Fault Diagnosis and Prognosis Capabilities for Wind Turbine Hydraulic Pitch Systems

TL;DR

This paper evaluates fault diagnosis and prognosis methods for hydraulic pitch systems in wind turbines, emphasizing the importance of early defect detection to improve reliability and reduce costly maintenance in offshore wind farms.

Contribution

It introduces a mathematical model and structural analysis for fault detection in hydraulic pitch systems, including sensor reduction strategies and robustness considerations.

Findings

Fault detection can be achieved with fewer sensors without losing diagnostic capability.

The model assesses the detectability of various faults and wear levels.

Robustness to model uncertainty enhances fault diagnosis reliability.

Abstract

Wind energy is the leading non-hydro renewable technology. Increasing reliability is a key factor in reducing the downtime of high-power wind turbines installed in remote off-shore places, where maintenance is costly and less reactive. Defects in the pitch system are responsible for up to 20% of a wind turbine downtime.Thus, monitoring such defects is essential for avoiding it. This paper presents a generic assessment of the diagnosis capabilities in hydraulic pitch systems, which are used in high-power wind turbines. A mathematical model of the non-linear system dynamics is presented along with a description of the most frequent faults that occur. Structural analysis is used to assess which defects can be detected in the pitch system. The structural properties are furthermore explored to investigate the possibility of reducing the amount of sensors without compromising the fault diagnosis capabilities. Robustness to model uncertainty is finally addressed and generic principles for estimating the detectable magnitude of wear and tear are presented.