An urban water distribution network (WDN) is a network of components (e.g. pipes, pumps, valves, tanks, etc.) that transport water from a source to the consumers. Due to the substantial cost associated with the material and installation of WDN, it is necessary to optimize its design by selecting the lowest cost combination of appropriate component configration while the hydraulic and resilience constraints are satisfied. Thus far, a large variety of algorithms have been proposed for this optimization problem, among which swarm intelligence algorithms (SIA) attract the most recent attentions. In the project, several new SIAs are tested on this problem for the first time and different Machine Learning techniques are also used to further improve the performance of these swarm intelligence search algorithms. Ten different algorithms are proposed in this thesis project for WDN optimization problem. All of the proposed algorithms are tested on two famous benchmark networks and their performances are compared extensively, the results show that some of the proposed algorithms are very promising in the real application, especially for large size water distribution networks. What is more, one of the proposed algorithms successfully achieves a new record of the best solution cost on the larger size network.