A vector space solution to incipient sensor failure detection with applications to automotive environments

An efficient method of sensor diagnosis in electronically controlled IC engines is demonstrated. The approach taken is based on the failure detection and isolation (FDI) theory, and the resulting diagnostic systems can be implemented in on-board digital computers of passenger vehicles. The proposed diagnostic system, referred to as the binary phase detection filter (BPDF), uses a pair of failure-sensitive filters. Based on the reference model of an internal combustion engine, BPDF not only isolates the failed sensor, but also estimates the associated amount of failure. A wide variety of sensor failures, including soft partial types (e.g. calibration or bias drift), can be detected using BPDF