Bayesian Statistics Applied to the Recovery of Copper and Cobalt-Bearing Ores: Establishment of the Probability Laws of Leaching Parameters

Currently, industries and researchers recognize the need for new ways to incorporate prior knowledge into processes. With the increasing accessibility of machine learning techniques through software packages, data-driven modeling applications are becoming more prevalent, incorporating more advanced techniques and analytics. Bayesian methods utilize advanced concepts of conditional probability for their inferences. This is crucial for developing robust and understandable solutions that can be used for prediction, optimization, or decision-making. In this paper, we discuss the implementation of Bayesian modeling in hydrometallurgy, focusing specifically on its application in the leaching process. Bayesian statistics provides the opportunity to incorporate prior knowledge into a model, making it a valuable approach in this field. Therefore, parameters that influence leaching, such as mineralogy, particle size, redox potential (pH), acidity, and oxido-reduction potential (ORP), were included as random variables in the model. This enables the designing of Bayesian networks with precisely defined a priori probability laws.