لطفا منتظر بمانید ...
0% Complete
صفحه اصلی
/
یازدهمین کنفرانس بین المللی کنترل ، ابزار دقیق و اتوماسیون
Identification of Solar Panel Conditions Using Fully Decentralized Federated Learning
نویسندگان :
Keyvan Kazemi
1
1- دانشگاه تهران
کلمات کلیدی :
Decentralized federated learning،Solar panel monitoring
چکیده :
The efficiency of solar panels is highly sensitive to their physical and operational conditions, with environmental and structural factors leading to significant performance degradation. Automated monitoring with deep learning provides a promising solution, but centralized training requires aggregating data from distributed solar farms, which is often impractical due to privacy, bandwidth, and ownership constraints. Federated learning (FL) mitigates these challenges by enabling collaborative training without raw data sharing, yet server-based FL relies on a central coordinator, creating bottlenecks and a single point of failure. To address this, we propose a fully decentralized FL framework for solar panel condition monitoring, where clients fine-tune models locally using transfer learning and iteratively exchange updates with neighbors through weighted averaging. Experiments demonstrate that decentralized FL achieves a final test accuracy of 0.763, closely matching the accuracy of server-based FL at 0.775 and centralized training at 0.757.
لیست مقالات
لیست مقالات بایگانی شده
Modeling and exponential reaching law sliding mode control of the lower limb in cycling
Seyyed Arash Haghpanah - Navid Eqra - Seyyed Ehsan Zolfaghari - Mojtaba Mahzoon
Load Frequency Control Using Hierarchical Type-II Fuzzy Controller optimized via Imperialist Competitive Algorithm
Mohammad Ali Labbaf Khaniki - Mohammad Manthouri - Kosar Safari
Fault diagnosis of photovoltaic modules using deep neural networks-VGG16
Samaneh Azimi - Mohammad Manthouri
طراحی کنترلکننده حالت لغزشی تطبیقی زمان محدود مرتبه کسری برای بهبود پایداری توربینهای بادی پایه کششی شناور فراساحلی تحت اغتشاشات وزش باد و امواج دریا
متین جزءاسلامی - مائده نفیسیفر - محمدرضا عسکری سپستانکی - ابوالفضل جلیلوند
A Hybrid Inverse Reinforcement Learning and Deep Deterministic Policy Gradient Framework for Model-Free Control of Robotic Manipulators
Samim Kamyab - Ghoncheh Zand - Masoud Shirzadeh
Robust Fuzzy Observer-Based Speed Control of Electric Vehicle with Unmeasurable Premise Variables
Zeinab Echreshavi - Mokhtar Shasadeghi - Mohammad Hassan Asemani
Trajectory Planning and Simulation of Palletizing Robot Using MATLAB and Unity
Mohammad Sheibani - Amin Mojarrad Bahre - Armin Nasrollahi - Ali Rajabi
Tracking Blood Glucose Concentration of Type 1 Diabetic Patients Using Reinforcement Learning
Peyman Vafadoost Sabzevar - Hamid Sadrian - Ahmad Hajipour
Robust Data-Driven QFT Synthesis in the Presence of High Uncertainty
Mahdi Sadigh Dallali - Ali Khaki Sedigh
Design of a Nonlinear Backstepping Versus Sliding Mode Controller for a Human Musculoskeletal Arm Model in Sagittal Plane
Behzad Saeedi - Milad Alizadeh - Majid Mohammadi Moghaddam - Majid Sadedel
بیشتر
ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 44.9.5