ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Optimisation framework for distinctive vertical axis wind turbine blade generation using hybrid multi-objective genetic algorithms and deep neural networks

Joseph, J and Pant, P and Omkar, SN (2019) Optimisation framework for distinctive vertical axis wind turbine blade generation using hybrid multi-objective genetic algorithms and deep neural networks. In: AIAA AVIATION 2020 FORUM, 15 June 2020 through 19 June 2020, Virtual, Online, pp. 1-31.

[img] PDF
AIA-AVI-FOR-1-PartF-1-31.pdf - Published Version
Restricted to Registered users only

Download (7MB) | Request a copy
Official URL: https://dx.doi.org/10.2514/6.2020-3119


There has been a considerable impetus to curb global climate change brought about by the deleterious environmental impact of fossil fuels. This has driven the use of renewable energy, especially wind energy, as it is amongst the most abundant forms of renewable sources available to mankind. However, harnessing the wind has been challenging, not only due to its unpredictable and intermittent nature but also due to variations in wind conditions from one location to another. These challenges result in significantly longer design times for developing wind turbines while still yielding sub-optimal designs due to the varying nature of the operating conditions. Furthermore, commonly used blade design routines are heavily dependent on empirical models, which do not account for the wake interactions between the different blades of the turbine. Therefore, to circumvent these problems, a radically new design methodology has been formulated, wherein the latest computational methods and machine learning algorithms have been used. This paper delves into this problem by establishing a computational design framework that can be used to develop blade profiles for vertical axis wind turbines (VAWT) with the aim of maximizing efficiency at the given operating conditions using hybrid optimisation methods. This framework realises genetic algorithms by invasive weed optimisation (IWO), and multi-objective implementation using non-dominated sorting (NSGA-II) while utilizing machine learning and artificial deep neural networks in function approximation to reduce the overall computational cost. The optimisation framework has been validated using an extensive array of known test functions and further by comparatively testing an optimised blade design for a set of operating conditions with commercially prevalent blades under similar design constraints. © 2019, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.

Item Type: Conference Paper
Publication: AIAA Aviation 2019 Forum
Publisher: American Institute of Aeronautics and Astronautics Inc, AIAA
Additional Information: cited By 0; Conference of AIAA Aviation 2019 Forum ; Conference Date: 17 June 2019 Through 21 June 2019; Conference Code:228939
Keywords: Approximation algorithms; Aviation; Climate change; Deep learning; Deep neural networks; Environmental impact; Fossil fuels; Genetic algorithms; Learning algorithms; Learning systems; Multiobjective optimization; Neural networks; Turbine components; Wakes; Wind power; Wind turbines, Computational design; Function approximation; Global climate changes; Hybrid multi-objective genetic algorithm; Non-dominated Sorting; Optimisation method; Use of renewable energies; Vertical axis wind turbines, Turbomachine blades
Department/Centre: Division of Mechanical Sciences > Aerospace Engineering(Formerly Aeronautical Engineering)
Date Deposited: 03 Feb 2021 10:34
Last Modified: 03 Feb 2021 10:34
URI: http://eprints.iisc.ac.in/id/eprint/67846

Actions (login required)

View Item View Item