Date of Award

9-12-2019

Publication Type

Doctoral Thesis

Degree Name

Ph.D.

Department

Mechanical, Automotive, and Materials Engineering

Keywords

Bearings, Blades, Failure Prognosis, Generators, Wind Turbines

Supervisor

Carriveau, R.

Supervisor

Ting, D.S-K.

Rights

info:eu-repo/semantics/embargoedAccess

Abstract

Wind energy is playing an increasingly significant role in the World's energy supply mix. In North America, many utility-scale wind turbines are approaching, or are beyond the half-way point of their originally anticipated lifespan. Accurate estimation of the times to failure of major turbine components can provide wind farm owners insight into how to optimize the life and value of their farm assets. This dissertation deals with fault detection and failure prognosis of critical wind turbine sub-assemblies, including generators, blades, and bearings based on data-driven approaches. The main aim of the data-driven methods is to utilize measurement data from the system and forecast the Remaining Useful Life (RUL) of faulty components accurately and efficiently. The main contributions of this dissertation are in the application of ALTA lifetime analysis to help illustrate a possible relationship between varying loads and generators reliability, a wavelet-based Probability Density Function (PDF) to effectively detecting incipient wind turbine blade failure, an adaptive Bayesian algorithm for modeling the uncertainty inherent in the bearings RUL prediction horizon, and a Hidden Markov Model (HMM) for characterizing the bearing damage progression based on varying operating states to mimic a real condition in which wind turbines operate and to recognize that the damage progression is a function of the stress applied to each component using data from historical failures across three different Canadian wind farms.

Download

Share

COinS