A dual-adaptive directed genetic algorithm for construction scheduling

Abstract

The precise and efficient generation of construction sequences is a crucial concern within the engineering field. However, the formulation of construction schedules heavily depends on the expertise and proficiency of project planners, leading to significant potential for inefficiencies and instability in project management. To deal with this challenge, this study automatically extracts constructability constraints from 3D models and proposes a dual-adaptive directed genetic algorithm (DADGA) to generate a structurally stable installation sequence. The proposed algorithm adaptively changes both the crossover and mutation probabilities based on the quality of individuals and evolutionary stages. In addition, the idea of directionality and the chief strategy artificially controls the direction of evolution, which greatly improves the efficiency and robustness of local search. The results of comparison experiments demonstrate that the DADGA outperforms the traditional genetic algorithm in terms of both efficiency and accuracy, and a practical example is also presented to showcase the capability of the DADGA in solving ultra-complicated construction scheduling problems.