Application of damage detection techniques using wind turbine modal data

Abstract

As any structure ages, its structural characteristics will also change. The goal of this work was to determine if modal response data from a wind turbine could be used in the detection of damage. The input stimuli to the wind turbine were fiom traditional modal hammer input and natural wind excitation. The structural response data was acquired using accelerometers mounted on the rotor of a parked and undamaged horizontal-axis wind turbine. The bolts at the root of one of the three blades were then loosened to simulate a damaged blade. The structural response data of the rotor was again recorded. The undamaged and damage-simulated datasets were compared using existing damage detection algorithms. Also, a novel algorithm for combining the results of different damage detection algorithms was utilized in the assessment of the data. This paper summarizes the code development and discusses some preliminary damage detection results.

Structural health monitoring of offshore wind turbines: A review through the Statistical Pattern Recognition Paradigm

“…reduction in mass or stiffness) [72,73]. Due to these changes, the structure is considered to be damaged; that damage being identifiable by comparison between the structure's modal parameters before and after an event.…”

Vibration-based damage detection algorithm for WTT structures

“…Damage-induced changes in structural parameters, such as reduction in stiffness or loosening of connection, will cause detectable changes in the modal properties 4,[6][7]17 . As compared to other modal parameters, natural frequency is relatively convenient to utilize for identifying damage locations.…”

Damage detection and localization algorithm using a dense sensor network of thin film sensors

“…6 Traditional approaches for SHM of wind turbine blades have focused on the global monitoring of structures using a limited number of sensors and applying a variety of post-processing techniques. [7][8][9] These techniques often lack the ability to localize damage and distinguish local failures from global events. Additionally, the large amounts of data collected and retained for many post-processing techniques can add unacceptable data storage costs.…”