An artificial neural network architecture for application in general diagnostics

The application of computing rapidly advanced the test field, particularly the development of automatic test equipment (ATE). Typically the application of computers has been focussed on the control of instrumentation, allowing a large number of complex tests to be performed, with relatively little human interaction. The main development in computing that has propagated to the test environment has been the increase in processing power, allowing a greater number of faster, more complex tests and diagnosis to be performed. One area of development, that has not been adopted so rapidly, is the application of Artificial Intelligence (Al). The foundations of modern artificial neural network (ANN) theory were developed half a century ago, and although the application of neural networks can be difficult, their use is becoming increasingly widespread. This paper discusses a methodology that will allow Al to be applied, in an ad hoc fashion, in to the contemporary test arena, eliminating the need to for a link between LRU design and ANN development.