A switch model approach for uego sensor diagnostics in the presence of asymmetric faults

In this paper a diagnostics approach for the universal exhaust gas oxygen (UEGO) sensor in spark ignition (SI) internal combustion engines is proposed. Due to possible additional lag or delay in the transition of the air-fuel ratio (AFR) from lean to rich or rich to lean, the sensor measurement affects the closed-loop performance and may lead to instability of the feedback system. The paper discusses each of these faults and proposes a universal switch model approach that splits the operation of the sensor into lean and rich separate models to monitor both responses and enable the diagnostics of the UEGO sensor. The approach transforms the SISO system to a two-input single-output system and the estimated parameters of the system will identify the lag and delay in both burn modes to detect, isolate and estimate the symmetric/asymmetric faults. The method is demonstrated on a simulated air fuel (AF) closed loop system regulated by a PI controller together with a Smith predictor for the delay compensation. The method is then applied on experimental data obtained by inducing faults on the sensor.