A method is proposed to reconstruct a cyclic time-inhomogeneous Markov pro- cess from measured data. First, a time-inhomogeneous Markov model is fit to the data, taken here from measurements on a wind turbine. From the time-dependent transition matrices, the time-dependent Kramers-Moyal coefficients of the corresponding stochastic process are computed. Further applications of this method are discussed.
@article{arxiv.1404.0203,
title = {Modeling and analysis of cyclic inhomogeneous Markov processes: a wind turbine case study},
author = {Teresa Scholz and Vitor V. Lopes and Pedro Lind and Frank Raischel},
journal= {arXiv preprint arXiv:1404.0203},
year = {2014}
}