以Wi-Fi通道狀態資訊及深度學習技術於智慧建築與能源監測研究;Using Wi-Fi Channel State Information and Deep Learning Techniques for Smart Buildings and Energy Management

NCUIR > Engineering College > Graduate Institute of Civil Engineering > Electronic Thesis & Dissertation > Item 987654321/91654

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/91654

Title:	以Wi-Fi通道狀態資訊及深度學習技術於智慧建築與能源監測研究;Using Wi-Fi Channel State Information and Deep Learning Techniques for Smart Buildings and Energy Management
Authors:	張智雄;Chang, Chih-Hsiung
Contributors:	土木工程學系
Keywords:	用電監測;通道狀態資訊;深度學習;智慧建築;建築資訊模型;Monitoring electricity consumption;Channel state information;Deep learning;Smart building;Building information modeling
Date:	2022-12-30
Issue Date:	2024-09-19 14:08:45 (UTC+8)
Publisher:	國立中央大學
Abstract:	因全球能源日漸枯竭，能源的開源節流成為當前首要課題，而從眾多探討節能之研究可知，影響家庭能源管理系統(Home Energy Management System, HEMS)節能效果的因素，主要為再生能源發電量及使用者用電行為之不確定性。故本研究之目的為藉由智慧建築(Smart building)能源監測之技術改善上述問題。隨著物聯網(Internet of Things, IoT)的興起，無線通訊中的Wi-Fi技術逐漸被廣泛使用，因Wi-Fi調變技術的演進，而有了通道狀態資訊(Channel State Information, CSI)技術，再結合深度學習(Deep learning)即可達到室內定位、人體活動識別(Human Activity Recognition, HAR)、步態識別(gait recognition)…等應用。其中，HAR及步態識別雖無法透過接收訊號強度指標(Received Signal Strength Indication, RSSI)達成，然而在土木領域中，大多仍依賴RSSI達到基於位置的服務(Location-Based Service)應用，其可搭配建築資訊模型(Building Information Modeling, BIM)模擬無線存取點(Wireless Access Point, WAP)佈設位置。本研究以CSI技術為基礎，開發可快速佈署之智慧建築能源監測(Smart Building Energy Monitoring, SBEM)系統，以非接觸式有效監測太陽能板溫度監測(Monitoring Temperature of Solar Panel, MTSP)，並透過五個CSI應用指標，分別為人體存在檢測、室內定位、HAR、站立面向分析及步態識別，達到用電行為監測(Monitoring Electricity Consumption Behavior, MECB)，在應用之前將先探討環境變異對CSI應用之影響。本研究所提出之TPCNN+GRU (Triple-Parallel CNN + Gate Recurrent Unit)深度學習架構對HAR與步態識別達到良好的準確率，且CSI技術不具可視性，有隱私保護(Privacy-preserving)的特性，對於日後SBEM系統之推廣有利。;In the state of global energy depletion, increasing the source of energy and reducing expenditure has become the primary topic. At present, it is known that the factors affecting the energy-saving effect of the Home Energy Management System (HEMS) are mainly the uncertainty of the amount of renewable energy power generation and the user′s electricity consumption behavior. Therefore, the purpose of this research is to improve the above-mentioned problems through the technology of smart building energy monitoring. With the rise of the Internet of Things (IoT), Wi-Fi technology in wireless communication has gradually been widely used. Due to the evolution of Wi-Fi modulation technology, Channel State Information (CSI) technology has emerged. It combined with deep learning can achieve indoor positioning, Human Activity Recognition (HAR), gait recognition and other applications. Among them, HAR and gait recognition cannot be achieved by Received Signal Strength Indication (RSSI). However, in the field of civil engineering, most of them still rely on RSSI to realize the application of Location-Based Service, and Building Information Modeling (BIM) can be used to simulate the location of Wireless Access Point (WAP) layout. Based on CSI technology, this research develops a rapidly deployable Smart Building Energy Monitoring (SBEM) system, which can effectively Monitoring Temperature of Solar Panel (MTSP) without contact, and through five CSI indicators, namely human presence detection, indoor positioning, direction of standing, HAR and gait recognition, to realize Monitoring Electricity Consumption Behavior (MECB). The impact of CSI application on environmental changes must be considered before application. The TPCNN+GRU (Triple-Parallel CNN + Gate Recurrent Unit) deep learning architecture proposed in this research achieves good accuracy for HAR and gait recognition. Moreover, CSI technology is invisible data, so it has the characteristics of privacy-preserving. It is beneficial to the promotion of SBEM system in the future.
Appears in Collections:	[Graduate Institute of Civil Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	18	View/Open

社群 sharing

Loading...