STRUCTURAL INTEGRITY EVALUATION OF WIND TURBINE BLADES USING PATTERN RECOGNITION ANALYSIS ON ACOUSTIC EMISSION DATA

摘要

Current wind turbine (W/T) blade certification practices require static and fatigue testing on new blades in order to assess whether the blade can sustain the applied loads. Within the scopes of a current EC-funded research project, acoustic emission (AE) monitoring has been extensively applied during testing of various W/T blades of similar design. All blades were loaded to failure by, either, gradually increasing the static test loads, or fatiguing the blade until it failed. It has already been reported that AE could locate the damage imposed on the blade during such tests (static and fatigue), and in most cases before the damage had become visible or audible, enhancing the assessment capabilities and the understanding of the failure process of the blades. Additionally, application of typical AE load-and-hold proof tests at intermediate loading stages, prior to failure, has enabled the assessment of the damage criticality for the particular proof load, denoted by high AE rates during load-holds. Furthermore, it has been observed that the AE behaviour of all tested blades during load-holds exhibited very similar trends right prior to failure, despite the fact that blades failed differently. The present paper reports on the use of (specially created for the Project) pattern recognition (PR) software, which has revealed the existence of a "critical" class of AE data appearing close to failure. This has enabled the formulation of evaluation criteria used for the-automated assessment of the blade's integrity, based on the amount of hits from critical classes appearing during the hold period. It is shown that, for similar blades, common grading criteria can be applied successfully, enabling a fast and effective "grading" (from "good" to "severely damaged"), and providing successful warnings of impending failure. This is particularly important for an effective analysis of fatigue tests that have lasted for months and have produced huge amounts of AE data. The software and the automated blade evaluation will be verified with future tests on large, commercial scale blades.