Generating multiobjective trade-offs: an algorithm for bicriterion problems

A technique for generating an approximate or an exact representation of the noninferior set for problems with two objectives is presented. The approach proceeds by using the slope of the line segment between adjacent, previously generated noninferior extreme points to find new noninferior solutions to improve the approximation or to establish the noninferiority of the line segment when the entire noninferior set is to be generated. The proposed technique compares favorably with other generating techniques when the analytical goal is to approximate the noninferior set. A major innovation is the computation of the maximum possible error, which the analyst may control, to obtain an approximation of a desired degree of accuracy. A small example and a river basin planning problem of moderate size are used to demonstrate the use of the algorithm. Further research is required to extend the algorithm to problems with more than two objectives.