An effective FDTD model for GPR to detect the material of hard objects buried in tillage soil layer

It is difficult to systematically classify and detect objects buried in tillage soil layers. There is no effective way to detect destructive objects in the soil. This paper presents a method and model of intelligently classifying hard objects of electromagnetic wave based on Finite Difference Time Domain (FDTD), especially the volume and shape of the object can destroy agricultural machinery in farming. It aims to improve the environmental detection technology of agricultural machinery and break through the limit of intelligent equipment. Firstly, in order to calculate the energy amplitude loss function of plane electromagnetic waves in farmland soil layer, the optimal Soil Layer and Electromagnetic Wave Model (SEM) is determined by using Blackman-Harris pulse as the excitation source; The absorption boundary condition of Uniaxial Perfectly Matched Layer (UPML) is also determined by test. Secondly, the forward modeling is performed by REFLEX and MATLAB software programming, the grid type forward modeling results, compared to electromagnetic wave results and real-time forward measured data by GPR in layered soil, there were verification and analysis. Measuring the relationship between different phase amplitude and electromagnetic wave energy between different materials, proved by experiment, the optimized 750 MHz excitation source function was verified experimentally and accuracy rate reached to 0.98 or 0.99 with FDTD model. The precision of actual depth and calculated depth of the object are not more than 0.02 m. The multi-channel amplitude analysis of five kinds of material of electromagnetic waves of metal, wood, PVC plastic, dry hard stone, and glass is obtained by SEM in different soil layers, and the classification accuracy is 100%, the optimal value of energy discrimination peak amplitudes of these five materials are 0.0183, 0.0028, 0.0043, 0.0069 and 0.0087. The research shows that the FDTD difference method is used to simulate the propagation mechanism of electromagnetic waves in the soil. The SEM of GPR forward modeling can be established efficiently and accurately. The results can provide theoretical and technical support for the GPR soil matter or objects classification system and agricultural machine farming modules to detect objects buried in the soil layer.