Keywords: evolutionary computing, genetic algorithm, lifeline restoration, optimization, uncertainty.

The main purpose of this research is the early restoration of lifeline systems after the earthquake disasters. Here, two issues are focused on, the first of which is an allocation problem that determines which groups will restore which disaster places, and the second is a scheduling problem concerning what order is the best for the restoration. In order to solve the two problems simultaneously, a genetic algorithm (GA) is applied, because these algorithms have been proven to be very powerful in solving combinatorial problems. However, the actual restoring process should be performed under uncertain disaster circumstances due to the lack of information, secondary disasters, and other unpredicted situations. In this study, an attempt is made to develop an efficient disaster restoration method by using a GA considering uncertainty (GACU) algorithm.

When a restoration team arrives at the site, the disaster circumstances may be different from those predicted, because devastated situations are constantly changing by the aftershock, fire disaster and bad weather, which are likely to make the damage worse. Such a change in the devastated area affects the scheduling process, because it takes days more than scheduled, and furthermore it may be impossible to restore if the restoration teams do not have enough ability. Therefore, in this study the amount of damage will be treated as an uncertain factor and the restoration scheduling problem is formulated as one of an optimization problem under uncertainties.

In the formulation, the necessary restoring days are considered to be an objective function to be minimized and the restoring ability for each team is considered in the constraint conditions. For such a combinational optimization problem, the GA has been proven to be very powerful, because it can provide good solutions with simple formulation and programming. It is, of course, necessary to improve the GA to solve large or complicated optimization problems by reducing the large or complicated searching space. Moreover, more improvement is needed to solve the problem treated here. This is due to the complexity of the problem which includes various uncertainties and the large number of variables to be determined. In order to deal with the uncertainties, a new GA method was developed by combining the GA computation and a Monte Carlo simulation. This method reduces considerably the computation time and improves the convergence.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £62 +P&P)