Model based failure detection of Diesel Particulate Filter

Abstract

Improvements in diesel engine technology have resulted in their expanded usage as powertrains in automotive applications. The Diesel Particulate Filter (DPF) is a common component of the exhaust after-treatment system of Diesel engines that removes the harmful Particulate Matter (PM) in the exhaust gas. To ensure that the filter is able to reduce PM levels of the diesel exhaust below regulated limits, On Board Diagnostics (OBD) of DPFs is required to provide alerts in the case of filter malfunction or failure. In the present study a method for performing the failure detection of Diesel Particulate Filter is proposed based on an adaptive model based technique. To detect a failure the coefficients of a healthy model of the pressure difference across the filter are compared with the adapted model coefficients since the presence of failure alters the dynamics of the system. This approach is robust to modeling errors, sensor noise and process variability and has OBD capability without the need of any additional sensors. The proposed approach is experimentally validated on a federal test procedure (FTP-75) drive cycle for healthy and failed filters in a heavy duty diesel engine test cell.