Anomaly detection in small modular reactors using artificial neural networks
使用人工神经网络在小型模块化 反应器中进行异常检测
- Tez No: 841780
- Danışmanlar: DOÇ. DR. DONG ZHE
- Tez Türü: Yüksek Lisans
- Konular: Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrol, Bilim ve Teknoloji, Computer Engineering and Computer Science and Control, Science and Technology
- Anahtar Kelimeler: Belirtilmemiş.
- Yıl: 2023
- Dil: İngilizce
- Üniversite: Tsinghua University
- Enstitü: Yurtdışı Enstitü
- Ana Bilim Dalı: Belirtilmemiş.
- Bilim Dalı: Belirtilmemiş.
- Sayfa Sayısı: 99
Özet
新技术带来新的收益，但在使用时有时会带来风险。这些风险需要被控制或尽 可能降低到最低点。核能能够在不需要太多土地的情况下产生高价值的清洁热能。 控制和减轻这项技术的风险是一个不断发展的领域，因为这项技术的收益是如此 之大。对于这种情况，可以使用异常检测来降低这些风险。它旨在在许多方面降 低成本、时间和风险。及早发现问题可以防止随之而来的巨大风险和成本。 本研究的目的是使用人工神经网络 (ANN) 结构、多层感知器 (MLP) 作为异常 检测器来检测 NHR200-II 电站 SMR 设计中的异常。为此，本研究分为三个阶段。 在第一阶段，NHR200-II 工厂模拟用于为异常检测器收集数据。收集的数据被分为 9 类，包括正常操作和电源变化。在第二阶段，选择了 MLP 网络结构用于测试。最 后一部分，MLP 网络已经用从模拟中获得的数据集进行了训练和测试。为了达到 最好的效果，MLP 网络已经用 13 个训练函数和 5 个性能函数进行了测试。此外， 不同的神经元结构与最佳训练性能函数相结合，以实现 MLP 网络的最高成功率。 获得的结果显示了具有不同配置的 MLP 结构取得了多大的成功。它还可以表 明 ANN 在 SMR 中作为异常检测器的潜在用途。该拟议系统可能能够提高 SMR 的 安全性、意识和效率。 关键词：异常检测、小型模块化反应堆、人工神经网络、多层感知器、核电站
Özet (Çeviri)
New technologies brings new gains and sometimes risk when it has been used. These risks needs to be controlled or reduced into lowest point possible. Nuclear energy is able to produce clean thermal energy at high values without requiring much land. Controlling and mitigating the risks of this technology is a growing area, as the gains of this technology are so great. For this instance, anomaly detection could be used to reduce these risks. It aims to lower the costs, time and risk in many aspects. Early detection of a problem could prevent huge risks and costs coming along. Objective of this study, using an artificial neural network (ANN) structure, Multilayer perceptron (MLP) as an anomaly detector to detect anomalies in an SMR design, NHR200-II plant. To do so, this study has divided into three phases. In first phase, NHR200-II plant simulation used to gather data for the anomaly detector. Gathered data has been classified into 9 classes including normal operation, abnormal conditions and power change scenarios. In second phase, MLP network structure has been chosen to use in testing. And for the final part, MLP networks has been trained and tested with the dataset obtained from the simulation. To achieve the best result, MLP network has been tested with 13 training function and 5 performance function. Also different neuron structures tested with the best training-performance function couple to achieve the highest success rate with the MLP network. Results obtained shows how much success did MLP structures with different configurations has reached out. It can also show that in potential usage of ANN as an anomaly detector in SMRs. This proposed system may able to increase safety, awareness and efficiency in SMRs.
Benzer Tezler
- İleri seviye izleme ve makine öğrenmesi ile mikroservis mimarilerinde anomali tespiti
Anomaly detection in microscervices architectures with advanced monitoring and machine learning
BURAK ÇAYIR
Yüksek Lisans
Türkçe
2024
Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolGazi ÜniversitesiBilgisayar Mühendisliği Ana Bilim Dalı
PROF. DR. BÜNYAMİN CİYLAN
- Estimation of remaining useful life by using neural network method for lithium based batteries in aviation applications
Havacılıkta kullanılan lityum tabanlı bataryaların yapay sinir ağları ile ömür kestirmine katkılar
HÜSEYİN SELÇUK POLATÖZ
Yüksek Lisans
Türkçe
2019
Elektrik ve Elektronik Mühendisliğiİstanbul Teknik ÜniversitesiElektrik Mühendisliği Ana Bilim Dalı
DR. ÖĞR. ÜYESİ DERYA AHMET KOCABAŞ
- Makine öğrenmesi yöntemleri ile web isteklerinde anomali tespiti
Anomaly detection in web requests using machine learning methods
ÇAĞLAR ABABAY
Yüksek Lisans
Türkçe
2022
Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolHaliç ÜniversitesiBilgisayar Mühendisliği Ana Bilim Dalı
DR. ÖĞR. ÜYESİ FİGEN ÖZEN
- Olasılıksal alt-uzay yöntemleri ile hiperspektral imgelerde anomali saptama
Anomaly detection in hyperspectral images with probabilistic sub-space methods
NUR ALABUĞA
Yüksek Lisans
Türkçe
2022
Elektrik ve Elektronik MühendisliğiGebze Teknik ÜniversitesiElektronik Mühendisliği Ana Bilim Dalı
PROF. DR. KORAY KAYABOL
- Anomaly detection for video-based surveillance using covariance features and modeling of sequences via LSTMS
Video bazlı gözetim sistemlerinde kovaryans öznitelikleri kullanımı ve dizilerin LSTM modellenmesi ile anomali sezimi
ALİ ENVER BİLECEN
Yüksek Lisans
İngilizce
2021
Elektrik ve Elektronik MühendisliğiSabancı ÜniversitesiElektronik Mühendisliği Ana Bilim Dalı
DR. ÖĞR. ÜYESİ HÜSEYİN ÖZKAN