Performance Analysis and Scheduling Optimization of Networked Control Systems
|School||Shanghai Jiaotong University|
|Course||Control Theory and Control Engineering|
|Keywords||Networked control system Fieldbus Industrial Ethernet Real-time Reliability Sampling-data system Stability Schedulability Optimization Fuzzy logic NashEquilibrium Pareto Optimization|
Motivated by higher and higher requirements on industrial applications, modern con-trol systems are becoming more and more complex and networked, control algorithmsmore and more sophisticated. Such industrial networked control systems are of integrated,which involves measurement, process control and optimization, production scheduling,management and decision. The framework makes it possible to share process data andmarketing information plant-widely, which improves availability, operating safety, relia-bility and environment protection of the production. Clearly, introducing networks intothe control systems changes not only message transmission modes but also dynamicsof the control systems. Moreover, to guarantee real-timeliness, reliability and operat-ing safety of the control systems, new requirements of quality of service(QoS) will beimposing on the networks.In this thesis, systematic works on the configuration of networked control systems,the analysis and synthesis problems for the networked control systems were considered.The issues of networks performance optimization were also addressed. The contributionsof the works are listed in the following.1. Chapter 1 concerned with the relationship between the structures of the industrialcontrol systems and the networks. The issues of real-timeliness, reliability andsafety of the control systems, communication protocols and their realizations ofsome typical industrial control networks were also studied in the chapter.2. For di?erent working modes of the networks and the control systems, generalsampling-data models were developed in the chapter 2. Based on these hybridmodels, the stability problems for the networked regular and singular linear controlsystems were considered. A less conservative time-delay dependent stability wereobtained in Chapter 2 for the systems considered.3. Moreover, we also concerned with the synthesis problems of the networked regularand singular linear systems for the di?erent operating modes of the networks andthe control systems in Chapter 2.4. From Chapter 3 to Chapter 5, the network scheduling problems were considered toimprove the QoS of the networks, and guarantee the real-timeliness and reliabilityof the control systems.? Subject to the bandwidth limitation and the requirement of real-time transmis-sion, a jitter-dependent bandwidth optimization problem for networked controlsystems was considered to guarantee the performance of the control systems,reduce the bandwidth occupancy and realize the service fairness. Taken theToken-based and CAN-based networks as examples, scheduling schemes, theirreliability and network stability were considered, respectively.? A new closed-loop, distributed priority fuzzy scheduling was proposed for CANnetworks, which makes a well tradeo? between the quality of service, bandwidthutility (network schedulability) and the service range of the networks. ? Based on Nash equilibrium and Pareto optimization methods, ?exible schedul-ing schemes were developed to adapt the ?uctuation of the network load andthe change of the service requirements, which improves the adaptation of thescheduling and provides a high e?cient and ?exible network service.