Icon

Predictive_​Maintenance_​Model_​Deployment

<p>Predictive Maintenance</p><p>Anomaly Detection. Time Series AR Deployment<br><br>This workflow deploys a previously trained auto-regressive model for anomaly detection:<br>- Select the date for deployment. Two months of its past values must be available.<br>- Loop over each frequency column<br>- Apply the previously trained auto-regressive model to the data<br>- Calculate 1st level alarms based on the prediction errors<br>- Calculate 2nd level alarms as the moving average of the 1st level alarms<br>- Trigger an action if a 2nd level alarm is active</p>

Predictive Maintenance - Deployment

This workflow deploys a previously trained auto-regressive model for anomaly detection.

The input data must include at least 2 months of past values.

This workflow calls a separate "Send_email_to _start_checkup" workflow in case of a 2nd level alarm.

Model deployment and calculating 1st level alarms
Calculating 2nd level alarms and triggering action
[500-600]+Amp -> Amp
Column Name Replacer
moving average on level 1 alarms and alarm level 2 generation
Alarm Level 2
visualize 2nd alarm spikes
Bar Chart
Call Workflow node
Trigger check up if level 2 Alarm =1
add original datetime column
Joiner
moving average on level 1 alarms and alarm level 2 generation
Alarm Level 2
Lag = 10
Lag Column
loop on all columns i.e. on all frequency bins for each sensor
Column List Loop Start
generation alarm level 1 for all time series
Alarm level 1
visualize 1st alarm spikes
Bar Chart
Stats and Models
Regression Predictor
Missing Value
Select a day > Extract 2 months of its past
Select Day
read errorstats
Table Reader
Loop End (Column Append)
read AlignedData.csv
CSV Reader

Nodes

Extensions

Links