An Implementation of a Complete Methodology for Wind Energy Structures Health Monitoring

Abstract

In this chapter a complete methodology for a SHM damage detection solution is explained, and how this is validated in a laboratory tower model. Several methodologies are proposed for the typical process of SHM. Starting with sensor placement (the best possible sensor locations are found), selecting the more representative data, classifying the different environmental and operational conditions, applying a damage detection methodology, including sensor fault detection. The paradigm of damage detection can be tackled as a pattern recognition problem (comparison between the data collected from the structure without damages and the current structure in order to determine if there are any changes). There are lots of techniques that can handle the problem. In this work, accelerometer data is used to develop statistical data driven approaches for the detection of damages in structures. As the methodology is designed for wind turbines, only the output data is used to detect damage; the excitation of the wind turbine is provided by the wind itself or by the sea waves, being those unknown and unpredictable.

Siamese Neural Networks for Damage Detection and Diagnosis of Jacket-Type Offshore Wind Turbine Platforms

“…The results show that the damage detection and damage diagnosis results are promising, as they achieved great performance on different structural state samples. Finally, to thoroughly test the functional characteristics of the algorithm, a comparison is made with four other methodologies given in [31], [32], [13], and [12] that use the same laboratory structure. The first methodology, given in [31], is based on principal component analysis and support vector machines.…”

“…The first methodology, given in [31], is based on principal component analysis and support vector machines. The second methodology, given in [32] (page 67), is based on the well-known damage indicators: covariance matrix estimate and scalar covariance. The third methodology, given in [13], is based on machine learning methods and the fractal dimension feature.…”

Structural Damage Classification in a Jacket-Type Wind-Turbine Foundation Using Principal Component Analysis and Extreme Gradient Boosting

“…The structure of a small-scale wind-turbine foundation is used to validate the damage detection and localization methodology developed in this study. This benchmark structure, based on the one designed by Zugasti [ 35 ], was considered by Hoxha et al [ 36 ], Vidal et al [ 7 ], and Puruncajas et al [ 13 ] to validate different approaches for structural damage detection and classification experimentally. The benchmark structure—placed in the CoDAlab laboratory (Escola d’Enginyeria de Barcelona Est, Universitat Politècnica de Catalunya, Barcelona, Spain)—is shown in Figure 3 a; it is composed of three parts: (i) the nacelle on the top, (ii) tower in the middle, and (iii) jacket on the bottom.…”

“…A 5 mm crack was introduced at four different bars located in the jacket structure, one at a time. It can be said that the 5 mm crack damage is small for this structure [ 35 ] and therefore hard to detect. Summarizing, four different damage scenarios are considered in this study, as shown in Figure 3 b.…”

“…This signal produces a vibration response that is captured using eight triaxial accelerometers (PCB Piezotronics, Model 356A17) with a sensitivity: 500 mV/g (51 mV/(m/s )). The location of the eight sensors was determined using the Sensor Elimination by Modal Assurance Criterion (SEAMAC) method [ 35 ], and their final location is shown in Figure 3 a. Taking as reference Figure 3 b as the front view of the structure, below in Table 1 the location of each of the 8 sensors is detailed.…”