Decentralised sliding mode controller design based on hybrid approach for interconnected uncertain non-linear systems

A new type controller, recurrent fuzzy neural networks-fuzzy-sliding mode controller (RFNN-FSMC), is developed for a class of large-scale systems with unknown bounds of high-order interconnections and disturbances. The main purpose is to eliminate the chattering phenomenon and to overcome the problem of the equivalent control computation. The proposed controller, which incorporates the recurrent fuzzy neural network (RFNN), fuzzy logic controller (FLC) and the methodology of sliding mode control (SMC), can eliminate chattering using a fixed boundary layer around the switch surface. Within the boundary layer, where the FLC is applied, the chattering phenomenon, which is inherent in a SMC, is avoided by smoothing the switch signal. Moreover, to compute the equivalent controller, a feed-forward RFNN is used. The stability of the whole system is analysed via the Lyapunov methodology. A bio-inspired algorithm called ant colony optimisation (ACO) is employed to calculate some key controller parameters. The effectiveness and efficiency of the proposed controller and optimisation method were tested using highly interconnected non-linear systems as examples.