This thesis explores the development of AI-driven anomaly detection (AD) methods for monitoring industrial equipment within Internet of Things (IoT) environments.
The collaboration with Statwolf S.r.l., a company specializing in data analytics and machine learning for industrial systems, focuses on designing scalable and reliable algorithms capable of identifying abnormal behaviors in connected machinery and production assets.

Modern industrial IoT infrastructures generate large volumes of sensor data, such as vibration, temperature, pressure, and operational metrics. Detecting anomalies in these streams is crucial for predictive maintenance, fault prevention, and process optimization.
The project investigates both unsupervised and semi-supervised anomaly detection approaches, aiming to detect deviations from normal operating conditions without the need for extensive labeled datasets.

Focus on eXplainable Artificial Intelligence approaches.

The final objective is to integrate the developed models into Statwolf’s analytics platform, providing real-time alerts, explainable results, and visual analytics for maintenance engineers.