An approach for state observation in dynamical systems based on the twisting algorithm

This paper discusses a novel approach for state estimation in dynamical systems, with the special focus on hydraulic valve-cylinder drives. The proposed observer structure is based on the framework of the so-called twisting algorithm. This algorithm utilizes the sign of the state being the target for the observation, not directly implying its applicability for state observation. However it is shown that in the discrete case this approach may be applied to obtain a real time derivative using an intermediate integral process. Furthermore, different types of gains for the algorithm are suggested and analyzed via Lyapunov arguments, and a special version of the observer is proposed specifically for hydraulic drives utilizing online available plant information. The proposed observer with different observer gains are subjected to simulation studies in order to evaluate the exact properties, and results confirm that the proposed observer structure is easily implemented, produces very limited phase shift and is highly robust toward measurement noise.