Page 177 - Data Science Algorithms in a Week

The Estimation of Cutting Forces in the Turning of Inconel 718 Assisted … 161

In order to achieve the best performance of GA-based ANN a parametric study for
determination of optimal set of GA parameters was carried out. The optimization process
takes place with the values of crossover probability, mutation probability, maximum
number of generations and population size. A parametric study is carried out so that the
value of one parameter is varied at a time, while other parameters have fixed values. The
fitness value of a GA is estimated based on the mean absolute percentage error of each
training data sample, representing deviation of the result (cutting force components) of
the GA-based ANN from that of the desired one. It is noted that the error could be either
positive or negative. Thus, absolute value of the error is considered as a fitness value of
GA solution. For main cutting force the optimal values of crossover probability, mutation
probability, number of generations, and population size were 0.15, 0.025, 2260, and 520,
respectively. For feed force optimal values of these parameters were 0.9, 0.01, 1480, and
590. Finally, for passive force the optimal values of crossover probability, mutation
probability, number of generations, and population size were 0.1, 0.015, 1920, and 260,
respectively. The results of the parametric study for main cutting force are shown in
Figure 4.

Figure 4. Results of parametric study for determination of optimal set of GA parameters.