Stochastic Subspace Identification Applied to the Weave Mode of Motorcycles

This paper presents a safe and practical method for the identification of the weave mode of motorcycles without the need for the test rider to provide a deliberate lateral input to excite a large perceptible weave response. The solution utilizes stochastic subspace identification (SSI) and relies on the smooth surface of the road under normal steady-state running conditions to randomly excite the steering system. Three SSI variants: covariance (COV), unweighted principal component (UPC), and the canonical variate analysis (CVA) are outlined and pole selection via stabilization diagrams is discussed. Then a motorcycle test protocol necessary to collect quality data for identification analysis is described. Strong correlation between stochastic identifications and traditional impulse-based weave testing of several straight running motorcycles under multiple trim states is shown. Because of the ability to use data collected under normal steady-state running conditions, the proposed stochastic technique has the potential for allowing the identification of weave modal properties under trim state conditions that are not possible with traditional weave testing, like hands-on the handlebars in straight running or when the motorcycle is cornering. Results from identifications under these hands-on trim states are presented, demonstrating the potential for deeper understanding of these conditions.