Anomaly Detection

Detect signal anomalies using neural networks and autoencoders

Extract sparse time-frequency features based on wavelets to detect anomalies.

Functions

deepSignalAnomalyDetector Create signal anomaly detector

`cwtLayer`	Continuous wavelet transform (CWT) layer
`modwtLayer`	Maximal overlap discrete wavelet transform (MODWT) layer
`stftLayer`	Short-time Fourier transform layer

`dlcwt`	Deep learning continuous wavelet transform
`dlmodwt`	Deep learning maximal overlap discrete wavelet transform and multiresolution analysis
`dlstft`	Deep learning short-time Fourier transform
`cwtfilterbank`	Continuous wavelet transform filter bank
`findchangepts`	Find abrupt changes in signal
`findpeaks`	Find local maxima
`modwt`	Maximal overlap discrete wavelet transform
`risetime`	Rise time of positive-going bilevel waveform transitions
`stft`	Short-time Fourier transform
`signalFrequencyFeatureExtractor`	Streamline signal frequency feature extraction
`signalTimeFeatureExtractor`	Streamline signal time feature extraction
`waveletScattering`	Wavelet time scattering

`edfheader`	Create header structure for EDF or EDF+ file
`edfinfo`	Get information about EDF/EDF+ file
`edfread`	Read data from EDF/EDF+ file
`edfwrite`	Create or modify EDF or EDF+ file
`signalDatastore`	Datastore for collection of signals

Model wavelet scattering network in Simulink

Detect Air Compressor Sounds in Simulink Using Wavelet Scattering (DSP System Toolbox)
Use the Wavelet Scattering block and a pretrained deep learning network to classify audio signals.

Use wavelet-extracted features and an autoencoder to detect arc signals in a DC system.

Use wavelet and deep learning techniques to detect transverse pavement cracks and localize their position.

Learn how to develop an alert system for predictive maintenance using wavelet scattering and deep learning.

Classify faults in acoustic recordings of air compressors using a wavelet scattering network paired with a recurrent neural network.

Use a learning-based tool to detect abnormal points or segments in time-series data.

Use a long short-term memory autoencoder to detect anomalies in data from an industrial machine.