Methodology for Diagnosing the Causes of Die-Casting Defects, Based on Advanced Big Data Modelling

The purpose of this paper was to develop a methodology for diagnosing the causes of die-casting defects based on advanced modelling, to correctly diagnose and identify process parameters that have a significant impact on product defect generation, optimize the process parameters and rise the products’ quality, thereby improving the manufacturing process efficiency. The industrial data used for modelling came from foundry being a leading manufacturer of the high-pressure die-casting production process of aluminum cylinder blocks for the world's leading automotive brands. The paper presents some aspects related to data analytics in the era of Industry 4.0. and Smart Factory concepts. The methodology includes computation tools for advanced data analysis and modelling, such as ANOVA (analysis of variance), ANN (artificial neural networks) both applied on the Statistica platform, then gradient and evolutionary optimization methods applied in MS Excel program’s Solver add-in. The main features of the presented methodology are explained and presented in tables and illustrated with appropriate graphs. All opportunities and risks of implementing data-driven modelling systems in high-pressure die-casting processes have been considered.