Prediction and verification of thin liquid film thickness on salt-deposited copper surface in an atmospheric hygrothermal environment

Abstract

This study used laser spectroscopy testing technology and independently constructed a testing apparatus to achieve stable measurements of the adsorbed liquid film thickness on copper surfaces. The apparatus measurement accuracy reached 91.7%. Using this apparatus, the thickness of the adsorbed liquid film on copper surfaces was measured and analyzed under different temperatures, relative humidity (RH), and salt (NaCl) deposition density conditions. According to the results, increased temperature, RH, and NaCl deposition increased the liquid film thickness. Furthermore, the liquid film thickness increased exponentially with increasing RH under the same temperature and NaCl deposition conditions. Surface fitting of the obtained liquid film thickness data yielded a fitting calculation formula for the adsorbed liquid film thickness on copper surfaces under typical atmospheric temperature conditions (25–45 °C). In addition, artificial neural networks (ANNs) and support vector machine models were constructed based on machine learning methods for predicting liquid film thickness. Comparative results indicated that the ANN prediction model exhibited higher accuracy, with a model determination coefficient (R²) reaching 0.99. Validation by comparing measured and predicted values under typical conditions for liquid film thickness showed that the machine learning-based prediction error was approximately 9.7%. This approach rapidly predicted adsorbed liquid film thickness on copper surfaces subjected to atmospheric humid and hot NaCl deposition.