Tazifor Tchantcho Martial, Zimmermann Egon, Huisman Johan Alexander, Dick Markus, Mester Achim, van Waasen Stefan
Central Institute of Engineering, Electronics and Analytics (ZEA-2), Forschungszentrum Juelich GmbH, 52428 Juelich, Germany.
Institute of Bio- and Geosciences Agrosphere (IBG-3), Forschungszentrum Juelich GmbH, 52428 Juelich, Germany.
Sensors (Basel). 2023 Aug 22;23(17):7322. doi: 10.3390/s23177322.
Electromagnetic induction (EMI) systems are used for mapping the soil's electrical conductivity in near-surface applications. EMI measurements are commonly affected by time-varying external environmental factors, with temperature fluctuations being a big contributing factor. This makes it challenging to obtain stable and reliable data from EMI measurements. To mitigate these temperature drift effects, it is customary to perform a temperature drift calibration of the instrument in a temperature-controlled environment. This involves recording the apparent electrical conductivity (ECa) values at specific temperatures to obtain a look-up table that can subsequently be used for static ECa drift correction. However, static drift correction does not account for the delayed thermal variations of the system components, which affects the accuracy of drift correction. Here, a drift correction approach is presented that accounts for delayed thermal variations of EMI system components using two low-pass filters (LPF). Scenarios with uniform and non-uniform temperature distributions in the measurement device are both considered. The approach is developed using a total of 15 measurements with a custom-made EMI device in a wide range of temperature conditions ranging from 10 °C to 50 °C. The EMI device is equipped with eight temperature sensors spread across the device that simultaneously measure the internal ambient temperature during measurements. To parameterize the proposed correction approach, a global optimization algorithm called Shuffled Complex Evolution (SCE-UA) was used for efficient estimation of the calibration parameters. Using the presented drift model to perform corrections for each individual measurement resulted in a root mean square error (RMSE) of <1 mSm for all 15 measurements. This shows that the drift model can properly describe the drift of the measurement device. Performing a drift correction simultaneously for all datasets resulted in a RMSE <1.2 mSm, which is considerably lower than the RMSE values of up to 4.5 mSm obtained when using only a single LPF to perform drift corrections. This shows that the presented drift correction method based on two LPFs is more appropriate and effective for mitigating temperature drift effects.
电磁感应（EMI）系统用于近地表应用中绘制土壤的电导率。EMI测量通常受随时间变化的外部环境因素影响，温度波动是一个重要因素。这使得从EMI测量中获取稳定可靠的数据具有挑战性。为减轻这些温度漂移影响，通常在温度受控环境中对仪器进行温度漂移校准。这包括记录特定温度下的表观电导率（ECa）值，以获得一个查找表，该表随后可用于静态ECa漂移校正。然而，静态漂移校正未考虑系统组件的延迟热变化，这会影响漂移校正的准确性。在此，提出一种漂移校正方法，该方法使用两个低通滤波器（LPF）来考虑EMI系统组件的延迟热变化。同时考虑了测量设备中温度分布均匀和不均匀的情况。该方法是在10°C至50°C的广泛温度条件下，使用定制的EMI设备进行总共15次测量而开发的。该EMI设备配备有八个分布在设备各处的温度传感器，在测量期间同时测量内部环境温度。为了对所提出的校正方法进行参数化，使用了一种称为洗牌复合进化（SCE-UA）的全局优化算法来有效估计校准参数。使用所提出的漂移模型对每个单独测量进行校正，所有15次测量的均方根误差（RMSE）<1 mS/m。这表明漂移模型可以正确描述测量设备的漂移。对所有数据集同时进行漂移校正，得到的RMSE<1.2 mS/m，这大大低于仅使用单个LPF进行漂移校正时获得的高达4.5 mS/m的RMSE值。这表明所提出的基于两个LPF的漂移校正方法对于减轻温度漂移影响更合适且有效。
