透過您的圖書館登入
IP:3.141.197.247
  • 學位論文

機器學習技術應用於瓦斯存量監控系統

Machine Learning Technology Applied to Gas Storage Monitoring Systems

指導教授 : 鄭智元
共同指導教授 : 徐培倫(Pei-Lun Hsu)
本文將於2029/12/31開放下載。若您希望在開放下載時收到通知,可將文章加入收藏

摘要


桶裝瓦斯在使用上會遇到瓦斯存量不足導致相關設備(瓦斯爐、熱水器等)無法使用。而桶裝瓦斯壓力過高容易引起問題,若不即時處理會發生氣爆及火災之意外。為解決上述之問題本論文設計及開發一套監視系統,用來即時獲得瓦斯存量資訊。 本論文應用圖形辨識及機器學習技術,做為瓦斯存量監視系統的主要應用核心。硬體部分利用Esp32-CAM開發板拍攝研究時需要的所有影像及監視當下的即時影像,而軟體部分使用影像處理與機器學習為核心,建構成瓦斯存量監視系統。桶裝瓦斯內壓力過高未能即時發現容易造成瓦斯外洩引發意外,桶裝瓦斯內瓦斯存量不足會讓正在使用相關設備無法繼續使用,為解決上述問題透過與LINE聊天機器人交談之方式,將桶裝瓦斯資訊自動傳給使用者,讓使用者得知目前桶裝瓦斯的狀況。根據實驗結果,桶裝瓦斯在正常運作之範圍內,使用者能透過LINE聊天機器人上輸入關鍵字來查詢目前桶裝瓦斯的資訊。當瓦斯壓力過高或過低時自動的透過LINE聊天機器人來告知使用者桶裝瓦斯的安全性警示或更換桶裝瓦斯提醒。

並列摘要


When using barreled gas, there will be insufficient gas inventory, which will make the related equipment (gas furnace, water heater, etc.) unusable. The high pressure of the barreled gas is likely to cause problems. If it is not handled immediately, there will be an accident of gas explosion and fire. In order to solve the above problems, this paper designs and develops a monitoring system for instant access to gas inventory information. This paper applies image recognition and machine learning technology as the main application core of the gas inventory monitoring system. The hardware part uses the Esp32-CAM development board to capture all the images needed for the study and monitor the current live image. The software part uses image processing and machine learning as the core to construct a gas inventory monitoring system. If there is no immediate discovery that the pressure inside the gas in the barrel is too high, it will cause accidents caused by gas leakage. The insufficient inventory of gas-filled gas can make the relevant equipment unable to continue to use. Therefore, this study automatically communicates the barreled gas information to the user by talking with the LINE chat robot, so that the user can know the current state of the barreled gas. According to the experimental results, the barreled gas is within the normal operating range, and the user can input the keyword through the LINE chat robot to query the current information of the barreled gas. When the gas pressure is too high or too low, the LINE chat robot will automatically alert the user to the status of the barreled gas or to remind the replacement of the barrel of gas.

並列關鍵字

Image Recognition Machine Learning LINE BOT

參考文獻


[1] 許文豪,圖形辨識概述與實作,國立清華大學資訊工程系,民國八十九年七月,檢索日期:2019/06/01。
[2] Hiroki Tanioka, Kang Xin,深度學習入門教室:6堂基礎課程+Python實作練習,Deep Learning、人工智慧、機器學習的理論和應用全圖解,臉譜出版,民國一百零八年一月二十七日,檢索日期:2019/05/28
[3] 李其軒,運動精彩瞬間影像辨識與擷取之研究,國立臺灣海洋大學電機工程系,民國一百零七年六月二十一日,檢索日期:2019/05/28
[4] 賴郁仁,指紋辨識改善之研究,東吳大學資訊科學系,民國九十七年六月二十三日,檢索日期:2019/05/28。
[5] 蘇立珩,基於機器學習與影像之機械手臂抓取,國立清華大學動力機械工程系,民國一百零七年七月十八日,檢索日期2019/05/29。

延伸閱讀