Condition monitoring, diesel engines, and intelligent sensor processing

The research briefly outlined indicates the feasibility of a condition monitoring system in which ANNs are trained to recognise certain known faults. Such a system could provide useful early warning of known fault conditions before they have a chance to develop and become compounded. Data from any one of three sensors could be used for fault classification, but better performance was obtained as a result of decision fusion in which a set of differently trained nets were combined to form a multinet system. These findings provide further confirmation of the benefits of multiple classifier systems that have been found in a number of other application domains.