<p style="text-align: justify">Wind energy has significantly expanded in the last 0 years becoming a mainstream source of clean, cost-competitive energy around the world. In 019 69.4 GW of wind energy capacity was installed globally which corresponds to a 19 per cent increase from installations in 018 and the second-best year for wind historically. The total capacity for wind energy globally is now over 651 GW. However, the maintenance of Wind Turbines (WTs) can be really challengeable. On the one hand, most of WTs are operated in remote areas, such as hills, fields and offshore in the sea. On the other hand, the dimension of their main parts (e.g. tip, hub and rotor) is usually enormous. In order to prevent an abrupt machine breakdown and abate the economic loss, a number of Condition Monitoring (CM) methods have been proposed based on vibrations, acoustics, thermography and lubrication analysis. Vibration analysis is nowadays the most commonly used method for gearbox monitoring. Envelope analysis and the corresponding Squared Envelope Spectrum (SES) is a classic technique to diagnose bearings’ defects. However, the bearings’ signals are often weak and buried under the strong influence of gears. Therefore in order to find the optimal frequency band for demodulation in the frames of Envelope Analysis, the fast Kurtogram has been proposed as a popular tool, calculating the kurtosis of signals decomposed by a 1/3-binary tree. In addition, Infogram and Sparsogram have peen presented as similar diagnostics methods estimating the Negentropy and the Sparsity, respectively. The analysis of Wind Turbines usually requires significant effort due to the complexity of the drivetrain system and the varying operating condition and therefore the identification of the optimum diagnostic methodology is still under research. Scattering Transform (ST) is inspired by the Convolution Neural Network (CNN) and has achieved a great success at image classification. ST decomposes the signals into a number of sub-signals by a cascade of wavelet filters with different bandwidths. As such, thanks to the scattering networks and the localization property of wavelets, ST is capable to identify the hidden information and magnify the bearing’s fault information. In this paper, the scattering transform is proposed for CM of wind turbines. Various indicators are estimated based on the outputs of the ST. The methodology is tested and evaluated on a publically available wind turbine dataset of six wind turbines and the performance and the effectiveness of the proposed diagnostic indicators are compared with state of the art methodologies, being capable to accurately detect the faults.</p>

Vibration-Based Condition Monitoring of Wind Turbines Based on the Scattering Transform

Paper Proceedings: https://asmedigitalcollection.asme.org/GT/GT2021/volume/84898

Turbo Expo 2021

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.