Geri Dön

Data processing of measurements collected in IoT systems with machine learning

Başlık çevirisi mevcut değil.

  1. Tez No: 798311
  2. Yazar: ÖVGÜ ÖZDEMİR
  3. Danışmanlar: DR. KáLMáN TORNAİ
  4. Tez Türü: Yüksek Lisans
  5. Konular: Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrol, Computer Engineering and Computer Science and Control
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2018
  8. Dil: İngilizce
  9. Üniversite: Pázmány Péter Catholic University
  10. Enstitü: Yurtdışı Enstitü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 47

Özet

Özet yok.

Özet (Çeviri)

In the Internet of Things systems, smart meters gather a huge amount of data about the elements of the system. Processing of the measurements enables to obtain comprehensive information about the users taking part in the system and improve the service intelligently. In Smart Grids, it is a significant task to process the data collected by smart meters for the analysis of energy usage. Every consumer in a smart grid tends to use the energy different ways. Hence, it is an important task to achieve classification of different types of consumers in Smart Grids at a high accuracy and efficiency. To manage the capacity, pricing and distribution planning efficiently, electric power should be provided to the consumers having different consumption patterns different ways. Related works are trying to solve the classification task using various machine learning methods. However, the thesis work focuses on solving the classification task with deep learning, using particularly convolutional neural networks, as well as considering the performance of the other related machine learning algorithms. Even though deep neural networks have come up with remarkable performance results in case of two-dimensional images for several image classification applications, the capabilities of the deep neural networks have not been comprehensively investigated in case of using time series for this type of classification task. Thereby, the thesis presents a deep learning approach to be able to solve the classification problem of power consumers in Smart Grid and demonstrates promising performance results as a result of different tests. The results are obtained using a publicly available database containing electricity consumption measurements collected from 16 types of consumers (buildings) in the United States. As the consumption measurements are time series, data has been transformed to twodimensional form. Implementation of convolutional neural networks and performance tests on the electricity consumption measurements have been carried out using Python and Keras deep learning environment. The purpose of the tests has been to observe the limit of the deep neural networks by evaluating different test scenarios regarding input length and seasonal impact on the data set and to obtain a detailed analysis for the solution of the classification of the measurements. For this purpose, implemented deep neural networks have been evaluated using different seasons to find how much deep neural networks are influenced by the seasonality. The model has also been evaluated using different input lengths to find the minimum limit. Besides, the performance has been tested when the number of the classes were increased based on the seasons. To demonstrate the performance of the model for another classification application, tests have been performed on the data set of measurements collected by smart sensors containing time series of human activities. Both results demonstrate that the convolutional neural networks are capable of solving the classification task on time series data at high 4 level of accuracy. The performance results on the electricity consumption data set have been compared with different classification algorithms as well, and the comparison indicated that the convolutional neural networks had the most remarkable performance results among the other machine learning algorithms.

Benzer Tezler

  1. Building sensor-based real-time predictive maintenance system by utilizing artificial intelligent techniques

    Yapay akıllı teknikleri kullanarak sensör tabanlı gerçek zaman tahminli bakım sistemi kurulması

    RAGHAD MOHAMMED KHORSHEED

    Doktora

    İngilizce

    İngilizce

    2021

    Endüstri ve Endüstri Mühendisliğiİstanbul Teknik Üniversitesi

    Endüstri Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ ÖMER FARUK BEYCA

  2. Nesnelerin interneti platformları için makine öğrenmesi tabanlı bir tahmin modülü

    A machine learning based prediction module for internet of things platforms

    HALİL GÜLAÇAR

    Yüksek Lisans

    Türkçe

    Türkçe

    2018

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ YUSUF YASLAN

  3. Machine learning techniques for estimation of human tiredness level from brain electrical activity

    Başlık çevirisi yok

    SAAD SHABAN

    Doktora

    İngilizce

    İngilizce

    2023

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolAltınbaş Üniversitesi

    Elektrik ve Bilgisayar Mühendisliği Ana Bilim Dalı

    PROF. DR. OSMAN NURİ UÇAN

  4. Akıllı ev cihazlarının haberleşmesinde hafif sıklet şifreleme algoritmalarının performans analizi

    In communication of smart home devices performance analysis of lightweight encryption algorithms

    ÖMER YEL

    Yüksek Lisans

    Türkçe

    Türkçe

    2024

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolSakarya Üniversitesi

    Bilgisayar Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ HÜSEYİN ESKİ

  5. Gerçek zamanlı kablosuz algılayıcı ağlarda saldırı tespit ve saldırı yanıt si̇stemlerinin tasarlanması

    Design of intrusion detection and response systems on real-time wireless sensor networks

    ÇAĞLAR OFLAZOGLU

    Doktora

    Türkçe

    Türkçe

    2023

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolHatay Mustafa Kemal Üniversitesi

    Enformatik Ana Bilim Dalı

    PROF. DR. BÜNYAMİN YILDIZ

    DOÇ. DR. İPEK ABASIKELEŞ TURGUT