Artificial Neural Networks for Modeling Mechanical and Microstructural Properties of Spark Plasma Sintered Powders

This paper presents a novel approach to modeling and optimizing the mechanical and microstructural properties of 316L stainless steel surfaces manufactured by spark plasma sintering (SPS). The integration of artificial intelligence techniques, particularly artificial neural networks (ANNs), with the optimization of material properties and the spark plasma sintering (SPS) process reflects the growing emphasis on intelligent manufacturing in advanced industrial applications. The surface functionality depends on the material's mechanical and microstructural characteristics. The optimization technique was developed through the processing of a comprehensive set of measurement data, forming the foundation for the artificial intelligence method. To model the relationships between SPS parameters (sintering temperature and holding time) and material properties (density, porosity, hardness, and surface-affected zone (AKA the possible carbide zone depth), a series of controlled experiments was conducted. The performance of neural network models was evaluated using their coefficients of determination (R² > 0.95) and the sum of squared errors (SSE < 0.02). These metrics were calculated by comparing actual measurement data with values predicted by the models. Validation experiments confirmed the reliability of the presented models and their relevance for implementation in industrial environments. The predictive model is valid for 316L stainless steel within the tested SPS setup and parameter range.