Improvement of LQR active anti-roll control of a single-unit heavy vehicle by means of a trained artificial neuronal network

Abstract

In this paper, a neuronal network is used to improve a Linear-Quadratic Regulator (LQR) active anti-roll control applied to a single-unit heavy vehicle suspension with linear and non-linear side force model. First, to keep the normalised rollovers between front and rear axles, equal to or below unity, the LQR control is proposed. After that, the training data collected from this controller are used as a training basis of a neuronal regulator. The artificial neuronal network controller is thereafter applied for the non-linear side force model, and it gives more satisfactory results than the LQR.