Time-Varying Sliding Surface Design with Support Vector Machine Based Initial Condition Adaptation

Abstract

In this paper, the incorporation of soft computing methodologies into sliding mode control structures with a time-varying sliding surface is considered. One of the main problems in the time-varying sliding surface design problem is the determination of the proper values of the design parameters, which are initial condition dependent. When the initial conditions of the system change, the parameters of the time-varying sliding surface have to be calculated again to obtain the desired performance. Therefore, a support vector machine structure is used for estimation of the parameters of the time-varying sliding surface to respond to a change in the system initial conditions, paving the way for a new integration of soft computing methodologies with sliding mode control theory. Simulations are then performed on a second order nonlinear model with an external disturbance and parameter variations to demonstrate the validity of the proposed method. It is illustrated that the sliding surface with the estimated system parameters performs better in moving from different initial conditions to the sliding phase than does the conventional sliding mode controller.