Introducing the problem of anomaly detection in predictive maintenance
Before we move on to the actual task of anomaly detection, we will explain the concept of an anomaly, discuss the challenges of anomaly detection, introduce IoT data for the application, and perform data preprocessing and exploration.
We’ll complete this introduction with the upcoming subsections:
- Introducing the anomaly detection problem
- IoT data preprocessing
- Exploring anomalies visually
In the first subsection, we’ll explain what characterizes an anomaly and introduce the anomaly detection application.
Introducing the anomaly detection problem
Anomalies can be of two types, static and dynamic:
- Static anomaly: This is a data signal that is different from its neighbors; for example, a sudden rotor breakdown due to a power outage.
- Dynamic anomaly: This shows slowly changing patterns over time; for example, due to a deteriorating rotor, the mechanical pieces...
