Control system design with limited communication.

摘要

Sensor-based systems are becoming more prevalent in manufacturing and consumer products. In complex systems with large numbers of sensors and actuators where control occurs over a limited communication medium such as a network, it is unrealistic to model all feedback and inputs as occurring simultaneously. In such systems a controller communicates with the devices periodically due to limited access or bandwidth. To model an output feedback system with limited communication a reformulation using a state space representation is presented that takes into account the dynamics caused by communication delays. Two model types are developed, one that assumes a static access schedule and one that assumes dynamic sensor access. Given a static schedule, one formulation presented uses a constant gain vector to stabilize the system. The static access control system can also be viewed as a step output varying system. Here an estimator is designed to achieve deadbeat convergence using step varying estimator gains. A system with dynamic access is modeled as a step varying process. An estimator is used to predict the system state at the actuator update time and standard eigenvalue placement techniques are used for control. To converge the estimated and real states, a formulation that recursively calculates the estimator gain is used to ensure deadbeat convergence. Experimental results are presented showing superior performance using the controller designed for dynamic access compared to a traditional controller.