Unsupervised Classification of Acoustic Emission Signal to Discriminate Composite Failure at Low Frequency

Abstract

The use of acoustic emission (AE) for damage assessment and detection technique in structural engineering is widely used and has earned a reputation as one of the reliable non-destructive techniques. AE source is produced based on the elastic wave propagation through the specimen which converted into the electrical AE signal by the AE sensors. Certain AE features belong to the signal allow their use to discriminate mode of damage in a composite material. However, the challenge encountered during analysis of AE signals attributed to the techniques like pattern recognition and classification method. In this paper, several orientation of laminated fiber specimens are undergoing tensile test. According to the information from tensile load test, significant features were monitored namely amplitude and energy in order to execute the classification method. The AE data are successfully cluster by k—means algorithm.