物聯網即時離子濃度輪廓技術探討環境蒸氣壓差與養液導電度對芝麻菜水耕栽培之研究

Abstract

目前水耕栽培產業普遍依據養液的導電度(Electrical Conductivity, EC)來進行養液管理，但單靠導電度無法確認養液中各種離子的濃度。若某些離子濃度不足或過高，不僅可能導致作物出現特定營養缺乏的生理性病害，也可能引發肥傷症狀。為解決此問題，本研究群先前開發出基於固態接觸式離子選擇電極的物聯網養液感測系統(IoT-based Nutrient Sensing System, IoNSS)，能夠即時監測養液的離子濃度。然而在商用溫室測試中發現，IoNSS容易受環境雜訊干擾，使系統量測的離子濃度精確度較差。此外，戶外氣候變化也可能造成溫室內部溫濕度驟變，進而導致環境蒸氣壓差(Vapor Pressure Deficit, VPD)改變，間接影響養液中即時離子濃度的變化。透過實驗分析發現環境雜訊主要為60Hz及其諧波頻率，因此本研究首先對IoNSS進行改良。為降低離子選擇電極的電位飄移，將3種類比低通濾波器(RC濾波器、Sallen-Key濾波器與多重回授濾波器)整合至IoNSS中。結果顯示，Sallen-Key濾波器能有效平衡雜訊抑制及安定時間，並提高IoNSS量測濃度的精確度。後續將開發出的物聯網環境模擬系統(IoT-based Environment Simulation System, IoESS)結合IoNSS，設計在適中VPD(0.9 kPa)與高VPD(1.6 kPa)環境同時以低EC(1 mS/cm)及高EC(2.5 mS/cm)養液栽培水耕芝麻菜，並即時監測養液離子濃度變化，探討環境VPD與養液EC對養液濃度及芝麻菜(Eruca sativa)體內離子的影響。結果顯示，在低EC養液中栽培時，高VPD環境使後期養液NO3-濃度下降趨勢減緩，並造成芝麻菜體內NO3-增加。Ca2+吸收量雖未受抑制，但養液Ca2+濃度下降幅度減緩，甚至上升。養液K+濃度波動，容易使濃度突然上升，且芝麻菜K+含量下降，可能影響葉片水分調節，增加失衡風險。然而，若在高VPD環境中提高養液EC，反而降低K+、NO3-、Ca2+吸收量，表示異常環境下，過量提升養液濃度，不但無助離子吸收，反而提高肥傷風險。因此，透過本研究的系統，未來可整合即時離子濃度、環境參數等資訊，並導入人工智慧(Artificial Intelligence, AI)模型，建立一個即時診斷平台，可預測目前環境下預期的養液離子濃度趨勢並提前發出預警，提升作物品質及產量。

Currently, hydroponic cultivation industries commonly manage nutrient solutions based on electrical conductivity (EC). However, EC alone cannot accurately reflect the concentrations of individual ions in the solution. Deficiencies or excesses of specific ions may lead to physiological disorders caused by nutrient imbalances or fertilizer burn symptoms in crops. To address this issue, our research team previously developed an IoT-based Nutrient Sensing System (IoNSS) utilizing solid-contact ion-selective electrodes, enabling real-time monitoring of nutrient solution ion concentrations. However, field tests in commercial greenhouses revealed that the IoNSS is susceptible to environmental noise interference, resulting in reduced accuracy of ion concentration measurements. Additionally, sudden changes in outdoor weather can cause rapid fluctuations in temperature and humidity inside greenhouses, altering the Vapor Pressure Deficit (VPD) and indirectly affecting ion concentrations in the nutrient solution. Experiment analysis identified 60 Hz and its harmonic frequencies as the primary sources of noise. To mitigate this, the IoNSS was enhanced by integrating three types of analog low-pass filters (RC filter, Sallen-Key filter, and multiple feedback filter) to reduce the potential drift of ion-selective electrodes. Among them, the Sallen-Key filter effectively balanced noise suppression and settling time, improving the precision of ion concentration measurements. Subsequently, the developed IoNSS was integrated with an IoT-based Environment Simulation System (IoESS) to investigate the effects of environmental VPD and nutrient solution EC on ion concentrations in the solution and in hydroponically grown arugula (Eruca sativa). The experiment was conducted under moderate (0.9 kPa) and high (1.6 kPa) VPD conditions, using both low (1 mS/cm) and high (2.5 mS/cm) EC nutrient solutions. Results showed that under low EC conditions, high VPD slowed the decline of NO3- concentration in the nutrient solution during the later growth stage, leading to an increase in NO3- content within arugula. Although Ca2+ uptake was not inhibited, the decrease in Ca2+ concentration in the solution slowed or even reversed. K+ concentration in the solution fluctuated, occasionally spiking, while K+ content in the plants decreased, potentially impairing leaf water regulation and increasing the risk of imbalance. Conversely, increasing EC under high VPD conditions reduced the uptake of K+, NO3-, and Ca2+, suggesting that excessive nutrient supply in stressful environments not only fails to promote ion absorption but also elevates the risk of fertilizer burn. In the future, this system could be integrated with real-time ion and environmental data, combined with Artificial Intelligence (AI) models, to establish a real-time diagnostic platform. Such a platform could predict expected ion concentration trends under current conditions and issue early warnings, ultimately improving crop quality and yield.