Ma Xiaohu, Fischerauer Alice, Haacke Sebastian, Fischerauer Gerhard
Faculty of Engineering Science, University of Bayreuth, 95440 Bayreuth, Germany.
Sensor Control GmbH, 56566 Neuwied, Germany.
Sensors (Basel). 2024 Dec 19;24(24):8111. doi: 10.3390/s24248111.
Molding sand mixtures in the foundry industry are typically composed of fresh and reclaimed sands, water, and additives such as bentonite. Optimizing the control of these mixtures and the recycling of used sand after casting requires an efficient in-line monitoring method, which is currently unavailable. This study explores the potential of an AI-enhanced electrical impedance spectroscopy (EIS) system as a solution. To establish a fundamental dataset, we characterized various sand mixtures containing quartz sand, bentonite, and deionized water using EIS in the frequency range from 20 Hz to 1 MHz under laboratory conditions and also measured the water content and density of samples. Principal component analysis was applied to the EIS data to extract relevant features as input data for machine learning models. These features, combined with water content and density, were used to train regression models based on fully connected neural networks to estimate the bentonite content in the mixtures. This led to a high prediction accuracy ( = 0.94). These results demonstrate that AI-enhanced EIS has promising potential for the in-line monitoring of bulk material in the foundry industry, paving the way for optimized process control and efficient sand recycling.
铸造行业中的型砂混合物通常由新砂和再生砂、水以及膨润土等添加剂组成。优化这些混合物的控制以及铸件后废砂的回收利用需要一种高效的在线监测方法，而目前尚无此类方法。本研究探索了人工智能增强型电阻抗谱（EIS）系统作为解决方案的潜力。为了建立一个基础数据集，我们在实验室条件下，使用EIS在20 Hz至1 MHz的频率范围内对包含石英砂、膨润土和去离子水的各种砂混合物进行了表征，并测量了样品的含水量和密度。将主成分分析应用于EIS数据，以提取相关特征作为机器学习模型的输入数据。这些特征与含水量和密度相结合，用于训练基于全连接神经网络的回归模型，以估计混合物中的膨润土含量。这导致了较高的预测准确率（ = 0.94）。这些结果表明，人工智能增强型EIS在铸造行业中对散装材料的在线监测具有广阔的潜力，为优化过程控制和高效砂回收铺平了道路。
