Using particle swarm optimization to determine visit times in community nurse timetabling

In the problem of community nurse timetabling where the visits have accompanying time windows, knowing the order of the visits is not enough, the actual times of the visits need to be known. This is normally solved using a simple greedy approach, but the solutions reached this way can exhibit serious flaws . We show how the determination of these times can be modelled as a convex optimization problem, and demonstrate how this can be solved using Particle Swarm Optimization. Initial results show that this approach is far superior to the greedy approach.