杭菊智慧烘培技術提升之研究

苗栗縣銅鑼鄉為台灣最大的杭菊產地，目前面積約有50公頃，為稻轉作頗具潛力之特色作物，為農民帶來了高經濟價值。杭菊烘培過程中，乾燥溫度控制對杭菊影響大，多個工作環節需倚賴人工判斷，例如加工製程條件判斷、控制、調整等。有鑑於智慧物聯網（AIoT）等開發技術興起，本研究導入人工智慧技術，收集杭菊的烘培曲線、建立最佳模型，藉由自動偵測、自動控制，做更深層的數據分析，減少人力需求，並提升杭菊的烘培技術與成效。本研究主要探討要因為杭菊低溫乾燥製程中杭菊成品的含水率，剔除目前製程中的不穩定因子，使杭菊含水率可以達到品質標準的要求。使用田口方法最終得到最佳參數組合為A3、B1、C3、D1），也就是在「一階烘培溫度（A）55℃、一階烘培時間（B）2小時、二階烘培溫度（C）55℃、二階烘培時間（D）1小時」製程下烘培杭菊，確實可以提昇杭菊烘培製程，使杭菊含水率滿足品質規格界限內。

關鍵字

人工智慧型；杭菊；智慧物聯網；烘培曲線；田口實驗

並列摘要

Tongluo Township, Miaoli County is the largest Hang Chrysanthemum producing area in Taiwan, with an area of about 50 hectares at present. It is a characteristic crop with considerable potential for rice conversion, bringing high economic value to farmers. During the baking process of Hang Chrysanthemum, the drying temperature control has a great influence on chrysanthemums, and many work links need to rely on manual judgment, such as the judgment, control, and adjustment of processing conditions. In view of the rise of AIoT and other development technologies, this research introduces artificial intelligence technology, collects the baking curves of Hang Chrysanthemum, establishes the best model, and conducts deeper data analysis through automatic detection and automatic control to reduce Manpower needs, and improve the baking technology and effectiveness of Hang Chrysanthemum. This study mainly discusses the moisture content of the finished product of Chrysanthemum in the low-temperature drying process of Chrysanthemum, eliminating the unstable factors in the current process, so that the moisture content of Chrysanthemum can meet the quality standard requirements. Using the Taguchi method, the best combination of parameters is finally obtained as A3, B1, C3, D1), that is, in the "first-order baking temperature (A) 55°C, first-order baking time (B) 2 hours, second-order baking temperature ( C) Roasting chrysanthemum at 55°C and second-stage baking time (D) 1 hour" can indeed improve the baking process of chrysanthemum and make the moisture content of chrysanthemum meet the quality specification limit.