Moisture Content Detection of Tomato Leaves Based on Electrical Impedance Spectroscopy

Abstract

ABSTRACT The rapid detection of moisture content in tomato leaves can provide an important basis for tomato cultivation. Compared with spectroscopy technology, image technology, and dielectric detection technology, electrical impedance spectroscopy (EIS) has the potential to analyze the internal components of leaf microstructure qualitatively and quantitatively, and the characteristics of fast detection and low cost. The moisture content in tomato leaves was detected effectively and accurately at the anthesis of the first inflorescence using the EIS method. The influence of reduction of moisture content on the impedance characteristics and equivalent circuit parameters was analyzed. The frequencies of different impedance characteristics were selected based on the out-of-bag (OOB) importance. Different regression algorithms, including partial least square regression (PLSR), back-propagation neural network (BPNN), and random forest (RF), were used for estimating the leaf moisture content (LMC). The LMC models based on different impedance characteristics and different methods were established and compared. The results demonstrated that the impedance modulus in the low-frequency region, the extracellular resistance (R e), and the intracellular resistance (R i) increased first and then decreased with the decrease in the moisture content of the tomato leaves. The RF model established with the combination data had the best performance, with the coefficient of determination (R 2), normalized root-mean-square error (NRMSE), and the ratio of performance of deviation (RPD) for prediction being 0.8861, 6.02%, and 2.95, respectively. Therefore, the EIS method combined with RF was a feasible and effective method to predict LMC, providing a promising tool for moisture content detection.