Geri Dön

Machine learning modelling of stall on airfoils

Başlık çevirisi mevcut değil.

PDF İndir

  1. Tez No: 841465
  2. Yazar: BERAT KAAN FIRAT
  3. Danışmanlar: Belirtilmemiş.
  4. Tez Türü: Yüksek Lisans
  5. Konular: Makine Mühendisliği, Mechanical Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2022
  8. Dil: İngilizce
  9. Üniversite: Imperial College London
  10. Enstitü: Yurtdışı Enstitü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 64

Özet

Research and developments have been growing for wind power industry in the form of onshore and offshore wind turbines as they provide cost-effective, clean, and sustainable energy. The increase in the need of clean and sustainable energy source and competition in renewable energy has created a need of a more efficient design for wind turbines that is capable of withstanding high loads while avoiding stability problems of the blades like stall. CFD analysis require many simulations and are mostly time consuming when the design space of possible airfoil geometries is considered. On the other hand, the recent growth of data driven methods and applications has led to advances in many scientific fields. Therefore, machine learning applications can be used together with a data set from a highfidelity data from CFD to build a surrogate model. However, more conventional methods require large data set to make the machine learning model learn the pattern and usually struggles to foresee the results for the unseen data. This project aims to combine the methods on CFD simulations with machine learning applications to create a predictive model that determines the stall of an airfoil by using novel machine learning techniques. In this technique, physically lower-fidelity solutions are being injected into the machine learning model to make the most efficient output from the model predictions with a given amount of data. To provide the data set of different airfoil geometries covering corresponding stall angles, a CFD solver has been utilised. Artificial neural networks were trained, and physics guidance is provided by a vortex panel method. It has been shown that when the neural networks are constrained by physics input, data is used more efficiently, and predictions are more accurate to the physical situations in addition to the fact that the models identify the pattern using data, making the concept of finding the stall quicker which would otherwise take quite long times when solely CFD simulations were performed. Therefore, it is more reliable to depend on the predictions from physics guided neural networks which provided stall predictions up to 5% error whereas pure neural networks model provides 15% error when both models have the same structure.

Özet (Çeviri)

Özet çevirisi mevcut değil.

Benzer Tezler

  1. Tez No

    759550

    Solving Navier Stokes equations with physics informed neural network for calculation of aerodynamic forces

    Aerodinamik kuvvetleri hesaplamak için Navier-Stokes denklemlerinin fizik bilgili nöral ağ ile çözümü

    SILA AKPINAR

    Yüksek Lisans

    İngilizce

    İngilizce

    2022

    Mekatronik MühendisliğiSabancı Üniversitesi

    Mekatronik Mühendisliği Ana Bilim Dalı

    PROF. DR. SERHAT YEŞİLYURT

  2. Tez No

    920832

    Data driven optimization of structural and dynamical properties of parts manufactured via selective laser melting process using machine learning and finite element analysis methods

    Seçici lazer ergitme prosesi ile üretilen parçaların yapısal ve dinamik özelliklerinin makine öğrenmesi ve sonlu elemanlar analizi yöntemleri kullanılarak veri odaklı optimizasyonu

    CAN BARIŞ TOPRAK

    Doktora

    İngilizce

    İngilizce

    2024

    Makine MühendisliğiHacettepe Üniversitesi

    Makine Mühendisliği Ana Bilim Dalı

    PROF. DR. CAN ULAŞ DOĞRUER

  3. Tez No

    66774

    Yapay sinir ağları ile ulaştırma taleplerinin modellenmesi

    Başlık çevirisi yok

    YUSUF KAAN DEMİR

    Yüksek Lisans

    Türkçe

    Türkçe

    1997

    İnşaat Mühendisliğiİstanbul Teknik Üniversitesi

    Ulaştırma Ana Bilim Dalı

    PROF. DR. HALUK GERÇEK

  4. Tez No

    857941

    Accelerating molecular docking using machine learning methods

    Kenetleme hesaplarının makine öğrenme metotları ile hızlandırılması

    ABDULSALAM YAZID BANDE

    Yüksek Lisans

    İngilizce

    İngilizce

    2022

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Bilgisayar Bilimleri Ana Bilim Dalı

    Assist. Prof. Dr. SEFER BADAY

  5. Tez No

    768818

    Rastgele yönlendirilmiş karbon nanotüp takviyeli kompozitlerin modellenmesi

    Modelling of randomly oriented carbon nanotube reinforced composites

    EMRE KÖROĞLU

    Doktora

    Türkçe

    Türkçe

    2022

    Uçak Mühendisliğiİstanbul Teknik Üniversitesi

    Uçak ve Uzay Mühendisliği Ana Bilim Dalı

    PROF. DR. HALİT SÜLEYMAN TÜRKMEN

Geri Dön