Application of genetic algorithm for resource-constrained scheduling

Goal of Scheduling problem is optimization of a function in deciding a set of activities that must be executed under different constraints. There are two main types of constraints including are precedences between activities, and the availability of finite resources. This research presents a genetic algorithm approach to resource-constrained scheduling using a direct, time-based representation. The new representation encodes schedule information as a dual array of relative delay times and integer execution modes. This representation includes time-varying resource availabilities and requirements. The genetic algorithm adapts to dynamic factors such as changes to the project plan or disturbances in the schedule execution. The genetic algorithm was applied to over 1000 small job shop and project scheduling problems (10–300 activities, 3–10 resource types). According to result based on computationally expensive, the algorithm performed fairly well on a wide variety of problems. In addition, the algorithm found solutions within 2% of published best in 60% of the project scheduling problems. The GA performed better than deterministic, bounded enumerative search methods for 10% of the 538 problems tested on project scheduling problems with multiple execution modes