Precedence constraints treatment in ant colony optimization

When Ant System algorithm is used to solve combinatorial optimization problems with precedence constraints, the computation of components list that can be used to extend the current partial solution is time consuming. For every component, the algorithm tests if the precedence constraints are met. In this work, a treatment list of components is proposed, aiming to avoid the test of precedence constraints for every component and for every ant. The list is not a fixed sequence of components, but it is equivalent to a total order between predefined groups of components. Search biases can appear and some solutions are proposed. The use of this technique is exemplified by an Ant Colony Optimization algorithm for the Tasks to Workstations Assignment(TWA) problem. The effectiveness of the proposed approach is validated by computational tests.