ABSTRACT This article presents a novel non-destructive microwave technique for predicting fluoride contamination in pure water. The proposed microwave-based sensing technique uses an open-ended coaxial probe (OECP) microwave sensor for monitoring fluoride concentration in water. The sensor output is the input of Artificial Neural Network (ANN) for predicting the complex dielectric constant of contaminated water, which has direct correlation with fluoride contamination in water. The ANN is trained through analytically generated sensor output for various lossy liquid materials and tested for experimental data obtained through laboratory prepared samples. Hence, the proposed technique has the capability to compute the amount of fluoride contamination faster, when compared to analysis only method. The results shows that a well-trained ANN is computationally efficient and capable of predicting the amount of fluoride level in the pure water. The results also has good agreement with the data published in the literature at room temperature.

中文翻译：

---

基于人工神经网络的无损微波技术监测水中氟化物污染

摘要 本文介绍了一种用于预测纯水中氟化物污染的新型无损微波技术。所提出的基于微波的传感技术使用开放式同轴探头 (OECP) 微波传感器来监测水中的氟化物浓度。传感器输出是人工神经网络 (ANN) 的输入，用于预测受污染水的复介电常数，该介电常数与水中的氟化物污染直接相关。人工神经网络通过针对各种有损液体材料的分析生成的传感器输出进行训练，并针对通过实验室制备的样品获得的实验数据进行测试。因此，与仅分析方法相比，所提出的技术能够更快地计算氟化物污染量。结果表明，训练有素的人工神经网络在计算上是高效的，并且能够预测纯水中的氟化物含量。在室温下，结果也与文献中公布的数据有很好的一致性。