固態離子選擇電極陣列於水耕作物之營養素組成分析

為了解決農業人口老齡化問題，智慧農業科技可將現有的生產經驗與知識保留並轉換為數位資訊，有利於建立專家系統以降低農業經營門檻與提高生產效率。本研究欲發展多離子感測試片進行高通量植物營養素組成分析之現場檢測平台，以找出較佳植物栽種條件及管控植物品質，故建立植物營養素檢測裝置。以植物工廠之水耕蔬菜作為感測樣本，自製固態離子選擇電極陣列試片 (solid-contact ion-selective electrode array chip, SCISE array chip) 作為感測器，搭配多通道電位感測模組，量測不同栽種條件下植物葉片汁液之離子濃度，並使用多變量統計分析離子組成找出植物之生理特性、健康狀態和品質等關聯。SCISE陣列試片可量測鉀、銨、鈣、鎂、氫、硝酸根、氯和鈉等八種離子濃度。為了檢測應用性，使用其量測未過濾植物汁液和養液 (濃度範圍：10-3~10-4M) ，並與離子分析儀 (IA-300) 過濾後的植物汁液和養液之結果比較，且該試片當天可重複使用至少30次。接著，以標準儀器量測之同光質不同養液栽種之植物葉片汁液之離子濃度，比較變異數分析 (ANOVA) 、多變量變異數分析 (MANOVA) 、主成分分析 (PCA) 和因素分析 (FA) 等統計分析之結果。我們發現FA可分群不同栽種條件之植物並找出栽種條件與生理特性之關係，透過盒鬚圖搭配AONVA鈉離子在統計存在顯著差異，因此為分群上的重要參數，有助於未來判斷不同栽種條件與品質管理。並以不同光質同養液栽種之植物驗證其結果，結果顯示不同光質比例應用於不同種類的蔬菜後可影響不同的生理特性，有助於預測光質對特定作物生理特性的影響並進而掌握品質。最後，使用逐步迴歸 (stepwise regression) 建立最適化模型和擬合公式，其結果有助於改良養液與光質配方，並可應用於發展專家系統。本研究發展植物營養素檢測裝置，可快速量測多種離子濃度並建立出代數值模型，未來可利用此技術判斷栽種條件之合宜性，並掌握植物生理特性和健康狀態，大到品質管理與產量提升。

關鍵字

植物工廠；營養素分析；固態式離子選擇電極；多變量統計；汁液分析

並列摘要

Smart agricultural technology can retain the existing production experience and knowledge, and convert it into digital information to solve the aging problem of agricultural population, which will help to establish an expert system to lower the operation threshold and improve agricultural productivity. We intend to develop an on-site detection platform for high-throughput plant nutrient composition analysis based on a multi-ion sensing chip to find better growth conditions and control plant quality. Thus, a plant nutrient detection device based on a solid-contact ion-selective electrode (SCISE) array chip combined with a portable multi-channel potentiometric model was developed in this study, which was used to measure ion concentration in plant sap growing under different conditions. Through multivariate statistics, the ion concentration would be transformed into the physiological characteristics, health status, and quality in plants. Hydroponic vegetables from the plant factory were used as sensing samples. The SCISE array chip consists of potassium, ammonium, calcium, magnesium, hydrogen, nitrate, chlorine, and sodium ion-selective electrodes. In order to prove this application, the crude vegetable saps and nutrient solutions (concentration range: 10-3~10-4M) were determined by the device and the results were compared with those measured by ion chromatography (IA-300). To test the repeatability, the same SCISE array chip was measured for 30 times on the same day. The ion concentrations of leaf sap in plants growing under the same light and different nutrient solutions were measured by ion chromatography. The results were compared in different statistical analyses (like analysis of variance (ANOVA), multivariate analysis of variance (MANOVA), principal component analysis (PCA), and factor analysis (FA)). We found that the results can group plants grown in different conditions and find out the relationship between growth conditions and physiological characteristics by factor analysis. There were statistically significant differences in sodium ions between plants grown in different conditions through boxplot combined with AONVA. Therefore, the sodium ion is an important parameter for grouping plants growing under different conditions, which is helpful for controlling the growth condition and quality of plants. The results were verified by plants growing under different light quality and the same nutrient solution. The results showed that different light quality ratios applied to different kinds of plants can affect different physiological characteristics, which is helpful to predict the effect of light quality on the physiological characteristics of specific crops and grasp the quality. Finally, an optimal model and a fitting formula were established by stepwise regression. The results would contribute to the improvement of nutrient solution formula and light quality and could be applied to the development of expert systems. In conclusion, this study developed a plant nutrient detection device that can quickly measure the concentration of various ion and establish the linear models and fitting formulas. In the future, our smart agricultural technology will be used to judge the suitability of growth conditions, and also master the physiological characteristics and health status of plants, resulting in, quality management and yield increase.