Least median squares curve fitting using a genetic algorithm

Least median squares (LMS) curve fitting is a method of robust statistics hat guards the process of data analysis from perturbations due to the presence of outliers'. This procedure has several advantages over classic least squares (LS) curve fitting, especially in the noisy problem environments addressed by today's process-control engineers. Although LMS curve fitting is a powerful technique, there are some limitations to the LMS approach. However, these limitations can be overcome by combining the search capabilities of a genetic algorithm with the curve fitting capabilities of the LMS method. Genetic algorithms are search techniques that model the search that occurs in nature via genetics. This paper presents a procedure for utilizing genetic algorithms in an LMS approach to curve fitting. Several examples are provided from a number of application areas, thereby demonstrating the versatility of the genetic-algorithm-based LMS approach.