Models for Predicting Remaining Useful Life
The remaining useful life (RUL) of a machine is the expected life or usage time remaining before the machine requires repair or replacement. Predicting remaining useful life from system data is a central goal of predictive-maintenance algorithms.
The term lifetime or usage time here refers to the life of the machine defined in terms of whatever quantity you use to measure system life. Units of lifetime can be quantities such as the distance travelled (miles), fuel consumed (gallons), repetition cycles performed, or time since the start of operation (days). Similarly time evolution can mean the evolution of a value with any such quantity.
Typically, you estimate the RUL of a system by developing a model that can perform the estimation based on the time evolution or statistical properties of condition indicator values, such as:
A model that fits the time evolution of a condition indicator and predicts how long it will be before the condition indicator crosses some threshold value indicative of a fault condition.
A model that compares the time evolution of a condition indicator to measured or simulated time series from systems that ran to failure. Such a model can compute the most likely time-to-failure of the current system.
Predictions from such models are statistical estimates with associated uncertainty. They provide a probability distribution of the RUL of the test machine. The model you use can be:
A dynamic model such as those you obtain using System Identification Toolbox™ commands. For more information about such models, see RUL Estimation Using Identified Models or State Estimators.
Specialized Predictive Maintenance Toolbox™ models designed for computing RUL from different types of measured system data. For more information about these models, see RUL Estimation Using RUL Estimator Models.
Developing a model for RUL prediction is the next step in the algorithm-design process after identifying promising condition indicators (see Condition Indicators for Monitoring, Fault Detection, and Prediction). Because the model you develop uses the time evolution of condition indicator values to predict RUL, this step is often iterative with the step of identifying condition indicators. For more information, see Feature Selection for Remaining Useful Life Prediction.