基于Transformer大模型的船体曲面分段车间排产与调度优化

Optimization of Production and Scheduling for Ship Surface Segmentation Workshop Based on Transformer Large Model

摘要:

提出基于Transformer大模型的船体曲面分段车间排产与调度优化方法。设计资源冲突消解策略处理车间资源冲突问题，合理设置工件移动路径并进行工序分割，然后将船体曲面分段建造资源冲突消解策略、工序分割结果以及当前分段建造任务数量、工作班组数量等作为输入，利用Transformer大模型，以总完工时间最小为目标构建排产与调度目标函数，并将资源均衡性作为重要考量指标，输出排产与调度优化结果。实验表明:该方法能科学管控任务和多班组协作，有效提升资源利用率与执行效率；资源均衡性表现优异，10种设备资源均衡度均在0.7以上；造任务的实际耗时均低于预计时间，能够显著缩短任务完成时间，为船舶制造行业提供了创新有效的生产优化途径。

Abstract:

A production and scheduling optimization method for ship surface segmentation workshop based on Transformer big model was proposed in this paper . A resource conflict resolution strategy is designed to address workshop resource conflicts, reasonable workpiece movement paths are set, and process segmentation is performed. Subsequently, the ship surface segmentation construction resource conflict resolution strategy, process segmentation results, and the current number of segmented construction tasks and work teams are used as inputs. The Transformer architecture in artificial intelligence models is utilized to construct scheduling objective functions with the goal of minimizing total completion time, while resource balance is considered as an important indicator, with scheduling and optimization results being outputted. The experiment shows that this method can scientifically control tasks and facilitate multi team collaboration, effectively improving resource utilization and execution efficiency; Excellent resource balance performance, with resource balance of all 10 devices above 0.7; The actual time consumption of the construction task is lower than the expected time, which can significantly shorten the completion time of the task and provide an innovative and effective production optimization approach for the shipbuilding industry.