Avionics Testing with Artificial Intelligence Support

Abstract

This paper describes a concept allowing for automated testing of avionics installed on a test bench where concepts of Artificial Intelligence are used. This test bench technology provides means for fast discovery of changes in behavior of new releases of avionics software where feature corrections can lead to modified behavior of other functions. The tested avionics is installed in a simulator test bench allowing to simulate behavior and systems of the respective airplane type. The test bench allows the pilot himself to test the functionality which can be repeated by the automation system itself. The technology behind the simulator collects all the data from the test bench internal buses under different conditions to be used as a reference to identity deviations from expected behavior of the avionic systems. The Artificial Intelligence methods used in our approach include state space search algorithms for determination of required control routines and data learning related to identification of deviations in behavior.