Sensor fault detection and isolation algorithm for a continuous damping control system

Abstract

This paper presents an analytical sensor fault-detection and isolation algorithm for the vertical accelerometers of a continuous damping control (CDC) system, which are essential but vulnerable components in the CDC system. Because the sensor configuration of the target CDC system does not provide sufficient redundancy for fault diagnosis, two different vehicle suspension models, a modified full car model and a roll-and-pitch plane model, are derived and combined to establish analytical redundancy. Parity equations for the vertical motion of a sprung mass are derived using the combined model and applied in order to detect faults of the vertical accelerometers. In order to isolate the faults properly in practical situations, this paper proposes a fault isolation algorithm which calculates estimates of the vertical acceleration at the centre of gravity of the sprung mass using vertical acceleration signals and compares the residuals of these estimates. This paper also provides a detailed analysis regarding the detectability and isolability of vertical accelerometer faults. An adaptive threshold method is proposed to improve the performance of the suggested fault-detection and isolation algorithm. Simulation and test results demonstrate the feasibility and effectiveness of the proposed algorithm.