H control for uncertain linear system over networks with Bernoulli data dropout and actuator saturation

• DOI: 10.1016/j.isatra.2017.12.024
• Publication Date: 2018-01-16T22:11:44Z
• Authors: Jimin Yu; Chenchen Yang; Xiaoming Tang; Ping Wang
• Keywords: Bernoulli distribution; Linear matrix inequality; Linearization; Dropout (neural networks)

This paper investigates the H∞ control problems for uncertain linear system over networks with random communication data dropout and actuator saturation. The random data dropout process is modeled by a Bernoulli distributed white sequence with a known conditional probability distribution and the actuator saturation is confined in a convex hull by introducing a group of auxiliary matrices. By constructing a quadratic Lyapunov function, effective conditions for the state feedback-based H∞ controller and the observer-based H∞ controller are proposed in the form of non-convex matrix inequalities to take the random data dropout and actuator saturation into consideration simultaneously, and the problem of non-convex feasibility is solved by applying cone complementarity linearization (CCL) procedure. Finally, two simulation examples are given to demonstrate the effectiveness of the proposed new design techniques.