Minimum entropy control of non-linear TITO systems with random delays

This paper presents a control system design strategy for non-linear two inputs and two outputs (TITO) systems with random delays. In order to cast the TITO feedback control systems into a general framework, the stochastic characteristics induced by uncertain time delays are represented by the entropy of tracking errors. The performance index of the control systems is constructed and emphasised. Back propagation (BP) neural networks are employed as PID controllers to deal with both non-linearity and randomness. The convergence condition in the mean-square sense is analysed. The methodology is illustrated by simulations.