Data-Based Predictive Hybrid Driven Control for a Class of Imperfect Networked Systems

A data-based predictive hybrid driven control (DPHDC) approach is presented for a class of networked systems compromising both computation and communication delays, packet dropouts, and disturbances. First, network problems are classified in a generic way, which is used to design a network problem detector capable of detecting online current delays and packet dropouts. Second, a single-variable first-order proportional-integral based adaptive grey model [PIAGM(1,1)] is designed to predict future network problems, and to predict system disturbances. Third, a hybrid driven scheme integrated optimal small buffer (OSB) is constructed to allow the system to operate without any interrupts due to large delays or packet dropouts. Furthermore, the OSB size is online optimized using an adaptive grey fuzzy cognitive map technique. Forth, a prediction-based model-free adaptive controller is developed to compensate for network problems. The DPHDC stability is theoretically proved, while its effectiveness is demonstrated through a case study.