Icon

03_​Time_​Series_​AR_​Deployment

Anomaly Detection. Time Series AR Deployment

This workflow applies a previously trained auto-regressive model to predict signal values. The model was trained for normal functioning conditions. After prediction, the first and second level alarms are calculated based on the differences between real values and predicted values.

- Read 313 spectral amplitudes time series produced by the Pre-processing/Time Alignment & Visualization workflow - define training set with only normal data (till August 2007) and maintenance set till July 22 2008 - First Level Alarm For each frequency bin: - read previously trained auto-regressive models (Lag = 10) - calculate mean and stddev of model on training set - Generate prediction values on non-training data - Extract in maintenance window (Sep 2007 - Jul 2008) - generate first level alarms from prediction errors off training error boundaries - calculate moving average on first level alarms on all frequency bins - Second Level Alarm - generate second level alarms on moving average signals - on today's date, if level 2 alarm = 1 => email for mechanical check up Anomaly Detection: Time Series AR DeploymentThis workflow applies a previously trained auto-regressive model to predict signal values. The model was trained for normal functioning conditions. Afterprediction, the first and second level alarms are calculated based on the differences between real values and predicted values. Node 340Node 341 First Level Alert Second level Alert Take Action Read IOT Data - Read 313 spectral amplitudes time series produced by the Pre-processing/Time Alignment & Visualization workflow - define training set with only normal data (till August 2007) and maintenance set till July 22 2008 - First Level Alarm For each frequency bin: - read previously trained auto-regressive models (Lag = 10) - calculate mean and stddev of model on training set - Generate prediction values on non-training data - Extract in maintenance window (Sep 2007 - Jul 2008) - generate first level alarms from prediction errors off training error boundaries - calculate moving average on first level alarms on all frequency bins - Second Level Alarm - generate second level alarms on moving average signals - on today's date, if level 2 alarm = 1 => email for mechanical check up Anomaly Detection: Time Series AR DeploymentThis workflow applies a previously trained auto-regressive model to predict signal values. The model was trained for normal functioning conditions. Afterprediction, the first and second level alarms are calculated based on the differences between real values and predicted values. Node 340Node 341 First Level Alert Second level Alert Take Action Read IOT Data

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
On NodePit since: 2019-10-03
Last update: 2026-01-13
Created with KNIME version: v4.0.0

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!