The use of wireless communications for real-time control applications poses\nseveral problems related to the comparatively low reliability of the communication channels.\nThis paper is concerned with adaptive and predictive application-level strategies for\nameliorating the effects of packet losses and burst errors in industrial sampled-data\nDistributed Control Systems (DCSs), which are implemented via one or more wireless\nand/or wired links, possibly spanning multiple hops. The paper describes an adaptive\ncompensator that reconstructs the best estimates (in a least squares sense) of a sequence of\none or more missing sensor node data packets in the controller node. At each sample time,\nthe controller node calculates the current control, and a prediction of future controls to apply\nover a short time horizon; these controls are forwarded to the actuator node every sample\ntime step. A simple design method for a digital Proportional Integral Derivative (PID)-like\nadaptive controller is also described for use in the controller node. Together these\nmechanisms give robustness to packet losses around the control loop; in addition, the\nmajority of the computational overhead resides in the controller node. An implementation of\nthe proposed techniques is applied to a case study using a Hardware in the Loop (HIL) test\nfacility, and favorable results (in terms of both performance and computational overheads)\nare found when compared to an existing robust control method for a DCS experiencing\nartificially induced burst errors.
Loading....