Acoustische anomaliedetectie voor industriële machines: robuustheid, generaliseerbaarheid en betrouwbaarheidseigenschappen

Acoustic Anomalous Detection (AAD) is the task of identifying whether the sound emitted from a target machine is normal or abnormal. Automatic detection of mechanical failure is an essential technology in the fourth industrial revolution, which involves artificial intelligence (AI)-based factory automation. Prompt detection of machine anomalies by observing acoustic signals is useful and necessary for monitoring the condition of machines. This PhD will address the following challenges: 1. In real-world factories, anomalies rarely occur and are highly diverse, which makes it difficult to obtain exhaustive patterns of anomalous sounds, and unknown anomalous sounds that were not observed in the given training data must be detected (i.e. unsupervised AAD). 2. In real-world cases, the operational conditions of machines or environmental noise often differ between the training and testing phases. For example, the operation speed of a conveyor can change due to seasonal demand, or environmental noise can fluctuate depending on the state of surrounding machines. 3. In test data, samples unaffected by domain shifts (source domain data) and those affected by domain shifts (target domain data) are mixed, and the source/target domain of each sample is not specified. Therefore, the model must detect anomalies with the same threshold value regardless of the domain (i.e., domain generalization). 4. Collecting data and detecting abnormal acoustic events in real-time using low-power IoT (IoT) systems in real-world factory applications requires techniques that can, on the one hand, minimize the amount of data to be transmitted and, on the other hand preserve essential data features.