Each term refers to a management procedure that implements a different operating philosophy (greater importance for economic factors, quality factors, crossing time factors, use of equipment, etc.). These terms represented the independent variables needed to conduct an experiment using neural models aimed at identifying the most suitable setting to meet a specific series of requirements under special operating conditions.
The independent variables were used to create a Ball Design in which only n-1 terms of the n total terms are really independent and distributed over a portion of a spherical hypersurface:
Starting from the predefined experimental points, an experiment was performed utilising the simulation and a learning database was developed for the neural network. The neural network architecture was obtained directly from a neural model studied for a similar problem.
The results of the experiment (summarized in figures 2 & 3) are satisfactory in terms of the performances achieved; figure 4 graphicaly illustrates the comparison of results obtained from five different interactive simulations.
