Output details

Nonlinear PCA with Local Approach for Diesel Engine Fault Detection and Diagnosis

New theoretical ideas are presented around nonlinear PCA, introducing Basseville’s statistical local approach to produce normally distributed scores, thereby easing fault detection. Our laboratory is the first to apply statistical-fault diagnosis to internal combustion engines. The paper includes unique experimental test results for a Volkswagen 1.9L turbocharged diesel engine. The research programme was supported by a joint EPSRC grant (EP/C005457/1) with Mechanical Engineering, £1M of engine test equipment from the Virtual Engineering Centre, and a supply of catalytic convertors from General Motors, USA. Results show enormous potential for enabling on-board diagnosis to meet increasing stringent legislation on automotive emissions.