| Soil characteristics mainly include water content,acidity and alkalinity,and salinity.Traditional soil feature extraction techniques and equipment are complex and can cause significant damage to the soil.Ultra-wideband（UWB）sensors are non-invasive sensors with signals that have good penetration and high resolution,making them suitable for non-destructive extraction of soil characteristics.The spatial distribution prediction of soil characteristics refers to using the characteristics of some positions of soil to predict other positions in space.Kriging interpolation is the most common spatial distribution prediction method in soil research,but it has many limitations and drawbacks in practical applications that need to be improved.Therefore,this article uses the UWB sensor Pulse On?P440 for soil characteristic signal acquisition,and based on this,carries out research on non-destructive extraction and spatial distribution prediction of soil characteristics.The specific work is as follows:1.Using UWB sensors and drones to collect soil echo signals for future research.2.To extract soil moisture features from ultra-wideband（UWB）echoes,a soil moisture extraction system based on time-frequency analysis and neural networks was constructed.Experimental studies were conducted to investigate the factors that affect the convergence speed and accuracy of the system.The results showed that the system was able to effectively extract soil moisture features from UWB echoes.The time-domain resolution of the time-frequency analysis method was identified as an important factor affecting the convergence speed and maximum accuracy of the system.3.In order to predict the soil moisture distribution in ridge-structured soil,a comparison was made between the kriging interpolation method and ordinary interpolation methods in terms of their accuracy in predicting soil moisture.The experimental results showed that the traditional kriging interpolation method had higher accuracy in predicting the spatial distribution of soil moisture in ridge-structured soil compared to ordinary interpolation methods.However,in some cases,there were still significant errors in soil moisture predictions.4.To address the issue of large prediction errors in traditional kriging interpolation,an improved kriging interpolation method was proposed and its predictive accuracy was compared with that of the traditional kriging interpolation method.The experimental results demonstrated that the proposed method effectively resolved the low accuracy issue of traditional kriging interpolation in predicting the spatial distribution of soil moisture in ridge-structured soil.Building upon this,the improved kriging interpolation method was combined with the UWB-based soil moisture extraction system proposed in this study to achieve the prediction of soil moisture spatial distribution.In summary,this article proposes a non-destructive method for extracting soil moisture based on time-frequency analysis and neural network,providing a new approach to extract soil moisture from ultra-wideband signals.The introduced Krige interpolation method with a stimulus factor excellently solves the problem of low prediction accuracy of traditional Krige interpolation method for soil moisture under complex ridge structure.Combining these two methods can achieve rapid monitoring of soil moisture in large-scale field ridge structures,which has great practical value in agricultural development. |
