Artificial immune systems in solving routing problems

Abstract

Successful planning of electrical distribution networks is a complex problem. Besides the well-known radial network layout, in Europe, particularly in Croatia, the distribution networks are also loop or link structured. In order to be solved, network optimization problem is translated into some known combinatorial problems. When networks are loop-structured, the traveling salesman problem (TSP) is commonly used. The combinatorial problems like TSP cannot be solved exactly and evolutionary algorithms have been successful in solving these problems. Although, they are shown as very efficient, progress is still expected concerning the stability and lesser dependency on input parameters. In addition, evolutionary techniques for solving combinatorial problems often do not pay attention to existing knowledge about the problem. An improvement of the classic genetic algorithm used for solving the TSP is shown here, inspired by artificial immune systems techniques. Special attention is also paid to object-oriented design of the application and achieved benefits.