Overcoming data gathering errors for the prediction of mechanical properties on high precision foundries

Mechanical properties are the attributes of a metal to withstand several loads and tensions. More accurately, ultimate tensile strength (UTS) is the force a material can resist until it breaks. The only way to examine this feature is the use of destructive inspections that render the casting invalid with the subsequent cost increment. In our previous researches we showed that the foundry process can be modelled as an expert knowledge cloud to anticipate the value of the UTS with outstanding results. Nevertheless, the data gathering phase for the training of machine learning classifiers is performed in a manual manner. In this paper, we present the use of Singular Value Decomposition (SVD) and Latent Semantic Analysis (LSA) with the aim of reducing the number of ambiguities and noise in the dataset. Furthermore, we have tested this approach comparing the results without this pre-processing step in order to illustrate the effectiveness of the proposed method.