The Diagnostic Feature Designer app lets you extract features from your data interactively. Within the app, you can prepare your data for feature extraction, extract features and visualize their effectiveness, and rank features using various statistical algorithms. You can generate code from the app to replicate and automate the computations for the most effective features. For more information on the app, see Explore Ensemble Data and Compare Features Using Diagnostic Feature Designer.
Use Live Editor tasks to perform phase space reconstruction interactively and to extract the approximate entropy, correlation dimension, and Lyapunov exponent without writing code. The tasks generate plots that let you explore the effects of changing parameter values and options, and automatically generate code that becomes part of your live script.
|Diagnostic Feature Designer||Interactively extract, visualize, and rank features from measured or simulated data for machine diagnostics and prognostics|
|Reconstruct Phase Space||Reconstruct phase space of a uniformly sampled signal in the Live Editor|
|Estimate Approximate Entropy||Interactively estimate the approximate entropy of a uniformly sampled signal in the Live Editor|
|Estimate Correlation Dimension||Estimate the correlation dimension of a uniformly sampled signal in the Live Editor|
|Estimate Lyapunov Exponent||Interactively estimate the Lyapunov exponent of a uniformly sampled signal in the Live Editor|
A condition indicator is any feature of system data whose behavior changes in a predictable way as the system degrades.
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.
This three-part tutorial shows you how to work with ensemble data and extract and rank features in Diagnostic Feature Designer.
Interpret feature histograms to assess how well each feature separates labeled groups of data.
This example shows how to isolate a shaft fault from simulated measurement data for machines with varying rotation rates.
Use Live Editor tasks to reconstruct phase space of a uniformly sampled signal and then use the reconstructed phase space to estimate the correlation dimension and the Lyapunov exponent.