INTRODUCTION
In modern complex control systems, sensor, actuator and control units are more and more connected by communication networks instead by traditional point-to-point connections. The significant advantages over traditional control architectures include reduced wiring and cost, increased modularity, easier maintenance and system diagnosis as well as high flexibility and reconfigurability. Examples are manufacturing systems, automobiles, aircrafts, and mobile phone networks. This novel ”Networked Control System” (NCS) architecture poses a number of new research challenges to the control system design as the signal transmission over a communication network can no longer be regarded as ideal. Untreated transmission time delay, data drop outs, and limited communication resources may cause undesirable behavior of the system, while the exact transmission parameters are possibly not exactly known. The large application potential of NCS has driven the recent great attentions of the control, communication and signal processing communities into this topic.
DIRECTIONS OF RESEARCH
Exploitation of the Limited Computational Power in the Plant Side
In real NCSs, although the controller is remotely placed, in the plant side there is some limited computational power. This can be exploited to locally implement computationally inexpensive control actions. Based on this fact a novel distributed control approach has been developed with its major advantages summarized into :
- Stability for arbitrary large time-varying delay and packet loss
- Improved performance compared to standard delay-dependent methods
- Reduced sensitivity of the input-output behavior to the transmission effects
Performance Guaranteed Stochastic Control Approach
The signal transmission behavior of used communication network usually determines the control performance of NCS. However, the communication quality depends on network loadings, which alter from time to time upon the data size and network participants. An intuitive way to increase the NCS control performance is to synchronize the system controller with the variations of network.
- Stochastic stability for random delay
- Switching controller to transmission delay
Optimal event-triggered Sampling and Control
In many future applications of networked control systems, we are facing scenarios where a vast number of distributed sensors and actuators can only communicate wirelessly with a central controller. Due to limited battery capacity or constraints on the transmission bandwidth, it is favorable to keep communication at a minimum.Therefore, a major issue in networked control systems is to find a tradeoff between minimizing communication in the feedback loop and maximizing control performance. Whereas most digital control systems use time-triggered sampling schemes which implies that updates are sent periodically over the network, a much more natural choice to meet such tradeoff is an event-triggered sampling scheme. This is because event-based control reacts instantaneously on abrupt changes of the system, but does not send updates, when the system state is small. The main questions to be answered in this project are:
- How to design scheduling and control schemes in an optimal way?
- Does there exist some kind of separation principle of scheduling and control?
- What can be said about robustness issues of an event-based controller in the presence of packet loss and time delay?
In order to tackle these questions, we resort to the following theoretical concepts:- Stochastic control
- Markov decision process
- Switched systems
- Asynchronous sampling and optimal stopping times
Arbitrated Networked Control Systems
This project concerns the analysis and synthesis of a class of cyber-physical systems where the objective is to realize high control performance in applications with distributed embedded systems (DES). DES are ubiquitous, driven by their increasing networking capabilities and ever-declining costs of computing and communications devices. DES are highly attractive due to the fact that they radically enhance the capabilities of the underlying system by linking a range of devices and sensors that will allow information to be collected, shared, and processed in unprecedented ways. Deploying control and non-control applications on a modern DES, which uses advanced processor and communication technology, introduces a host of challenges in their analysis and synthesis, and leads to a large semantic gap between models and their implementation. The presence of multiple processors and communication buses in a DES necessitates arbitration and the use of schedules with disparate protocols so as to serve multiple physical applications. This research, carried out in collaboration with the Active-Adaptive Control Laboratory at MIT, concerns the development of a class of Arbitrated Network Control Systems (ANCS) that overcomes these challenges by providing a cyber-physical framework. Fundamental to this framework is a co-design of both the control methods as well as the architecture of the DES. Main features of the ANCS include (i) hierarchy and (ii) multi-modality, both of which are necessary components in the design of a DES where advanced controllers need to be implemented. How adaptive controllers will be implemented in DES so as to ensure a satisfactory Quality of Control and minimal resource utilization will be explored. Some highlights of our recent results can be found here:
