Main Content

Design Condition Indicators at the Command Line

Explore data at the command line to identify features that can indicate system status or predict future states

You can derive condition indicators at the command line from signal analysis or model fitting. If you have rotating machinery, you can extract specialized features that incorporate characteristics of your system, such as characteristic fault frequencies, or derive gear condition metrics with sensitivities to specific fault modes. You can derive more general features in the time domain, the frequency domain, or in the time-frequency domain. You can also derive condition indicators by fitting static or dynamic models to your data, and examining model parameters or model behavior to distinguish fault states or predict system degradation. Use command-line feature selection and ranking commands to evaluate the effectiveness of your features. For more information, see Condition Indicators for Monitoring, Fault Detection, and Prediction.

To extract features, you must often first filter or transform your data first into the form that the specific feature extraction operation requires.

Functions

expand all

Filtering

filter1-D digital filter
designfiltDesign digital filters
butterButterworth filter design

Interpolation

interp11-D data interpolation (table lookup)

Time-Domain

tsaTime-synchronous signal average
tsadifferenceDifference signal of a time-synchronous averaged signal
tsaregularRegular signal of a time-synchronous averaged signal
tsaresidualResidual signal of a time-synchronous averaged signal
ordertrackTrack and extract order magnitudes from vibration signal
rpmtrackTrack and extract RPM profile from vibration signal
envelopeSignal envelope

Frequency-Domain

pspectrumAnalyze signals in the frequency and time-frequency domains
envspectrumEnvelope spectrum for machinery diagnosis
orderspectrumAverage spectrum versus order for vibration signal
modalfrfFrequency-response functions for modal analysis
ssestEstimate state-space model using time-domain or frequency-domain data
arEstimate parameters when identifying AR model or ARI model for scalar time series

Time-Frequency

pkurtosisSpectral kurtosis from signal or spectrogram
spectrogramSpectrogram using short-time Fourier transform
emdEmpirical mode decomposition
pentropySpectral entropy of signal
kurtogramVisualize spectral kurtosis
hhtHilbert-Huang transform
bearingFaultBandsGenerate frequency bands around the characteristic fault frequencies of ball or roller bearings for spectral feature extraction (Since R2019b)
gearMeshFaultBandsConstruct frequency bands around the characteristic fault frequencies of meshing gears for spectral feature extraction (Since R2019b)
faultBandsGenerate fault frequency bands for spectral feature extraction (Since R2019b)
faultBandMetricsSpectral metrics for the specified fault frequency bands of the power spectral density (PSD) (Since R2019b)
gearConditionMetricsStandard metrics for gear condition monitoring (Since R2019a)

Time-Domain

meanAverage or mean value of array
movmeanMoving mean
medianMedian value of array
stdStandard deviation of timeseries data
rmsRoot-mean-square value
movmadMoving median absolute deviation
peak2peakMaximum-to-minimum difference
skewnessSkewness
kurtosisKurtosis
dtwDistance between signals using dynamic time warping
rainflowRainflow counts for fatigue analysis
approximateEntropyMeasure of regularity of nonlinear time series
correlationDimensionMeasure of chaotic signal complexity
lyapunovExponentCharacterize the rate of separation of infinitesimally close trajectories
phaseSpaceReconstructionConvert observed time series to state vectors

Frequency-Domain

powerbwPower bandwidth
modalfrfFrequency-response functions for modal analysis
bandpowerBand power
meanfreqMean frequency
medfreqMedian frequency
sfdrSpurious free dynamic range
sinadSignal to noise and distortion ratio
snrSignal-to-noise ratio
thdTotal harmonic distortion
obwOccupied bandwidth
findpeaksFind local maxima

Time-Frequency

pentropySpectral entropy of signal
pkurtosisSpectral kurtosis from signal or spectrogram
spectrogramSpectrogram using short-time Fourier transform
tfmomentJoint moment of the time-frequency distribution of a signal
tfsmomentConditional spectral moment of the time-frequency distribution of a signal
tftmomentConditional temporal moment of the time-frequency distribution of a signal
instfreqEstimate instantaneous frequency

Model Fitting

ssestEstimate state-space model using time-domain or frequency-domain data
nlarxEstimate parameters of nonlinear ARX model
arxEstimate parameters of ARX, ARIX, AR, or ARI model
armaxEstimate parameters of ARMAX, ARIMAX, ARMA, or ARIMA model using time-domain data
arEstimate parameters when identifying AR model or ARI model for scalar time series
pemPrediction error minimization for refining linear and nonlinear models
modalfitModal parameters from frequency-response functions
modalfrfFrequency-response functions for modal analysis
segmentSegment data and estimate models for each segment

Recursive Model Fitting

recursiveARCreate System object for online parameter estimation of AR model
recursiveARMACreate System object for online parameter estimation of ARMA model
recursiveARMAXCreate System object for online parameter estimation of ARMAX model
recursiveBJCreate System object for online parameter estimation of Box-Jenkins polynomial model
recursiveLSCreate System object for online parameter estimation using recursive least squares algorithm
recursiveOECreate System object for online parameter estimation of Output-Error polynomial model
recursiveARXCreate System object for online parameter estimation of ARX model

Recursive State Estimation

unscentedKalmanFilterCreate unscented Kalman filter object for online state estimation
extendedKalmanFilterCreate extended Kalman filter object for online state estimation
particleFilterParticle filter object for online state estimation

Model Dynamics

dampNatural frequency and damping ratio
polePoles of dynamic system
zeroZeros and gain of SISO dynamic system

Simulation

simSimulate response of identified model
residCompute and test residuals

Feature Selection

pcaPrincipal component analysis of raw data
pcaresResiduals from principal component analysis
sequentialfsSequential feature selection using custom criterion
fscncaFeature selection using neighborhood component analysis for classification
tsnet-Distributed Stochastic Neighbor Embedding

Classification Feature Ranking

anova1One-way analysis of variance
bhattacharyyaDistanceOne-dimensional Bhattacharyya distance between two independent data groups to measure class separability (Since R2020a)
kruskalwallisKruskal-Wallis test
perfcurveReceiver operating characteristic (ROC) curve or other performance curve for classifier output
rocmetricsReceiver operating characteristic (ROC) curve and performance metrics for binary and multiclass classifiers (Since R2022a)
ranksumWilcoxon rank sum test
relativeEntropyOne-dimensional Kullback-Leibler divergence of two independent data groups to measure class separability (Since R2020a)
ttest2Two-sample t-test
correlationWeightedScoreAdjust feature ranking scores using correlation factor (Since R2020a)

Topics

Condition Indicators Basics

  • Condition Indicators for Monitoring, Fault Detection, and Prediction
    A condition indicator is any feature of system data whose behavior changes in a predictable way as the system degrades.
  • Signal-Based Condition Indicators
    A signal-based condition indicator is a quantity derived from processing of signal data. The condition indicator captures some feature of the signal that changes as system performance degrades.
  • Model-Based Condition Indicators
    A model-based condition indicator is a quantity derived from fitting system data to a model and performing further processing using the model. The condition indicator captures some feature of the model that changes as system performance degrades.

Condition Indicators for Rotating Machinery