Multi-mine Planning using a Multi-objective Evolutionary Algorithm

We describe a planning system for multi-mine scheduling that works by iteratively interrogating a single-mine planner for each individual mine-site. At the heart of the system is a multi-objective evolutionary algorithm that runs in every iteration to derive a set of requests to present to the single-mine planners. These requests are optimised to build up information about the individual mines: ideally they should be likely to be accepted, and they should contribute significantly to the process when they are accepted, although these two goals are often in conflict. We describe the structure and operation of the evolutionary algorithm in detail and we illustrate the behaviour of the planning system through two case studies.