Knowledge-based prediction of pore diameter of nanoporous anodic aluminum oxide

• Application of ANN modeling for prediction of the pore diameter.
• ANN saves time and reduces the cost of experimental studies.
• ANN can predict the pore diameter with acceptable errors.
• ANN’s and experimental study’s results are matched together.

Using nanostructured materials, especially nanoporous aluminum oxide, becomes more popular in recent years. The main purpose of this paper is developing an artificial neural network (ANN) model and conducting an experiment to predict the pore diameter of nanoporous aluminum oxide membrane. For this reason, a total of 32 experimental data are collected and used to develop the proposed model. The process parameters such as electrolyte concentration, temperature and anodization potential are considered as input, while the pore diameter is accounted for output. A comparison of ANN, experimental study and two previous empirical formulas indicates that ANN has a good predictive capability of the pore diameter. It can also forecast the experimental result with an acceptable error. The results also reveal that both empirical formulas are too conservative.