Preprocess Data

Clean and transform data to prepare it for extracting condition indicators at the command line and in the app

In algorithm design for predictive maintenance, data preprocessing is often necessary to clean the data and convert it into a form from which you can extract condition indicators. You can perform data preprocessing on arrays or tables of measured or simulated data that you manage with Predictive Maintenance Toolbox™ ensemble datastores. For an overview of some common types of data preprocessing, see Data Preprocessing for Condition Monitoring and Predictive Maintenance.

The Diagnostic Feature Designer app lets you perform many preprocessing operations interactively. The processing tools in the app include filtering, time-domain processing, frequency-domain processing, and interpolation. App time-domain processing options include specialized filtering for rotating machinery. For more information on the app, see Explore Ensemble Data and Compare Features Using Diagnostic Feature Designer.

Apps

Diagnostic Feature Designer

Interactively extract, visualize, and rank features from measured or simulated data for machine diagnostics and prognostics

Functions

expand all

Basic Preprocessing

`fillmissing`	Fill missing entries
`filloutliers`	Detect and replace outliers in data
`smoothdata`	Smooth noisy data
`movmean`	Moving mean
`detrend`	Remove polynomial trend
`rescale`	Scale range of array elements

Filtering

`filter`	1-D digital filter
`designfilt`	Design digital filters

Time-Domain Preprocessing

`tsa`	Time-synchronous signal average
`tsadifference`	Difference signal of a time-synchronous averaged signal
`tsaregular`	Regular signal of a time-synchronous averaged signal
`tsaresidual`	Residual signal of a time-synchronous averaged signal
`ordertrack`	Track and extract order magnitudes from vibration signal
`rpmtrack`	Track and extract RPM profile from vibration signal

Frequency-Domain Preprocessing

`pspectrum`	Analyze signals in the frequency and time-frequency domains
`envspectrum`	Envelope spectrum for machinery diagnosis
`orderspectrum`	Average spectrum versus order for vibration signal
`modalfrf`	Frequency-response functions for modal analysis
`bearingFaultBands`	Generate frequency bands around the characteristic fault frequencies of ball or roller bearings for spectral feature extraction
`gearMeshFaultBands`	Construct frequency bands around the characteristic fault frequencies of meshing gears for spectral feature extraction
`faultBands`	Generate fault frequency bands for spectral feature extraction

Time-Frequency Preprocessing

`pentropy`	Spectral entropy of signal
`pkurtosis`	Spectral kurtosis from signal or spectrogram
`kurtogram`	Visualize spectral kurtosis
`spectrogram`	Spectrogram using short-time Fourier transform
`hht`	Hilbert-Huang transform
`emd`	Empirical mode decomposition

Topics

Data Preprocessing for Condition Monitoring and Predictive Maintenance
Use signal-processing techniques to preprocess data, cleaning it and converting it into a form from which you can extract condition indicators. Knowledge of your system can help you choose an appropriate preprocessing approach.
Explore Ensemble Data and Compare Features Using Diagnostic Feature Designer
Follow this workflow for interactively exploring and processing ensemble data, designing and ranking features from that data, and exporting data and selected features, and generating MATLAB^® code.
Organize System Data for Diagnostic Feature Designer
Organize measurements and information for multiple systems into data sets that you can import into the app.
Process Data and Explore Features in Diagnostic Feature Designer
Filter and transform data within the app. Extract features from the imported and derived signals, and assess feature effectiveness.

Featured Examples

Multi-Class Fault Detection Using Simulated Data

Use a Simulink model to generate faulty and healthy data, and use the data to develop a multi-class classifier to detect different combinations of faults.

Open Live Script

Analyze and Select Features for Pump Diagnostics

Use the Diagnostic Feature Designer app to analyze and select features to diagnose faults in a triplex reciprocating pump.

Open Live Script

Wind Turbine High-Speed Bearing Prognosis

Build an exponential degradation model to predict the Remaining Useful Life (RUL) of a wind turbine bearing in real time. The exponential degradation model predicts the RUL based on its parameter priors and the latest measurements.

Open Live Script