Icon

Fourier Transform for Anamoly Detection

In this workflow I demonstrate how the Fourier Transform along with basic aggregations and rule settings can be used to automatically detect anomalies on when armed with only "normal" data.

Preprocess data so series are of the proper length.The Fourier Transform requires series of length 2^n. We use 512. Use the FFT component to compute theFourier Transform. Calculate mean and standard deviation by frequencyDetermine how many standard deviations away example test data isAverage the distance across all frequenciesIf over 2 then flag anomally.Note that in this example it is 14! one examplecolumncreate multiplewindows of length 512# of rows amultiple of 512find meand andvariance for eachfrequencyone examplecolumn# of rows amultiple of 512Join withteststd frommeanaverage stdfrom meanmore than 2 stdflag anomalyNormal FunctionBad Bearing Fast FourierTransform (FFT) Column Filter Chunk Loop Start Column Filter Missing Value Loop End Row Filter GroupBy Column Filter Row Filter Fast FourierTransform (FFT) Missing Value Column Filter Joiner Math Formula GroupBy Rule Engine Table Reader Table Reader Preprocess data so series are of the proper length.The Fourier Transform requires series of length 2^n. We use 512. Use the FFT component to compute theFourier Transform. Calculate mean and standard deviation by frequencyDetermine how many standard deviations away example test data isAverage the distance across all frequenciesIf over 2 then flag anomally.Note that in this example it is 14! one examplecolumncreate multiplewindows of length 512# of rows amultiple of 512find meand andvariance for eachfrequencyone examplecolumn# of rows amultiple of 512Join withteststd frommeanaverage stdfrom meanmore than 2 stdflag anomalyNormal FunctionBad BearingFast FourierTransform (FFT) Column Filter Chunk Loop Start Column Filter Missing Value Loop End Row Filter GroupBy Column Filter Row Filter Fast FourierTransform (FFT) Missing Value Column Filter Joiner Math Formula GroupBy Rule Engine Table Reader Table Reader

Nodes

Extensions

Links

Download

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

Download Workflow
Created by: corey
Created at: 2021-12-14
On NodePit since: 2021-12-18
Last update: 2024-05-01
Created with KNIME version: v4.4.2
Tags: IoTTime SeriesFFTFourier TransformDFTMachine MaintenancePredictive MaintenanceAnomaly Detection

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

Try NodePit Runner!