Column Flotation Model Tuning Using A Genetic Algorithm

Researchers at the U.S. Bureau of Mines have successfully used a genetic algorithm to tune computer models of three column flotation units simulating a separation circuit. The computer models rely on empirical constants that are selected so that the models accurately predict the performance characteristics of individual column flotation units. Unfortunately, these empirical constants have traditionally been selected using a time consuming trial and error process. A new and more efficient approach to selecting empirical constants for computer models has been developed. A genetic algorithm has been used to select the empirical constants for column flotation models used to simulate a three column circuit. Genetic algorithms are search algorithms based on the mechanics of natural genetics. They rapidly locate near optimum solutions in difficult search spaces. Bureau researchers have used these innovative search algorithms to dramatically reduce the time needed to select empirical constants for column flotation models. The approach is applicable to the selection of empirical constants for numerous other computer models.