A condition monitoring approach for real-time monitoring of degrading systems using Kalman filter and logistic regression

What is it about?

This research provides a new approach for diagnostics and prognostics of mechanical systems by providing a bridge between the well-known Kalman Filter and the Logistic Regression, when we consider observations for both the degradation of the system and its health at every time. The merging of these two methods can lead us to better predictions regarding the system's lifetime.

Featured Image

Why is it important?

The main contribution in this paper is a new mathematical framework that gives a closed-form solution to the expanded maximum likelihood equation that provides estimates for the unknown parameters of the state-space model represented by the Kalman Filter and those of the Logistic Regression. Throughout the literature there are many examples of methods that describe such hybrid mathematical approaches, but this is the first paper that provides a closed-form solution approach that can be solved analytically and not dealt with numerical approximations.