Neighbor Evaluation Procedures in Local Search based Algorithms for solving the JSPTWT

Abstract

Local search algorithms perform iteration processes in which a neighborhood search is repetitively executed. The computational effort of the neighborhood search should be as low as possible and as high as needed in order to make continuously improvements over a long period of time. The generation and assessment of neighboring solutions are major steps in this process and consume most of the computation time. The generation of new schedules depends on the complexity of the perturbation scheme of the used neighborhood operator (see Chapter 4). The assessment of the generated solutions, i. e. the determination of the resulting total weighted tardiness value, can be performed exactly or approximately.