Remaining useful life prediction for milling tool based on an improved inverse Gaussian process

The present work proposes a novel remaining useful life (RUL) prediction method for milling tools by improving inverse Gaussian (IG) stochastic degradation process, to address the insufficient accuracy of existing RUL predicting methods under limited training sample data. In contrast to the standard IG process, which establishes the degradation increment relative to the initial state of degradation, the proposed improved IG process adopts an assumption that the degradation increment is more greatly affected by the current state of degradation rather than the initial state. This yields a three-parameter increment of wear, and the parameters are estimated by maximum likelihood estimation. Finally, the RUL of tool is predicted based on slice sampling technology. Applications of the proposed RUL prediction method to a benchmark milling tool wear dataset and a milling tool wear experiments demonstrate that the method obtains superior RUL prediction performance relative to two other state-of-the-art methods.