Autoencoder and Incremental Clustering-Enabled Anomaly Detection

Many machine-learning-enabled approaches towards anomaly detection depend on the availability of vast training data. Our data are formed from power readings cycles domestic appliances, such as dishwashers or washing machines, and contain no known examples anomalous behaviour. Moreover, we limited to machine’s voltage, amperage, current readings, drawn a retrofitted outlet in 60-s samples. No rich sensor previous insights available basis, limiting our ability leverage existing work. We design system monitor behaviour electrical appliances. This requires special consideration different same machine can exhibit behaviours, it accounts for this by clustering unseen cycle patterns into siloed datasets corresponding learned parameters. They then passed real-time an autoencoder ensemble reconstruction-based detection, using error reconstruction means flag points time. The correctly identifies trains appropriate clusters streams real-world dataset injected with stochastic, proportionate anomalies.