Maintenance Models for Real Time Optimization of Wind Farm Maintenance

Abstract

The thesis intends to study maintenance models for real time optimization of wind turbine maintenance. And then the result could be used to improve the availability of wind farm projects. The experiments of bearing degradation were run in RAMS lab for obtaining real degradation for models testing. It provides a good opportunity to understand the degradation mechanism and chance to practice maintenance models with real data. Some advices are given for improving the experiments in future. Features are extracted from the observations of experiment and selected by the value of monotonicity. Three methods for Cumulative Distribution Function of the first passage time are investigated and implemented. The first passage time is the first time that a stochastic process reaches a certain level. The selected stochastic process are wiener process and Geometric Brownian motion. The results are calculated through designed models with assumed parameters. All results are compared and discussed to give advice on future applications. The first passage time model is compared with digital twin model with the experiment data. The final results are consistent with each other. The benefit of digital twin is self correction during the predict process especially in the later stage. All these models could be used to improve maintenance strategy later.