基于无尺度复杂网络建模的单元制造系统调度及优化

In order to support advanced manufacturing modes, the complex networked modeling and optimization scheduling on resource nodes of cell manufacturing systems in networked environments are studied. A scale-free complex networked model of resource nodes in the cell manufacturing system is proposed from the perspective of complex network theory and its dynamics parameters are yielded. Global optimization scheduling methods for multi-node, large-dimensional, multi-objective and constraint in networked environments are proposed. Specifically, a scale-free complex networked model of cell manufacturing resource nodes is proposed and its complex networked features and dynamics parameters are analyzed using methods of statistical physics. The invulnerability of the proposed model is tested. Then a mathematical model for the multi-node, large-dimensional, multi-objective and constraint scheduling problems is presented and the non-linear and non-convexity properties of the proposed model are transformed. A ε-dominance evolutionary algorithm with adaptive ε parameter adjusting strategies for large-dimensional, multi-objective and constraint scheduling problems is designed. Combining with scheduling rules for the scale-free complex network model, an optimization scheduling strategy with reasonable population size, searching speed and global optimization of solutions for the multi-node, large-dimensional, multi-objective and constraint scheduling problems is presented. Finally, TOC, polychromatic set theory and LCODE model are adopted to improve the execution chain of the cell manufacturing system. This application provides innovative views and approaches for complex networked modeling and scheduling of cell manufacturing system and can be further extended to optimize manufacturing resource scheduling problems. Moreover, it is also a typical application case of complex network theories.

为适应先进制造模式，对网络环境下单元制造系统资源节点复杂网络建模及调度优化进行研究，从复杂网络视角建立单元制造资源节点无尺度模型并获取动力学参量，寻求网络环境下多资源节点、高维目标和多约束的全局优化调度方法。首先构建单元制造系统资源节点无尺度复杂网络模型，采用基于统计物理方法分析其复杂网络特性、获取动力学参量，并对其抗毁性进行分析测试；建立资源节点调度优化高维目标数学模型并进行线性及凸化转换；设计自适应ε调整的ε-支配高维目标进化策略；结合无尺度复杂网络模型的调度规则，寻求兼顾求解规模、求解速度及全局优化性能的高维目标调度优化方法，采用约束理论、多色集合理论和线性耦合常微分方程一步协调与改善单元制造系统执行链。所采用的研究方法和手段为解决网络环境下单元制造系统建模与调度优化提供新的角度和方法，可进一步深化并扩展制造系统资源调度优化的建模与求解方法，也为复杂网络相关理论和方法提供应用范例。

为适应信息化时代的制造模式，针对网络环境下大规模单元制造系统的调度优化需求，从复杂网络视角建立单元制造资源节点复杂网络模型，研究网络环境下多资源节点高维目标和大规模调度优化方法。研究内容包括：1）建立了单元制造系统资源节点无尺度复杂网络模型；采用统计物理方法分析了复杂网络特性、获取了动力学参量，对其抗毁性进行分析测试。2）针对网络环境下单元制造系统资源节点大规模调度优化问题，建立复杂网络模型，提取与调度优化相关的复杂网络特征，提出一种基于模块度的改进小世界遗传算法和一种基于瓶颈度导向局部搜索的改进布谷鸟搜索算法； 针对网络环境下跨单元调度高维目标优化问题，采用主成分分析法对高维目标优化问题进行降维处理，设计了改进精英策略的PCA-NSGAⅡ算法，解决了网络环境下跨单元制造中高维目标与大规模调度优化问题。（3）针对网络环境下单元制造系统中的不确定因素，采用主动调度策略，提出了基于博弈的前摄性鲁棒调度方法；遵照复杂网络构建规则构建复杂调度网络模型，依据复杂网络结构特征与任务属性制定调度规则，这些规则可用于改进算法和训练神经网络；为了快速得到调度方案，提出一种基于复杂网络特征的双重神经网络调度器；（4）对复杂网络故障节点、复杂网络定位及故障免疫问题进行研究，建立了虚拟单元制造系统节点网络中基于SIRS的误差传播模型，通过数值模拟对无尺度复杂网络误差传播规律进行了仿真分析；为了优化机械产品研发过程中变更传播路径问题，基于复杂网络理论，提出一种结合复杂网络变化概率(CP)和变化影响(CI)的改进Dijkstra算法对变更传播路径进行优化，以降低产品设计过程总成本。基于上述研究开发了一个跨单元制造系统复杂网络模型与调度优化平台。研究方法与成果为解决网络环境下单元制造系统建模与调度优化提供了新的角度和方法，研究成果丰富了复杂网络理论和调度优化理论与算法，为促进先进制造模式的应用提供了范例。