Performance Monitoring for Vehicle Suspension System via Fuzzy Positivistic C-Means Clustering Based on Accelerometer Measurements

This paper focuses on fault detection and isolation for vehicle suspension systems. The proposed method is divided into three steps: 1) confirming the number of clusters based on principal component analysis; 2) detecting faults by fuzzy positivistic C-means clustering and fault lines; and 3) isolating the root causes for faults by utilizing the Fisher discriminant analysis technique. Different from other schemes, this method only needs measurements of accelerometers that are fixed on the four corners of a vehicle suspension. Besides, different spring attenuation coefficients are regarded as a special failure instead of several ones. A full vehicle benchmark is applied to demonstrate the effectiveness of the method.