工程學

Agro-guard aims to revolutionize disease identification and community-based projects in the field of agriculture. Integrating Machine learning, Computer vision, clustering, and community-based technology, this project helped farmers to detect their plant disease with their solution and for early warning of plant disease which was spreading in their community which helped in crop management. The research project is divided into three parts.First,Integrating Machine learning to detect and classify plant disease with their solutions.Second,Integrating Density-Based Spatial Clustering of Applications with Noise (DBSCAN),to identify disease and analyze the pattern within agricultural regions.Third,Establishing notification system to notify real-time alerts to farmers about disease spreading in particular region.The research is crucial because it solve one of the crucial problem of our community which is untimely detection of disease.The finding of the research highlight the effectiveness of Agro-Guard framework in early disease detection and community detection.The machine learning models achieved high accuracy in identifying common plant disease and clustering results the pattern in diseases that were very important for notifying the community.The significance of these findings is that it can build powerful system which will overall grow the production of crops and plants due to timely update of the disease prevailing in the community.It contributes in sustainability production of crops and plants which ultimately ensure the good livelihood of farmer.

Due to global population growth and industrialization, excessive inflow of causative nitrogen into rivers, and the increase in water temperature due to global warming, the occurrence of harmful algal blooms (HABs) is increasing. HABs can cause not only ecological destruction but also various social and economic problems. Additionally, consuming water from lakes with abundant toxic cyanobacteria can lead to liver damage, vomiting, abdominal pain, and even death if consumed over a long period. The first recorded occurrence of animal mortality due to HABs was in Australia in 1878, and since then, livestock and wildlife have suffered damages from HABs worldwide. Furthermore, the United States' Lake Erie has experienced frequent HAB occurrences since 2011, and in 2007, China faced social disruption when a massive HAB outbreak in Lake Tai, one of the freshwater lakes, resulted in a suspension of the water supply. In order to address these HAB occurrence issues and assess the severity of HAB events, several systems have already been established and potential solutions have been proposed. However, these systems have limitations such as being highly systematic and advanced in terms of equipment and configuration. They are often located only in periodically affected areas, and they involve substantial costs. Therefore, we aim to overcome these limitations and design a system that can effectively manage HABs.

本研究探討臺灣常見的鋼筋華廈、鋼筋及鋼骨大樓，透過於地下室外牆與連續壁間填入非牛頓流體、牛頓流體及輕黏土，比較建物受震時加速度，發現地下結構中設置非牛頓流體減震裝置較牛頓流體、輕黏土更減震。而模擬器搖晃20-50秒時，非牛頓流體能顯著的減震，超過50秒後，非牛頓流體可能因沉澱而減震效果下降；在100秒後，無減震裝置的建築加速度上升，非牛頓流體再次出現明顯的減震效果；不同重心的建築質量分布導致不同的擴溶現象，使減震效果發生變化；較高的建物因力臂較長，重心高時產生較大的加速度。接著觀察光穿透吉利丁凍的偏折情形，發現受力面與地震方向垂直時，牆面受力明顯；若受力面和搖晃方向不垂直，柱的部分受力大，且觀察到力量有轉移的現象。最後，為建築設計超聲波測距模組，即時監測建築下陷或傾斜情形，以利即時修繕及重建。

從家用微波爐到3C產品，皆有微波電磁干擾 (Microwave Electromagnetic Interference, MWEMI)的防蔽設計需求。依據2020年科展競賽作品中的薄層微量雙金屬催化技術可製得 Graphene/Ag-doped異質結構，但該作品未曾研究此材料的防蔽 EMI 能力。本研究優化此技術，成功將銀奈米結構沉積在石墨烯的缺陷及晶界邊緣，相關技術與實驗參數（化學氣相沉積法的加熱溫度、時長和通氣量）已發表於2023年科展競賽作品。本研究進階發現 Graphene/Ag-doped異質結構具高透明度和屏蔽 MWEMI 的能力，單層膜可屏蔽60%之 MWEMI，效能優於文獻上記載的2~3層石墨烯。而依據2023年文獻說明單層奈米碳管 (MWCNTs)異質結構經氟化處理後，可大幅提升 MWEMI 屏蔽效能，因此我們也將 Graphene/Ag-doped材料經由四氟化碳電漿處理，氟化後的屏蔽效能又比單層石墨烯-銀高出15倍，可遮蔽高達99.9%的 MWEMI。

