: A novel discrete-time (DT) temperature control scheme with the injection of high-frequency dither
signal is proposed to regulate thermal disturbance of a continuous-time (CT) water cooling system. The
control scheme is designed with a two-stage technique, which is composed of an off-line identification stage
and an on-line adaptive control stage. The dithered plant is identified by a Random-Simplex Network (RSN)
approach that guarantees global optimal modeling for the off-line identification stage. In which, this CT
nonlinear delayed system with disturbance is represented in the form of a discrete-time Two-Stage Neural
Network (TSNN) with tapped delays. Furthermore, we provide DT theories to ensure the convergence and
uniformly ultimate boundedness of the closed-loop system in the sense of Lyapunov. Experimental results
show that the proposed strategy has excellent control performance and that dither successfully suppresses the
effect of the nonlinear dynamics in the system.
