Data Fusion Based on an Iterative Learning Algorithm for Fault Detection in Wind Turbine Pitch Control Systems.

Leonardo Acho,Gisela Pujol-Vázquez

doi:10.3390/s21248437

Leonardo Acho, Gisela Pujol-Vázquez

Open Access

https://doi.org/10.3390/s21248437

Copy DOI

Abstract

In this article, we propose a recent iterative learning algorithm for sensor data fusion to detect pitch actuator failures in wind turbines. The development of this proposed approach is based on iterative learning control and Lyapunov’s theories. Numerical experiments were carried out to support our main contribution. These experiments consist of using a well-known wind turbine hydraulic pitch actuator model with some common faults, such as high oil content in the air, hydraulic leaks, and pump wear.

Highlights

Data fusion is a mathematical discipline that deals with the acquisition, processing, and combination of synergies of information gathered from sensors [1]
The adaptive iterative learning control scheme is based on performing repetitive tasks to obtain a parameter estimation
Based on the simplest model used for the pitch actuator system (3), and because the iterative process identifies a parameter related to the system time-constant, the best option for the iterative process is to use a stepped pitch reference command, as shown in the previous simulations

Summary

Introduction

Data fusion is a mathematical discipline that deals with the acquisition, processing, and combination of synergies of information gathered from sensors [1]. Data fusion can be defined as the combination of data and information from different sources, to obtain improved information [2]. This data fusion is usually done to analyze and understand a phenomenon [3,4,5], for instance a system malfunction. There are different analysis methods that combine data from different sources, where the most common options are algorithms based on optimization [10], multiblock (or multitable) methods [11], and statistical data fusion [12]. We used an original statistical parametric identification to perform data fusion, where covariance of sensory information is not required, which is generally not available

Objectives

Results

Discussion

Conclusion