Impact detection using ultrasonic waves based on artificial immune system

This paper presents a structural health monitoring system for judging structural condition of metallic plates by analyzing ultrasonic waves. Many critical accidents of structures like buildings and aircrafts are caused by small structural errors; cracks and loosened bolts etc. This is a reason why we need to detect little errors at an early stage. Moreover, to improve precision and to reduce cost for damage detection, it is necessary to build and update the database corresponding to environmental change. This study focuses our attention on the automatable structures, specifically, applying artificial immune system (AIS) algorithm to determine the structure safe or not. The AIS is a novelty computational detection algorithm inspired from biological defense system, which discriminates between self and non-self to reject nonself cells. Here, self is defined to be normal data patterns and non-self is abnormal data patterns. Furthermore, it is not only pattern recognition but also it has a storage function. In this study, a number of impact resistance experiments of duralumin plates, with normal structural condition and abnormal structural condition, are examined and ultrasonic waves are acquired by AE sensors on the surface of the aluminum plates. By accumulating several feature vectors of ultrasonic waves, a judging method, which can determine an abnormal wave as nonself, inspired from immune system is created. The results of the experiments show good performance of this method.