船体分段组立工程计划智能排产建模与优化方法研究

为解决船厂在分段组立计划排产环节存在智能化程度低、排产方案不合理导致资源浪费，无法实现精益排产的难题。针对分段组立工程准时排产问题，基于极大代数法理论，提出一种多资源约束的“拉式——准时”调度新模型。构建以所有组立总空闲时间最小化为目标的组立生产“连续性”评价函数；针对分段组立工程多资源均衡问题，以对其涉及核心资源的均衡利用为“均衡性”排产准则，采用最大熵原理及数据挖掘方法，建立组立工程各生产线的资源熵数学模型；最后，构建“连续——均衡”分段组立模式的多目标函数，以严格满足交货期为约束，设计基于混沌与差分进化的NSGA-II算法对该问题求帕累托解集，引入基于信噪比的机制，求解满足决策者偏好的最稳健计划，以实现智能排产。项目成果可为智能船厂的排产问题提供有力的理论基础，具有重要的理论意义与工程应用价值。

Lean shipbuilding is the cornerstone of ship intelligent manufacturing. Workshop-level lean scheduling is one of the important research areas of smart shipyard. In order to solve the problem that a low degree of intelligence and an unreasonable scheduling of ship block assembly lead to waste of resources in the shipyard, it is difficult to achieve lean shipbuilding. Aiming at the problem of on-time production scheduling for ship block assembly process, a new resource-constrained model named "pull and on-time" scheduling is proposed based on max-plus algebraic method, and an function is estimated to evaluate continuous production with the goal of minimizing the total idle time of all blocks/sub-blocks/pre-subblocks. In view of the multi-resource equilibrium problem of ship block assembly process, taking the balanced utilization of core resources as the "equilibrium" scheduling criterion, the principle of maximum entropy and data mining methods are used to establish a resource entropy mathematical model for each production line. Finally, the multi-objective function of the "continuous-equilibrium" mode for ship block assembly is constructed, with strict compliance with the delivery deadline as the constraint, and the NSGA-II algorithm based on chaos and differential evolution is designed to find the Pareto solution set for this problem. And the most robust plan that meets the preferences of decision makers is solved based on the signal-to-noise ratio mechanism to achieve intelligent scheduling. This project achievement will provide a strong theoretical basis for the scheduling problem of smart shipyards, and it has important theoretical significance and engineering application value.