Automotive control system diagnostics using neural nets for rapid pattern classification of large data sets

The problem of diagnosing faults in the electronic control systems now present in most new cars is considered. The authors have addressed this problem by developing an efficient data acquisition system for automotive applications which can obtain the full record of data exchanged between any electronic controller and the mechanical system in the vehicle. The task of analyzing the data sets obtained from the systems under test is essentially a classification problem and is therefore well suited to the application of neural nets. The authors present typical data extracted from the vehicles in short comprehensive tests and show how a variety of neural net paradigms have been used to classify the data as to the fault present and, in some cases, the severity of the fault.<>