2021 article

Towards Continuous Plant Bioimpedance Fitting and Parameter Estimation

2021 IEEE SENSORS.

co-author countries: United States of America 🇺🇸
author keywords: bioimpedance parameter extraction; equivalent circuit modelling; precision agriculture; cyberphysical systems
Source: Web Of Science
Added: February 28, 2022

The push to advance artificial intelligence, internet of things, and big data analysis all pave the way to automated and systematic optimization in precision agriculture and smart farming applications. These advancements lead to many benefits, including the optimization of primary production, prevention of spoilage via supply chain management, and detection of crop failure risk. Noninvasive impedance sensors serve as a promising candidate for monitoring plant health wirelessly and play a major role in this optimization problem. In this study, we developed a software pipeline to support impedance sensing applications and, as a proof of concept, applied this to track longitudinal consistent bioimpedance data from the V4 leaf midrib in maize plants. The script uses the single-shell equivalent circuit model to represent the extracellular fluid, cellular membrane, and intracellular fluid as a simplified resistive-capacitive circuit, where these elements’ parameters are estimated with complex nonlinear least squares. The double-shell model extends the single-shell model to account for the effects of the relatively large plant cell vacuole. Limit cases for impedance are utilized for specific parameters as an alternative method of estimation. We investigated a complex analysis-based modification to the objective function and model optimization for the data pipeline automation. Various weighing functions are applied and checked against one another. Additionally, a custom graphical user interface was developed to assist with parameter initialization for correcting potential convergence issues and understating the influence of each parameter on the dataset. We demonstrated that the analysis of an example longitudinal dataset was able to reveal a time series for parameter fitting.