請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/82972| 標題: | RFID結合影像辨識的智慧監控方法用於事業廢棄物清理之研究 A Study of intelligent monitoring method for industrial waste disposal using Radio Frequency Identification (RFID) and image recognition |
| 作者: | 陳柏崴 Bo-Wei Chen |
| 指導教授: | 闕蓓德 Pei-Te Chiueh |
| 關鍵字: | 智慧監控; 事業廢棄物清理; 無線射頻辨識; 影像辨識; 雜湊演算法 Intelligent monitoring; Industrial waste disposal; Radio frequency identification; Image recognition; Hash algorithm |
| 出版年 : | 2021 |
| 學位: | 碩士 |
| 摘要: | 隨著經濟發展與人口成長,廢棄物的產生量逐年上升,廢棄物被非法棄置的現象更是不斷發生,因此廢棄物的清除與處理作業必須要有一套更創新且有效的監控管理方法,以避免廢棄物被隨意處置或非法棄置。 過去大多數對於監控廢棄物清理作業的研究,主要都是監控廢棄物的清運車輛,而廢棄物被送往處理廠後,是否有被妥善處理,卻是沒有被深入探討的。廢棄物若是被清理機構非法處置,事業機構也會需要負連帶責任,因此事業機構過去都會派遣相關人員,利用跟車、訪廠等方式來確定委託的廢棄物是否被妥善清理。然而人力查核的方式不僅耗時,且需要花費相當高的成本才能監控每筆聯單的廢棄物。 本研究的目的為建立一種追蹤廢棄物清理流程的方法,結合無線射頻辨識(Radio Frequency Identification, RFID)技術、即時影像監控設備,以及自動影像辨識技術,去追蹤廢棄物從產源產生後,被清除機構載到處理機構處理的整個清理過程。根據研究的目標,選擇一事業機構、一清除機構,以及一處理機構來作為個案分析並進行實際訪廠,盤查廢棄物清除與處理的作業流程,並利用個案的清理流程規劃相對應的RFID與影像辨識查核方法。 過去利用RFID技術追蹤廢棄物流向的研究都是將RIFD標籤黏貼或綑綁於廢棄物的包裝上,然而環境的干擾、或是作業過程的擠壓,都有可能導致RFID標籤的讀取率下降。因此本研究規劃將RFID標籤以吊牌的方式,讓作業人員隨身攜帶,於作業當下即時記錄並追蹤每一次廢棄物清理的作業情況與作業的時間區間,將每一個作業步驟都進行事件點管理,以進行影像監控,並進一步利用圖像辨識的雜湊演算法來判斷較容易被非法棄置的廢棄物載運過程,以及需要花費大量時間查核的廢棄物暫存過程,是否有異常狀況發生。最後透過實際場域測試、成本效益分析,來確認此智慧監控廢棄物清理方法的可行性。 本研究所建立的智慧監控廢棄物清除與處理流程的方法,不只能夠即時監控廢棄物的清運端,也能即時監控廢棄物的處理端。結果也顯示利用雜湊演算法所建立的影像辨識模型,對於廢棄物清理流程中的異常狀況辨識表現良好,比傳統利用人力查核的方式節省更多的查核時間與成本。由於此次與個案只有短期的合作,若未來能夠長期導入不同的實際場域,蒐集更多數據,可望能讓影像辨識模型更優化,對異常狀況的預測更佳準確。未來可以再嘗試結合不同的感測器或影像辨識技術,對異常狀況做進一步的分析,以提升監控的效率與品質,然而還是要考慮到成本與效益,以及技術的方便性和可行性,以比較不同查核設備與方法的優缺點。 With economic development and population growth, the amount of waste generated is increasing year by year, and the phenomenon of illegal disposal of waste continues to occur. Therefore, the removal and treatment of waste must have more innovative and effective monitoring and management method. Avoid illegal disposal of waste. In the past, most of the research on monitoring waste disposal operations mainly focused on monitoring waste clearance machinery. However, after the waste is sent to the treatment plant, whether the waste is properly treated has not been discussed in depth. If the waste is illegally disposed by the disposal organization, the business organization will also be jointly and severally liable. Therefore, the business organization used to send relevant personnel to use car-following and factory visits to determine whether their waste has been properly cleaned up. However, the way of manpower checking is not only time-consuming but also expensive. The purpose of this research is to establish a method for tracking the waste cleaning process, combining radio frequency identification (RFID) technology, real-time video surveillance equipment, and automatic image recognition technology to track the entire waste disposal process. According to the research objectives, select a business organization, a clearance organization, a disposal organization as case analysis and actual visits to the factory, check the operation process of waste clearance and treatment, and use the case disposal process to plan corresponding checking method. In the past, the research on using RFID technology to track the flow of waste was to stick or bind RIFD tags to the waste packaging, but it was disturbed by the environment or squeezed by the packaging may cause the reading rate of RFID tags to decrease. Therefore, this research plans to use RFID tags as tags for operators to carry with them to record and track the status of each waste disposal operation and the time interval of the operation in real-time at the moment of operation and carry out incident management for each operation step. In order to carry out video monitoring, and further use the hash algorithm of image recognition to determine whether there is any abnormal situation in the waste transportation process that is more likely to be illegally discarded, and the waste temporary storage process that takes a lot of time to check. Finally, through actual field testing and cost-benefit analysis, to verify the feasibility of this smart monitoring waste cleaning method. According to the research results, the method for intelligently monitoring the waste removal and treatment process established by this research can monitor the entire waste disposal process. Image recognition can save more time and cost than using traditional manpower verification methods in the past. Since this time there is only short-term cooperation with the case, if we can import different actual fields in the future and collect more data, it is expected that the image recognition model will be more optimized, and the prediction of abnormal conditions will be more accurate. In the future, we can try to combine different sensors or image recognition technologies to have a closer analysis of abnormal conditions, which will have the opportunity to improve the efficiency and quality of monitoring. However, cost and benefit and the convenience and feasibility of technology should be considered, compared to the advantages and disadvantages of different inspection equipment and methods. |
| URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/82972 |
| DOI: | 10.6342/NTU202101392 |
| 全文授權: | 未授權 |
| 電子全文公開日期: | 2026-07-21 |
| 顯示於系所單位: | 環境工程學研究所 |
文件中的檔案:
| 檔案 | 大小 | 格式 | |
|---|---|---|---|
| ntu-109-2.pdf 未授權公開取用 | 3.22 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。
