Plant vitality is an important indicator of plant health. Previous studies have often assessed plant vitality using related physiological parameters, but few studies have examined the effects of changes in plant vitality on stem water content (StWC), which can be measured online, in real time, and nondestructively using a novel fringing impedance sensor. In the present study, the sensor calibration results showed a linear fitting relationship between the sensor output voltage and StWC, with coefficients reaching 0.96. The coefficients of correlations between StWC and four plant physiological parameters related to plant vitality (net photosynthetic rate, transpiration rate, stomatal conductance, and intercellular carbon dioxide concentration) were more than 0.8, indicating that StWC can be used to characterize plant vitality to a certain extent. A comparison between plants with normal vitality and weakened vitality showed that the self-regulation ability of plants gradually weakened as the plant vitality decreased, the diurnal mean of StWC lowered, and the diurnal range of StWC increased. In conclusion, StWC can be used as a new parameter to assess plant vitality.

To address the problems in the calibration of soil water content sensors, in this study, we designed a low-cost edge electromagnetic field induction (EEMFI) sensor for soil water content measurement and proposed a normalized calibration method to eliminate the errors caused by the measurement sensor’s characteristics and improve the probe’s consistency, replaceability, and calibration efficiency. The model calibration curve-fitting coefficients of the EEMFI sensors were above 0.98, which indicated a significant correlation. The experimental results of the static and dynamic characteristics showed that the measurement range of the sensor varied from 0% to 100% saturation, measurement accuracy was within ±2%, the maximum value of the extreme difference of the stability test was 1.09%, the resolution was 0.05%, the delay time was 3.9 s, and the effective measurement diameter of the EEMFI sensor probe was 10 cm. The linear fit coefficient of determination of the results was greater than 0.99, and the maximum absolute error of the measurement results with the drying method was less than ±2%, which meets the requirements of soil water content measurement in agriculture and forestry fields. The field experiment results further showed that the EEMFI sensor can accurately respond to changes in soil water content, indicating that the EEMFI sensor is reliable.

Deploying deep learning models in embedded terminals is essential for applications with real-time reasoning requirements. In order to make the model run efficiently in the embedded end with limited resources, we propose a model compression method combining multi-factor channel pruning and knowledge distillation. In the process of network sparsity, this method uses the double factors of the BN layer to improve the pruning standard and guides the local pruning of the model according to the new standard to ensure the compression rate. In order to further improve the accuracy, we use the knowledge transfer method in the idea of knowledge distillation to fine-tune the model and use the more continuous parameter distribution of the student model to ensure accuracy. We use several deep learning models to test. The experimental results show that the proposed model compression method has the advantages of fewer parameters and higher accuracy, which reduces the resources occupied by the model application to the embedded end. More importantly, this method not only realizes the efficient operation of various models in the embedded end, but also ensures the high accuracy of the model.

Soil water sensors based on the standing wave rate (SWR) principle are affected by temperature in long-term operation. To address this problem, a temperature compensation model based on the binary regression analysis method is proposed. The measurement results of the temperature-compensated standing wave rate (TCSWR) sensor at different temperatures and soil volumetric water content are analyzed, and the least-squares principle is used to identify the parameters to be determined in the compensation model for temperature for the SWR soil water sensor. A portable tapered TCSWR sensor with built-in temperature compensation model was developed on this basis. The calibration results show that the standing wave measurement circuit of the TCSWR sensor can effectively respond to changes in soil water, and the coefficient of the fitted equation exceeds 0.95. A comparison of the results before and after temperature compensation proves that compensation can significantly reduce the measurement error of the TCSWR sensor and improve the measurement accuracy. The static and dynamic characteristics of the TCSWR sensor show that the measurement range of the TCSWR sensor is7.50%-31.50%, the measurement accuracy is ±0.63%, the stability is good, the resolution is a minimum of 0.05%, and the dynamic response time is less than 1 s. The absolute error of the TCSWR sensor measurement is less than 1% in comparison with similar sensors, demonstrating that the measurement results of the TCSWR sensor are reliable.