Multi-damage detection of composite blades via the curvature modal shape approach in combination with surface interpolation and extreme learning machine

Abstract

In recent years, there has been a rise in renewable energy, which has led to an increase in equipment maintenance. The damage detection technique for wind turbine blades (WTBs) can reduce maintenance costs. The majority are not immediately applicable to structures that are in an operating condition. Motivated by the current constraints, a novel two-stage multi-damage quantitative detection approach in WTBs during operation has been proposed. The multiple damage locations are first identified by calculating the curvature modal shape (CMS) difference for the WTBs after the interpolation of the operating deflection shape (ODS) using surface interpolation (SI). Then the relationship between the natural frequencies and damage severities for the WTBs is determined utilizing the finite element method (FEM). The established correlation provides a natural frequency database that can be utilized for identifying the corresponding damage severities through the application of the Extreme Learning Machine (ELM) method, upon measuring the natural frequencies of the damaged WTB. The efficacy of the suggested damage detection methodology has been confirmed through simulations conducted on WTBs featuring two or three instances of damage. Moreover, the ODS along with natural frequencies of the WTBs exhibiting two or three damages are experimentally assessed using a Scanning Laser Doppler Vibrometer. The simulation and experiment verify that the presented method has satisfactory effectiveness in damage detection.