Modelling of the anodizing process of aluminum using neural networks

The aim of the research work was to present some possibilities of control and optimization of the technological process of aluminum anodic oxidation using neural networks and Design of Experiments (DoE) in order to evaluate and monitor the influence of the input factors on the resulting AAO (Anodic aluminum oxide) film thickness. Three types of neural units (first order neural unit, second order neural unit, third order neural unit) were used to create the prediction model describing the thickness of the final aluminium oxide layer formed during the process of anodic oxidation of aluminum. The paper also deals with the evaluating of minimal range of training data used for learning process, so the neural unit can produce sufficiently reliable model.