60 likes | 64 Views
This paper discusses the design of decentralized controllers for real-time interconnected systems. The aim is to ensure each system can track the reference input without knowing the exact state of other systems. The task period for each system can be dynamically adjusted to optimize control performance. The proposed approach is tested on double-tank systems, continuous stirred-tank reactors, and leader-heading robot systems with wireless sensors.
E N D
Decentralized Controller Design in Real-Time Interconnected Systems Xi Chen
Reference [1] F. Zhang, K. Szwaykowska,, W. Wolf and V.Mooney “Task Scheduling for Control Oriented Requirements for Cyber-Physical Systems”, Real-Time Systems Symposium, pp.47-56 ,2008 [2]Y. Sun and N. El-Farra , “ Quasi-decentralized model-based networked control of process systems” Computer and Chemical Engineering, pp. 2016-2029, September 2008
Idea • First: design two interconnected systems : • Double-tank Systems • Continuous Stirred-tank Reactors • Leader-Heading Robot Systems with wireless sensor • Second: Modify the system model in order to make sure: • The information of states are inexplicitly interconnected between the systems • The control signals are locally generated and are not shared between different systems • Desired state of each system is known by other systems
Idea (continue) • Thirdly: The double systems are implemented in Truetime Simulator where the sampling time variation will influence the control performance. • Discrete the model (it is not sure, because the method for decentralize adaptive control system seems to be applied only in continuous system.) • Analyze the relation between system’s stability and the sampling time variation. (this can be done even the model is not discrete)
Idea (continue) • Fourthly: • First , fix the sampling time which does not decrease the stability , and implement the control law in the systems by utilizing the decentralized control method introduced in the paper ”Exact Output Tracking in Decentralized Adaptive Control System” • Second, implement the decentralized controllers by dynamically changing sampling time for each control system, where the sampling time are calculated dynamically by optimizing the control performance
Idea (continue) • Finally, make the conclusions : • For interconnected systems implemented in real-time network or computer, the decentralized adaptive controllers could be designed to make sure each system can track the reference input without knowing the exact state of other system. • Moreover, the task period arranged to each system could be dynamically adjusted to ensure the optimized control performance • The systems could be theoretically extended to include multiple interconnected control systems co-executed in the shared plant