The Use of Genetic Algorithms in Underground Mine Scheduling

The problem of production scheduling is sufficiently complex to negate the use of many conventional optimising techniques as there are numerous and diverse factors that must be considered when trying to produce an optimal schedule. These factors include: grade distribution, orebody geometry, production rate capacity, processing constraints and market price of the minerals amongst a variety of others. In addition it may be observed that many of these factors incorporate a degree of uncertainty and any schedule produced must inevitably consider risk as a primary consideration. The very nature of the problem has limited the use of conventional optimisers. Genetic algorithms use methods based upon evolution and natural selection to produce `optimal' solutions to complex system problems. They are extremely flexible and robust in their nature and it is these characteristics that are important in this particular problem domain. A particular strength of the genetic optimisation approach is the ability to handle the multiple and non-linear constraints that are typically encountered in mine scheduling problems. This paper concentrates on the development of a prototype application of a genetic algorithm to the problem of scheduling the output from multiple production points in an underground mining operation. In particular it.describes a system (GO-SCHED) that can be customised to handle a wide variety of underground scheduling problems. GO-SCHED, a Windows-based system, can interface with a variety of databases and spreadsheets for data input and output and provides a simple intuitive approach to the solution of the scheduling problem.