Heuristic approaches for a multi-mode resource availability cost problem in aircraft manufacturing
研究了飞机制造中带时间约束的多模式项目调度问题，设计了优先规则启发式和偏随机密钥遗传算法，实验表明遗传算法在复杂网络和短工期下表现更优。
• A time-constrained project scheduling problem in aircraft manufacturing is studied. • Generalized temporal constraints and a multi-mode situation are considered • A priority rule-based heuristic and a biased random-key genetic algorithm are designed. • Problem instances motivated by settings found at a large aircraft manufacturer are used. • Computational experiments demonstrate that the genetic algorithm performs well. A multi-mode time-constrained project scheduling problem with generalized temporal constraints arising in aircraft manufacturing is studied in the paper. We propose a priority rule-based heuristic (PRH) and a biased random-key genetic algorithm (BRKGA) for its solution. A serial generation scheme (SGS) is used for computing schedules from a priority order of the tasks with given resource capacities and mode assignments. The SGS cannot guarantee that the maximum project duration and maximum time lags are respected. Starting with the highest possible resource capacities, the PRH performs the SGS in a repeated manner, reducing the least used resource capacity by one unit until the schedule becomes infeasible. Different priority rules are used for determining both mode assignments and task priority orders. We encode these two decisions as well as the resource capacities in the BRKGA and apply the SGS for decoding. Project duration and maximum time lag violations are penalized in the fitness function. Extensive computational experiments based on problem instances motivated by settings found at a large aircraft manufacturer demonstrate that the BRKGA outperforms the PRH under almost all experimental conditions, especially for problem instances with more complex networks and shorter maximum project durations.