本研究以 zig-zag path 管道自製、設計聲子晶體，並結合空腔(cavity)形成類似電路學的 RLC 共振結構，透過 COMSOL 軟體模擬出其理論之特徵頻率，使空腔中心聲壓達到穩定的放大效果。利用無響室的環境進行實驗，藉由放大電路將聲壓轉為電訊號並以聲源頻率、裝置距離和空腔尺寸作為變因，發現我們自製的聲子晶體可成功將電壓放大約 2.6 倍。此種聲子晶體具有體積小、構造簡單、可擴充為多晶體結構等特性，在未來可結合壓電材料，將其製成兼具發電與降噪功能之環保裝置。

本研究旨在開發一款能自動跟隨人的AI智慧運輸車，應用在 醫院、商場等環境，能減輕工作人員的負擔。本系統融合YOLOv8影像辨識技術和物聯網，使運輸車具備跟隨人體的能力。該系統通過物聯網，將車上攝影機接收的影像傳送至電腦，進行即時運算，不必在車上配備高階的微型電腦或GPU以降低成本。使用YOLOv8模型辨識人體與手勢，以arduino ESP32開發板作為主控單元控制減速馬達，使運輸車能自動跟隨人員。本研究不像傳統自動導引車（AGV），因為依賴固定路徑，而無法用在多變的場景，且不適合與人協同工作。將本系統應用於醫療院所或零售業商場等 ，經常出現 變障礙物的場所時，能與人員配合執行任務，例如當作病人的點滴架、輔助護理師工作的醫務車，以及賣場中協助工作人員上貨的籠車，能有效降低勞力負擔。

光柵作為常見的分光元件，應用於許多光學儀器中。然而，傳統光柵彈性較差且硬度較高，限制了其應用範圍。本研究利用具有高彈性和易形變特性的 PDMS 作為柔性光柵的材料，對不同厚度和彎曲程度的光柵進行一系列測試。為了探討厚度和彎曲曲率對繞射效果的影響，進行了不同厚度柔性光柵的繞射點分析實驗。實驗結果顯示，增加柔性光柵的厚度會提升其彎曲時第一亮紋的變化率；相反，減少厚度則會降低該變化率。隨後的研究進一步探討不同施力方式是否會影響柔性光柵的分光效果。通過拉伸和壓縮光柵，發現拉伸會使光柵的軌距變大，而壓縮則會使軌距變小。此外，研究還探討了利用 PDMS 複製類似光柵的結構是否具有分光效果。實驗結果顯示，複製的指紋確實展現了類似特性，期望未來能夠將這些特性實際應用於相關領域。

This paper describes a power-profile-based approach to CanSat picosatellite energy usage estimation. We measured the power profile of individual CanSat components on a test bench. The collected data was used to create a power consumption model for a simple flight control algorithm. The power consumption model was implemented in Python programming language and was used to simulate the CanSat flight. The simulation results were compared with the measured power profile of a fully integrated CanSat running the same flight control algorithm. The model showed an error of 2.2% and proved to be suitable for use in battery capacity optimization.

本研究專為因聽覺神經腫瘤手術造成單側聽力喪失(Single-Sided Deafness, SSD)的聽損患者設計了一款辨別聲音方位的輔助裝置。SSD患者因為有一側的聽力完全喪失，因此現有的助聽器無法提供有效的協助。 本研究採納各式麥克風電路的優點，設計出一款麥克風模組來接收聲音。並藉由帽子周圍的六個麥克風模組形成陣列。利用頭影效應所帶來的差異，運算出聲音的方位，並且透過馬達震動來提醒使用者聲音的方位。本作品延續並參考了之前的作品[一] [二]並經過改良，經過實驗後發現，本裝置可以偵測到以使用者為圓心半徑23公尺左右、76分貝的聲音，偵測距離約為之前作品的2.3倍，可以為使用者提供3到6秒的反應時間。希望可以藉由這種方式，讓SSD的聽損患者更安心地走在路上。

隨著矽基電晶體逐漸微縮，其元件效能將接近其物理極限，二硫化鉬 (MoS2) 等二維材料藉著其獨特的特性（如寬的能隙、高電流開關比及優異的載子遷移率等），可作爲矽的替代材料用於未來的電子科技應用。本研究旨在製造MoS₂ 的場效電晶體並研究元件之低溫特性。我們成功利用機械剝離法製備並轉移二維 MoS2薄膜至二氧化矽/矽基板上，並且製造MoS₂ 場效電晶體，並量測其室溫(300 K)至極低溫(~ 4 K)的電流特性，元件在此溫度範圍中具有優異的特性，能有效地調控電流調控，表現出良好的下閘極控制能力，同時具有低次臨界擺幅及高電流開關比(~ 106)。在極低的溫度(4 K)下，該電晶體仍能保持良好的運作，顯示出MoS₂應用於低功耗且高元件效能的低溫電子元件的潛力。