Background: Friction brake is the most important safety device of mechanical systems. In
order to avoid accidents caused by braking failure, it is highly important to forecast the remaining useful
life of brakes accurately.
Objective: The purpose of this study is to provide an overview about life prediction methods of brake
friction pair based on physical models and statistical analysis.
Methods: In this paper, the widely used life prediction methods of brake friction pair based on physical
model and statistical analysis are summarized. To be specific, the fatigue life of brake disc/drum and
the wear life of brake pad are analyzed in depth based on their physical models. Meanwhile, three life
prediction methods based on statistical analysis which are linear regression, grey prediction and neural
network, are discussed.
Results: Life predictions based on the physical model often ignore random, mutation and nonlinear
factors in building failure model, while life predictions based on statistical analysis don’t need to explore
the detailed failure mechanism. Curve fitting, grey prediction and neural network are mainly used
in life prediction based on the statistical analysis.
Conclusion: Data mining technology such as neural network plays a more important role as a result of
its comprehensive consideration of braking conditions and braking frequency, and studies on real-time
monitoring system for life predictions have practical significance for forecasting the catastrophic failure