Investigation by Means of Artificial Neural Networks on the Influence of the Shot Peening Parameters on the Hardness of Seamless Tubes Manufactured in TX304HB Stainless Steel

Author:  Diego Ferreño, Ruth González, Isidro A. Carrascal, Diego García, Rubén Eraña, Federico Gutiérrez-Solana
Source:  ICSP-13
Doc ID:  2017101
Year of Publication:  2017
Abstract:  
Introduction: Shot peening (SP) is a widely used cold working process. A number of analytical solutions were developed in the past to understand the relation between the input parameters and the final outcome;unfortunately, these models were necessarily oversimplifications of the complex reality behind a real SP providing only an approximate description of the process, since they were severely restricted by the underlying hypotheses. In contrast, the evolution of the Finite Element (FE) method in the last 20 years, linked to the development in computational capabilities, has provided accurate and more general solutions. Nevertheless, in many instances FE is still too rigid for the needs of the industry during an actual industrial SP process. Artificial intelligence (AI) methods have been used as an alternative way to deal efficiently with complex and ill-defined problems in very diverse fields of engineering as in (Maleki and Sherafatnia,2016) and (Kalogirou, 2003). Among their advantages it is worth mentioning that AI methods are able to deal with noisy and incomplete data, with nonlinear problems, and that, once they have been trained, they can perform predictions and generalizations at high speed. An artificial Neural Network (NN) represents a computational approach to solve problems imitating the human brain. In this sense, a NN consist of a number of simple processing units called neurons (or neural units) arranged in layers connecting the inputs to the outputs. An important necessary (but not sufficient) condition for the reliability of a NN as a predictive tool is that the data should be representative of the complete input–output space. This method has been uccessfully employed in many engineering applications. For the reader interested, (Iliadis and Jayne, 2013) gathers the contributions to the Engineering Applications of Neural Networks conference, showing a large number of examples proving how NN provide practical solutions in a wide range of applications.