Artificial Neutral Networks Application for Plasma Cutting Modeling

Abstract

In order to obtain the required cut quality during the plasma cutting process, the particular regularities between the elements of the cutting quality and the process input parameters should be explored. This research should take into consideration the large number of the significant factors that influence the process output parameters. In this paper, the experimental research deals with the investigation of the impact of the plasma cutting current, cutting velocity and metal thickness on the kerf surface roughness.
Based on the experimental results, the sets of tests were done in order to determine the most favourable Artificial Neural Network structure and architecture. The particular Artificial Neural Network process modelling was applied for plasma cutting modelling. The results were verified by the simulation of the Artificial Neural Network, which was done by the set of data that was not used for the network training. The simulation was done for the purpose of verifying the experimental results, where the simulated data showed the good agreement with the experimental results.