Plant growth monitoring, prediction, and self-regulation utilizing MXene/CNTs/TPU flexible strain sensors integrated with deep learning algorithms and soft actuators

Abstract

Smart agriculture utilizes sensors and software to control agricultural production through mobile or computer platforms, enabling unmanned, automated, and intelligent management. Recently, research and development in plant growth monitoring technologies have garnered significant attention. The challenge lies in achieving long-term monitoring, phased predictions, and plant self-regulation without harming the plants. The present study demonstrates the fabrication of plant-compatible and breathable tensile and bending strain sensors using composite nanofiber membranes (CNMs) composed of Ti₂C₂T_x (MXene), carbon nanotubes (CNTs), and thermoplastic polyurethanes (TPU) through electrospinning and ultrasonic immersion techniques. The MXene and CNTs synergistically form a dual-network conductive structure on the TPU nanofiber membrane, thereby imparting the composite membrane with remarkable tensile sensitivity (5.41, 7.39, and 3.39 within the ranges of 0%–20%, 20%–50%, and 50%–70%, respectively) as well as exceptional bending sensitivity (1.79, 0.89, and 0.46 within the ranges of 0°–30°, 30°–90°, and 90°–120°, respectively). The tensile strain sensor, combined with a deep learning Long Short-Term Memory (LSTM) model, establishes a platform for plant growth monitoring and prediction. The bending strain sensor, integrated with a shape memory alloy (SMA)-based soft actuator, forms a plant sensing-actuating system to assist in plant leaf growth. This work leverages MXene/CNTs/TPU CNMs to flexibly prepare strain sensors for specific applications, combining deep learning and soft actuators to achieve plant growth prediction and self-regulation. This research holds significant importance in advancing the development of smart agriculture.