Stochastic modeling and prediction of catalytic converters degradation

Abstract

This paper proposes a stochastic model for describing the degradation process of catalytic converters over time, where the degradation is indirectly measured through the emission of complex hydrocarbons (HC) in legislated driving cycles. The proposed model is the superposition of two processes, the former being a dependent increments process which describes the actual degradation process, and the latter a white noise process which models the experimental errors. In particular, the proposed model assumes that the degradation growth in a small usage interval depends on the degradation level at the beginning of the interval, rather than on the age of the converter. The model has been applied to the real case of catalytic converters mounted on three copies of a vehicle prototype used for a development test program at the Elasis Research Center. Inferential procedures and prediction results are presented and discussed in the last part of the paper.