Additive Manufacturing (AM) is a revolutionary innovation that has transformed the manufacturing industry and has applications in a wide variety of fields. The aerospace industry uses metal AM because of the design freedom it offers and its ability to produce lightweight parts. However, like any other manufacturing technique, AM is also susceptible to faults and defects. Ensuring the quality of the manufactured component is extremely important in the aerospace industry because of its conservative nature.

This thesis presents a system that can identify anomalies in the Wire Feed Directed Energy Deposition/Laser Beam (W-DED/LB) process using acoustic emission (AE) sensors and machine learning (ML). The experiment was done by melting a Waspaloy® wire during the W-DED/LB process onto an Inconel® 718 substrate.

Four acoustic emission sensors and a free-field microphone were used to monitor the experiment. Since nickel-based superalloys are prone to cracking during the AM process, the AE sensors are capable of detecting the elastic waves produced by these cracks within the ultrasonic frequency range. Vallen AE Suite, a companion software for data ac-quisition and analysis provided by the AE sensor manufacturer, and LabVIEW software were used to record data from the AE sensor and microphone, respectively. Two unsupervised clustering algorithms were used to analyse the AE data:

1) HDBSCAN clustering; and 2) Agglomerative hierarchical clustering.

Agglomerative clustering was effective in identifying the anomalies, whereas HDBSCAN did not perform well on the dataset.

The results were evaluated using the transient waveforms of the points identified as anomalies and through X-Ray and X-Ray Computed Tomography (CT) inspection. Anomalies were detected during both X-Ray and CT inspections. The proposed system, which uses AE sensors and machine learning, shows promise for monitoring the W-DED/LB process of Waspaloy®.