面向多移动智能物流资源调度的动态渐进群集智能优化方法研究

With the application of industrial IoT, the physical objectives during production process, such as materials and logistics equipment, enhances the information and automation level, and then become mobile smart logistics resources. These resources improve the flexible and intelligent level of production, and also can be utilized to enhance the robustness of industrial IoT. However, no studies have yet been made for the multiple mobile intelligence logistics resource scheduling which considers the logistics task execution and the network robustness enhancement simultaneously. With the complex constraints, e.g., dynamic assignment, procedure, layout, network connectivity, it is difficult to be solved, even for modern intelligence optimization algorithms, which have to face difficulties in constructing unfeasible solution, low search efficiency, etc. This project aims to build the scheduling model for multiple mobile smart logistics resources by considering the dynamic task and network robustness constraints using MOP theory and a dynamic incremental swarm intelligence optimization method. This project investigates the topology and robustness features of industrial IoT by using complex network theory, then establishes a scheduling model for multiple mobile smart logistics resources considering the network robustness and dynamic task constraints.; By analyzing the coupling relationship among individual, cluster and network in a swarm, the project builds an adaptive swarm intelligence optimization (SIO) algorithm model with multi-scale incentives as an optimization engine for the above schedule model.; This project also establishes an incremental optimization framework with integration of the multi-objective model and the SIO model, and design a dynamic incremental optimization mode for scheduling of mobile smart logistics resources to guarantee the stability of production organization. This study provides new view for scheduling mobile smart logistics resources under the new manufacture mode and SIO mechanism research.

工业物联网提升了生产中物料、物流设备等物理对象网络化、智能化水平，形成了移动智能物流资源。这些智能资源，在提升生产柔性和智能化水平的同时，也具备提升工业物联网鲁棒性的能力。目前，同时考虑物流任务和网络鲁棒性多目标优化的多移动智能物流资源调度问题，尚未有研究。因其调度将受动态任务、工序、布局及网络等多因素制约，即使采用智能算法也面临可行解难构建、优化效率低等难题。因此，本项目针对前述调度目标，运用多目标规划理论构建调度模型，并研究具有渐进式群集智能方法。项目将运用复杂网络研究工业物联网拓扑鲁棒性特征，构建具有作业任务和网络鲁棒性约束的多移动物流资源调度模型；通过对群集智能多层面间耦合关系分析，构建多尺度激励的自适应群集优化模型，作为优化引擎；提出渐进优化思想，构建针对移动物流资源调度的动态渐进式群集智能优化算法。本研究可望为新制造模式下智能移动物流资源调度及群集智能算法研究提供新思路。

群智能优化研究、创新及在资源调度优化中的应用是优化理论研究和应用的热点。目前对于群体搜索机理还缺少深入理解，对群智能算法应用还有待丰富。本项目围绕群智能算法机制研究及复杂物流资源调度优化开展，项目根据研究目标深入研究了群智能优化算法机理，提出了多尺度群智能搜索算法构建及改进思路，分析了不同尺度因素的相互影响，以此为基础又研究和提出了具有拓扑时变和搜索扰动的混合粒子群算法、融合局部学习和概率性选择变异的动态优化算法及具有解空间动态缩减的优化策略；结合生产环境下的移动物流资源调度问题，研究了具有任务约束的智能物流资源调度模型，提出多任务分配的AGVs调度模型及优化策略，并结合智能制造环境下任务的突发性特点，构建了考虑应急任务扰动的AGVs调度模型及优化方案，最后结合多类物流资源调度问题进行了群智能算法应用设计研究，解决了冷链运输车辆路径优化问题及协同库存的集合覆盖选址-分配方案优化设计。项目研究成果不仅有助于丰富和深化群智能优化理论及物流资源调度技术，而且将促进群智能优化调度算法的应用和推广，具有良好的理论和实际价值。