Automated Drift Compensation System for Electrical Conductivity and pH Probes in Hydroponic Systems

Highlights Used one- and two-point normalization to compensate for drifts in EC and pH probes in hydroponic systems. Developed automated sensor drift compensation systems for hydroponic systems. Evaluated the compensation systems for open- and closed-hydroponics. Application of the normalization methods increased the accuracy of the EC and pH measurements. Abstract. In hydroponic systems, the nutrient solutions are dispensed based on pH and electrical conductivity (EC) to ensure that the nutrient components in the solution are optimal for crop growth. However, the pH and EC probes may exhibit signal drifts when immersed in the solution for a considerable period, which deteriorate the measurement accuracy and the nutrient and water use efficiency. Notably, the existing hydroponic systems require manual calibration of the pH and EC probes and do not provide any information regarding their accuracy. Consequently, faults in the probes are challenging to detect. In this study, two types of automated drift compensation systems for EC and pH probes, based on one- and two-point normalization, were developed to accurately monitor changes in the pH and EC values in hydroponic solutions and enhance sensor management. The proposed framework can provide information regarding the drifts of the pH and EC probes through automated normalization solution measurements. The effectiveness of the developed system in scenarios involving plant growth with open and closed hydroponics was evaluated through comparison with a standard method involving sampling and laboratory analysis. The results indicated that the one-point normalization strategy had a simple structure and was effective in compensating drifted offsets of sensors. A two-point normalization strategy could effectively predict the sensor drifts as well as variations in the sensitivities, which could be used in the compensation processes. Application of the one-point normalization-based system increased the accuracy of the EC and pH measurements, with root mean square errors decreasing from 50 to 28.5 µS·cm -1 and 0.43 to 0.17 pH in the open hydroponic systems and from 68.7 to 38.3 µS·cm -1 and 0.15 to 0.11 pH in the closed systems, respectively. The two-point normalization-based system decreased the RMSEs from 44.6 to 33.4 µS·cm -1 and 0.55 to 0.20 pH for the EC and pH measurements, respectively. In addition, the decreased slope and coefficient of determination in the linear regression according to the system applicatio