Abstract

The heap bioleaching process at Minera Escondida Ltda. (MEL) has been operating since 2006 and during this time detailed monitoring has been performed. The development of the monitoring program has been continuously updated based on the state of the art including microbiology and molecular biology approaches. Up to now, we can say that this is a unique set of data at the world level and is an extremely valuable asset to raise knowledge about complex bioleaching processes. The huge industrial data recorded by several years represents an opportunity for technology improvement. A systematic approach using machine learning tools to perform supervised and unsupervised approaches useful for the analysis of High Dimensional Feature Space was developed for turning data into insight and to deliver experience-based learning with the aim to serve as the foundation for optimal industrial decision making even in presence of inherent process variations. The development of this Decision Support System considers a database for data logging and storage, a knowledge base for collecting the set of decision rules which combine the insights obtained from the data and the expert knowledge, the creation of a system for managing knowledgeable to reproduce the expert reasoning and finally, a friendly interface for interacting with the user able to receive the questions/queries and to transform expert reasoning into recommendations/information. The user can accurately retrieve data and design similar matches to the historic operation to get e.g. the expected metallurgical performance of a strip (leaching unit) based on its mineralogical/mineralization characteristics. The development of smart mining requires the understanding to reduce future uncertainty taking advantage of the information held within the data and creating capacities for knowledge to be shared. The next step is the transference to the end-user for them to adopt the technology in order to generate innovation and improving productivity.