Roughness is an important physical property of soil surface, and the estimation of soil surface roughness (SSR) has widely applications in the working performance evaluation and parameter control of agricultural machinery. This paper analyzed the influence of SSR on the reflection characteristics of ultrasonic beam and the subsequent changes in the receiver output voltage signal. Then, taking the peak output voltage , median value time interval , single-cycle voltage integral and peak time interval of the received voltage signal as inputs, an SSR estimation model based on adaptive neuro-fuzzy inference system (ANFIS) was established. The experimental results showed that the average estimation error was less than 18% with the measurement distance in 40 – 45 cm. The proposed method is different from laser scanning, depth camera and other non-contact measurement methods to obtain a 3D shape of the soil surface through the point cloud information, but directly estimates the SSR according to the reflection characteristics of the ultrasonic beam. Therefore, it is convenient and efficient, and the measurement component is mainly an ultrasonic sensor. The measurement principle determines that it has a good anti-interference performance to the ambient light intensity, dust and machine vibration, and can well meet the real-time requirements.

中文翻译：

---

基于超声回波信号特征的农业土壤表面粗糙度估计

粗糙度是土壤表面的重要物理性质，土壤表面粗糙度（SSR）的估计在农业机械的工作性能评估和参数控制中具有广泛的应用。分析了SSR对超声波束反射特性的影响以及随之而来的接收器输出电压信号的变化。然后，以接收电压信号的峰值输出电压 、中值时间间隔 、单周期电压积分和峰值时间间隔为输入，建立了基于自适应神经模糊推理系统（ANFIS）的SSR估计模型。实验结果表明，测量距离在40~45 cm范围内，平均估计误差小于18%。该方法不同于激光扫描、深度相机等非接触式测量方法通过点云信息获取土壤表面的3D形状，而是根据超声波束的反射特性直接估计SSR。因此方便、高效，测量元件主要是超声波传感器。该测量原理决定了其对环境光强、灰尘和机器振动具有良好的抗干扰性能，能够很好地满足实时性要求。