基于Steiner树的复杂电缆网布线的多目标粒子群优化方法研究

Cable harness configuration and routing planning (CHCRP) is one of the mostcomplicated and time-consuming work in the development of complex mechanical and electrical products, so its automated layout routing and optimization will improve the design efficiency and quality of product effectively. However, there are few applications of the automated routing technology in engineering owing to the problems of the present researches such as little consideration of cable harness routing process, difficult to get the global optimal configuration and path, and lack of multi-objective optimization method for cable harness routing. For the defense development requirement, this subject will research the multi-objective particle swarm optimization method for cable harness layout routing based on Steiner tree. A novel multi-objective optimization problem (MOP) model of CHCRP with engineering rules constraint is to be presented. With the import of Pareto optimal solution concept, to design the multiple objective particle swarm optimization (MOPSO) for CHCRP by research the key technology such as the modified particle coding, updating, best solutions selecting, and etc. The optimization mechanism of the presented MOPSO for CHCRP problem is to be revealed by the evolutionary status detection technology of the population to be researched, and the study will obtain high performance MOPSO for CHCRP by the improve strategies. At last the validation of research mentioned above will be verified according to some practical cable harness routing problem cases. The subject supplies novel research thinking and theoretical support for the routing scheme optimization and assembly performance analysis for cable harness, which has important practical significance and values for realizing the rapid and low cost product development.

电缆网的布线设计是复杂机电产品研发中最复杂、最耗时的部分之一，其自动布线及布局设计优化可有效提高产品设计效率和质量。然而目前的研究存在布线工艺考虑不足，难以得到全局最优的结构和路径，以及缺少布线多目标优化方法的问题，制约了自动布线技术的工程应用。本课题结合国防研制需求，拟开展基于Steiner树的复杂电缆网布线的多目标粒子群优化方法研究。提出新的工程规则约束的电缆网布线多目标优化问题模型；引入Pareto最优解概念，基于粒子群算法设计电缆网布线的多目标优化算法，研究新的粒子编码、运动策略和最优解选择策略等关键技术；研究算法进化状态的检测技术，揭示基于粒子群算法的复杂电缆网布线多目标优化机制，并研究算法性能的改进技术，得到高性能的布线多目标粒子群优化算法，并结合布线问题实例进行算法的有效性验证。本课题的研究将为电缆网自动布线的优化设计提供新的研究思路和理论支持，具有重要的理论价值和工程意义。

复杂机电产品中电缆网布局的优化设计需要同时满足结构、工艺、电气和成本等多方面的设计需求，是极具挑战性的难题。针对目前研究中该优化问题的描述缺乏实际工程设计准则、优化求解质量和效率不足的问题，本项目系统研究了基于Steiner树的复杂电缆网布线的多目标粒子群优化方法。研究了机电产品布线离散空间模型的建模和预处理技术，基于Steiner树构建了电缆网布线的信息模型，建立了具有共性工程规则约束的电缆网布线多目标优化问题的数学模型；研究了电缆网MBD模型的参数化自动建模技术，并设计开发了基于Creo和CATIA等典型商用CAD平台的自动建模接口；针对电缆网布局多目标优化问题中离散/非解析/无梯度/NP-hard的难求解特性，提出了基于MOPSO/D的电缆网布局多目标优化方法，其中通过对种群在优化中的运动监测，分析获得了算法操作对解集收敛性与分布性的影响，揭示了粒子群算法在电缆网布线问题中的多目标优化机制；设计开发了电缆网布局自动优化原型系统，并针对实际布线问题实例对算法的有效性和性能进行了验证。本项目研究成果为电缆网的布线自动设计技术提供了新的理论和方法支撑，也有望为航空航天等领域的复杂机电产品的研制提供必要的技术支持和保障，具有重要的国防应用价值。