Development of an artificial neural network using parametric correlation technique for the determination of machined torsional spring stiffness

Creative Commons License


JOURNAL OF THE FACULTY OF ENGINEERING AND ARCHITECTURE OF GAZI UNIVERSITY, vol.36, no.1, pp.105-118, 2021 (SCI-Expanded) identifier identifier identifier


Development of an artificial neural network (ANN) for the determination of the spring constant of machined helical springs, which may be preferred over conventional springs due to their high performance and operating efficiency, is presented. Initially, finite element analyses were performed with various dimensional parameters and the obtained spring constant values were verified with the tests performed with the designed test setup. Parametric correlation analysis was performed using the confirmed finite element results and the effect of each spring dimensional parameter on the spring constant was determined. The parameters required for ANN training was determined according to the this correlation result. The spring constant results obtained from the developed ANN was compared with the finite element results confirmed by the tests and it was determined that the ANN was successful in the determination of the spring constant. The importance of parametric correlation analysis has been revealed in the development of ANN.