Sensor fault estimation based on ellipsoid bundles for dynamic point-the-bit rotary steerable systems

The dynamic point-the-bit rotary steerable system (DPRSS) achieves high-performance trajectory control in oil and gas exploitation. However, the sensor faults significantly degrade the reliability of the DPRSS. In this paper, the problem of sensor fault estimation for the DPRSS is investigated. Firstly, the DPRSS is modeled by a descriptor system with unknown but bounded uncertainties, which include measurement noises, model perturbations, and unknown inputs. Next, by integrating the decoupling technique, a sensor fault estimator is put forward based on the ellipsoid bundles. Besides, order reduction and bound calculation methods are proposed to derive intervals of the states and faults. Then, with a defined minimization criterion of the ellipsoid bundle, a sufficient condition is presented to design the parameters of the estimator. Finally, numerical simulations and experiments on a DPRSS prototype are provided to demonstrate the effectiveness of the proposed fault estimation method. •The sensor fault estimation problem is studied for DPRSS with UBB uncertainties.•Based on ellipsoid bundles, a fault estimator is designed for descriptor systems.•Tighter bounds are derived to improve the accuracy of the interval estimation.•Simulations and experiments are performed to verify the proposed method.