease of reliability. The need for DRP and the EPA collaborative scheduling is verified for EPA system to achieve the reliability of NCS.3. Based on these necessities, the optional intervals of the DRP macrocycle and the EPA macrocycle are selected. The different collaborative scheduling algorithms are provided to achieve the collaborative scheduling of the DRP and the EPA, according to the mathematical relationships of these different intervals. The real-time performance is improved by the conflict of the DRP and the EPA data and storage of EPA data in switch avoided.4. According to the DRP recovery mechanism and collaborative scheduling algorithm of the DRP and the EPA, the DRP self-healing time models before and after the collaborative scheduling are compared. Moreover, these models in conjunction with the experimental data, which verified that the EPA delay time and DRP self-healing time are less than MADB and MATI respectively. So the real-time performance and reliability are achieved by collaborative scheduling basically.5. On the basis of achieving the collaborative scheduling algorithm of the DRP and the EPA, respectively, the model between the residual synchronization error and the DRP self-healing time and the model between the residual synchronization error and the EPA delay are researched. Based on that, combined with experimental data, analysis shows that an effective DRP self-healing time and the EPA delay optimization is the decrease of the residual error, when it is at the specified interval.6. Based on the validity provement of decrease of the residual synchronization error, a large-scale ring network clock synchronization method is researched. Based on existing boundary and transparent transmission algorithm, the backup path in the ring topology is used as a feedback path. The linear feedback and path-weighted feedback algorithm are introduced. On this basis, the residual error models of synchronization are developed. By simulation and test verifications prove that two algorithms achieve the optimization of the EPA delay and the DRP self-healing time.Finally, all the contents of this study were summarized and analyzed, and put forward ideas and suggestions for further research.