Native anomaly detection
Amazon IoT SiteWise native anomaly detection is a machine learning (ML) feature for monitoring industrial equipment that detects abnormal equipment behavior and identifies potential failures. With native anomaly detection, you can implement predictive maintenance programs and identify suboptimal equipment processes.
Amazon IoT SiteWise native anomlay detection doesn’t require extensive ML knowledge or experience. You simply select the properties to train a custom ML model that finds potential failures for you. Amazon IoT SiteWise native anomlay detection automatically creates the best model to learn your equipment’s normal operating conditions. The model is optimized to find abnormal equipment behavior that occurred in the historical data. Using either the Amazon IoT SiteWise console or the SDK, you run the model to process new time-series data according to your desired schedule.
To use Amazon IoT SiteWise native anomaly detection, you do the following:
Select the properties and the time period that you would like to train against.
Add the periods of historical failures shown in the data (label data), if it exists.
Train your ML model using Amazon IoT SiteWise native anomlay detection.
Setup your inference schedule to test your live data streams against your trained model.
Native anomaly detection monitors fixed and stationary industrial equipment that operates with limited variability in operating conditions. Supported equipment includes rotating machinery such as pumps, compressors, motors, computer numerical control (CNC) machines, and turbines. Process industry applications include heat exchangers, boilers, and inverters. Native anomaly detection is a back-end analytics service integrated into Amazon IoT SiteWise and supplements existing maintenance systems.