Fault Prediction Of Wind Turbine Gearbox Based On Hidden Semi-Markov Models

Fault prediction is a hot issue in fault diagnosis of equipments. The gearbox, which is a key part of the wind turbine,has an important influence for the whole turbine in terms of its monitoring condition. Furthermore, it will take lots of time to repair the gearbox because of its high failure rate.As a result,many certain methods should be used in monitoring its condition and prediciting its fault trend,which are the basis of making rational strategies in maintenance and extending its useful life.This paper makes a research on the method of fault prediction in wind turbine based on the Hidden Semi-Markov Models that with a purpose of monitoring the gearbox condition using a whole life model theory and current data so that make predictions for its remain useful life,which is an essential reference and basis on the operation and maintenance of wind turbines.The main contents of the paper are as follows:(1) A brief introduction is made in development of wind turbine gearbox and its common structures,while making a detailed introduction in its type of the fault,rules of vibration and fault evolution with examples of gears,bearings and so forth.(2)Introducing the theory of the Markov, the Hidden Markov models and the Hidden Semi-Markov Models that refers to this paper.What is more,a Continuous HSMM is proposed to make predictions including the general idea and specific steps for wind turbine gearbox with improved algorithm on the basis of the traditional HSMM.(3)The method based on the HSMM is tested with real data in bearing load test and monitoring gear of wind turbine gearbox, which is used in estimating the remaining useful life with a whole life model theory.As a consequence,it manifests that experimental results are ideal so that the method is proved to be reasonable and effective.