The application of genetic algorithms to optimisation problems in geotechnics

Abstract

The paper presents an introduction to a relatively new optimisation technique, known as genetic algorithms, and discusses its potential for application to geotechnical problems. The method of genetic algorithms is a search technique based on the mechanics of natural selection and natural genetics implemented by coding each state of a particular optimisation problem as a string of binary digits. The objective function provides a measure of the ‘fitness’ of each state. Further generations of the binary string are created by a process of reproduction, crossover and mutation that favours the survival of the fitter strings. An optimal, or near optimal solution is identified after a relatively small number of generations that represent only a small fraction of the complete set of possible enumerations. After a brief explanation of the principal elements of genetic algorithms the paper outlines the background theory to the identification of discontinuity frequency extrema in fractured rock masses — an optimisation problem that is computationally demanding. The technique is then implemented through a case example involving a rock structure containing up to 100 discontinuities, each one treated as a set. It is shown that genetic algorithms provide an efficient and computationally powerful optimisation technique.