Project Scheduling by Evolutionary Algorithm Using a Niche and Cooperation Technique
Iwao Okutani, Koji Jinde, Makoto Tsue · Journal of Construction Management JSCE · 1999
In this study project scheduling problem is solved by an evolutionary algorithm improved by niche and cooperation method. In the algorithm an individual is defined as an scheduling pattern. At the niche stage children individuals are generated by a simple transformation using tangent function from parents individuals and a superior individual is selected in order to produce parents in the next generation from the given pair individuals which are very close to each other with respect to distance measured by starting time of each job in the project. At the cooperation stage the selected individuals at the niche stage execute crossover operation repeatedly to give birth to the sole better individual. From the testing calculation using five scheduling networks it is found out that the proposed algorithm outperforms the tabu seach algorithm compared for three networks based on the maximum fitness which is composed of project finishing time, maximum manpower required over the whole project duration and smoothness of manpower expressed by its variance.