Takagi–Sugeno Fuzzy Observer-Based Magnetorheological Damper Fault Diagnosis Using a Support Vector Machine

Abstract

In this article, a magnetorheological (MR) damper fault diagnosis algorithm is proposed. Recently, various studies have been proposed fault diagnosis methods for this damper system, however, these methods use a displacement sensor in addition to the two accelerometers mounted on the commercial vehicle. The use of this sensor configuration limits the application of the algorithm proposed so far to commercial vehicles. In order to overcome this limitation, this article proposes an MR damper fault diagnosis algorithm that uses only two accelerometers. Based on these sensors, the states of the vehicle suspension system are estimated by a Takagi-Sugeno (T-S) fuzzy unknown input observer. This observer scheme can estimate the states of the damper even under conditions affected by damper hysteresis and unknown road elevation. Using the Lyapunov stability theorem, the stability of the proposed observer with an unmeasured premise variable is verified. In addition, a support vector machine (SVM) classifier is used to determine damper condition without empirically set thresholds. In this article, a fault flag is generated by data-driven machine learning algorithms, reducing the design effort while at the same time achieving optimal performance. The proposed algorithm is verified using a quarter-car test rig and test results confirm that the proposed algorithm exhibits robust performance in various road conditions. Consequently, the MR damper fault diagnosis algorithm proposed in this article can reduce the effort in designing a diagnosis algorithm while using inexpensive production sensors applied to a vehicle.