Predictive Maintenance Toolbox
Design and test condition monitoring and predictive maintenance algorithms
Predictive Maintenance Toolbox™ lets you label data, design condition indicators, and estimate the remaining useful life (RUL) of a machine.
The toolbox provides functions and an interactive app for exploring, extracting, and ranking features using data-based and model-based techniques, including statistical, spectral, and time-series analysis. You can monitor the health of rotating machines such as bearings and gearboxes by extracting features from vibration data using frequency and time-frequency methods. To estimate a machine's time to failure, you can use survival, similarity, and trend-based models to predict the RUL.
You can analyze and label sensor data imported from local files, cloud storage, and distributed file systems. You can also label simulated failure data generated from Simulink® models. The toolbox includes reference examples for motors, gearboxes, batteries, and other machines that can be reused for developing custom predictive maintenance and condition monitoring algorithms.
Fault Detection and Remaining Useful Life (RUL) Estimation
Detect anomalies, diagnose the root cause of faults, and estimate RUL using machine learning and time-series models.
RUL Estimation Models
Estimate the RUL of a machine to help you predict its time to failure and optimize maintenance schedules. The type of RUL estimation algorithm used depends on the condition indicators extracted from the data, as well as how much data is available.
Fault Diagnosis Using Classification Models
Isolate the root cause of a failure by training classification and clustering models using support vector machines, k-means clustering, and other machine learning techniques.
Fault and Anomaly Detection
Track changes in your system to determine the presence of anomalies and faults using changepoint detection, Kalman filters, and control charts.
Condition Indicator Design
Extract features from sensor data using signal-based and model-based approaches. Use extracted features as inputs to diagnostic and machine learning algorithms.
Diagnostic Feature Designer App
Extract, visualize, and rank features to design condition indicators for monitoring machine health.
Signal-Based Condition Indicators
Extract features from raw or preprocessed sensor data using rainflow counting, spectral peak detection, spectral kurtosis, and other time, frequency, and time-frequency domain techniques.
Model-Based Condition Indicators
Fit linear and nonlinear time-series models, state-space models, and transfer function models to sensor data. Use the properties and characteristics of these fitted models as condition indicators.
Reference Examples for Algorithm Development
Develop condition monitoring and predictive maintenance algorithms for batteries, gearboxes, pumps, and other machines.
Bearings and Gearboxes
Develop algorithms for classifying inner and outer race faults, detecting gear tooth faults, and estimating RUL.
Pumps, Motors, and Batteries
Develop algorithms for detecting leaks and clogs in pumps, tracking changes in motor friction, and estimating battery degradation over time.
Data Management and Labeling
Access data wherever it lives. Generate simulation data from Simulink models to represent machine failures in the absence of real sensor data.
Data Organization and Labeling
Import and label data from local files, Amazon S3™, Windows Azure® Blob Storage, and Hadoop® Distributed File System.
Failure Data Generation from Simulink and Simscape
Simulate failure data using Simulink and Simscape™ models of your machine. Modify parameter values, inject faults, and change model dynamics.
Diagnostic Feature Designer
Interactively extract, visualize, and rank features from measured or simulated data for machine diagnostics and prognostics
Gear Condition Metrics
Extract standard gear condition indicators from time-synchronous averaged signals