کنفرانس مهندسی برق ایران

صفحه اصلی / سی و یکمین کنفرانس بین المللی مهندسی برق

Improving Wind Turbines Blades Damage detection by using YOLO BoF and BoS

نویسندگان :

Reza Mohammadi¹ Saeed Sharifian²

1- دانشگاه صنعتی امیرکبیر 2- دانشگاه صنعتی امیرکبیر

کلمات کلیدی :

Convolutional Neural Network (CNN)،Wind Turbines،Deep Learning،Drone Inspection،Damage Detection

چکیده :

—In this study, we introduce a novel wind turbines blade’s damage detection framework for reducing inspection time and improving accuracy in comparison with human. With recent developments and improvements in UAVs technology, aerial imagery can be used to achieve high resolution images from blades. The acquired images, can be analyzed by novel deep learning and image processing techniques like YOLO, to detect the failures and damages on surface of blades. But because of small training set and the structure of a damage in a image, we need to use advanced deep learning methods to achieve more accurate analysis than human eyes and also previous works. The proposed method is found to be effective based on experimental results in terms of suggesting damage locations and types on a surface, the system can achieve a level of precision better than a human-level inspection.

لیست مقالات

لیست مقالات بایگانی شده

Integration of Deep Learning Techniques in Stock Market Forecasting: xLSTM-CNN with RevIN and Adaptive Wavelet Denoising

Alireza Mohammadi - Ali Doustmohammadi - Masoud Shafiee

Reduction of Common-Mode Voltage in Cascaded H-Bridge Inverter Under Faulty Conditions

Ashkan Raki - Yousef Neyshabouri - Hossein Iman-Eini - Mahdi Aslanian

تجزیه و تحلیل امواج فیبریلاتور دهلیزی به منظور طبقه‌بندی AF با استفاده از موجک لیدر

سارا میهن دوست

بهبود تخمین موقعیت هواپیما به کمک تشخیص موقعیت‌های پرت در داده‌های ADS-B ترافیک هوایی

سید سجاد حسینی رستمی - میترا میرزارضایی - بابک نجار اعرابی

بهبود نمونه برداری از سیگنال روی گراف مبتنی بر نظریه دوایر گرشگورین

مهدیه صادقیان - حمید سعیدی سورک

Multiswarm Binary Butterfly Optimization Algorithm for Solving the Multidimensional Knapsack Problem

Shakiba Shahbandegan - Madjid Naderi

Control of vienna rectifier with Discontinuous space vector modulation based on circuit level decoupling

Ali Roshandel - Mohammad Roshandel - Ebrahim Afjei

An improved ECG segmentation method based on adaptive Hermite functions

Abazar Arabameri - Sajad Haghzad Klidbary

Noninvasive Blood Pressure Classification Based on Photoplethysmography Using Machine Learning Techniques

Hanieh Mohammadi - Bahram Tarvirdizadeh - Khalil Alipour - Mohammad Ghamari

Autoencoders for Input Reduction in Interval Type-2 Hyperbolic Fuzzy System Identification and Control: Experimental Results

Behnaz Mohammadi - Nazanin Ildarabadi - Mohammad-R Akbarzadeh-T

ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 40.4.2