具有多性能特性的大规模多态系统可靠性高效评估和结构优化研究

Multi-state system (MSS) models have been widely applied for representing major equipment and technical systems and intensively studied in recent years. However, the huge number of possible states and components brings “dimension damnation” problem to the reliability evaluation for the large-scale MSS. Moreover, many complex systems like integrated energy systems have multiple performances, whose reliability cannot be accurately evaluated by a single performance MSS model. Furthermore, the coupling between multiple performances makes the system structure optimization much more complex. Therefore, the studies into the modeling of multi-performance MSS (MPMSS), the efficient reliability evaluation and structure optimization of large-scale MPMSS are of high importance to the system reliability theory.. The proposed project conducts studies on: (1) The modelling of MPMSS. The reliability model and the corresponding evaluation method for the MPMSS are proposed considering the influence mechanism of multiple performances on the MSS. (2) The efficient reliability evaluation of large-scale MPMSS. Facing the “dimension damnation” problem, three efficient reliability evaluation methods are proposed based on cluster analysis of system states. (3) The structure optimization of large-scale MPMSS. A novel method is developed to design a system structure with minimal cost and subjected to a desired reliability level. (4) The high-performance computing for the large-scale MPMSS. A three-level parallel computing framework is proposed for the reliability evaluation and structure optimization of large-scale MPMSSs, which includes problems, algorithms and platforms levels. The proposed studies have significant meaning for complex large-scale systems to achieve the economic operation with high reliability.

多状态系统模型能够细致地刻画现代重大装备和复杂工程系统的性能特征，是当前系统可靠性领域的重要研究方向。然而多态的引入和元件数量的攀升，使得大规模系统的可靠性评估面临维数灾的计算难题。同时，复杂多态系统如综合能源系统还可能呈现出多性能的特征。多种性能的耦合不仅使系统可靠性评估更加复杂，还使得满足多性能需求的系统结构优化愈发困难。 .项目通过研究具有多性能特征的多态系统与元件间的映射关系，构建多性能多态系统的可靠性模型与相应的可靠性评估方法；基于相似状态聚类，并通过算法比较整合，建立大规模多性能多态系统可靠性高效评估体系；开展综合考虑经济性与可靠性的多性能多态系统结构优化研究；提出基于问题并行，算法并行和平台并行的三层次可靠性并行计算架构，研究大规模多性能多态系统可靠性评估和结构优化的高性能计算方法，实现大规模复杂系统经济高效的规划目标，满足安全可靠的运行需求。

多性能多状态系统模型能够细致地刻画现代重大装备和复杂工程系统的特征，是当前复杂系统可靠性领域的重要研究方向。然而多状态和多性能参数的相互耦合极大地增加了系统可靠性研究的难度，不仅使大规模系统可靠性评估面临计算负担大的计算难题，还使得满足多性能需求的系统结构优化愈发困难。因此亟需开展具有多性能特性的大规模多态系统可靠性高效评估和结构优化研究。本项目取得的研究成果包括：1）提出了多性能多态串并联系统的基本数学定义和可靠性模型，量化了多性能参数对系统可靠性的影响，实现将传统多状态系统可靠性理论向多性能多态系统的拓展；2）针对大规模系统带来的计算负担过高的问题，提出了大规模多性能多态系统可靠性高效评估算法，在保证精度的前提条件下提高了可靠性评估的计算效率；3）考虑可靠性约束和最优经济性目标，建立了大规模多性能多态系统的结构优化模型，提出了结构优化高效求解算法，满足工程系统经济安全的规划需求；4）基于并行架构，提出了大规模多性能多态系统并行计算方法，将可靠性高效评估中的并行任务分给不同的进程，为适应大规模多性能多态系统的高性能计算提供技术支撑。.项目的实施使多性能多态系统可靠性理论得到了进一步的发展，尤其是大规模复杂系统的可靠性高效评估和结构优化研究。通过可靠性建模、高效评估算法、高效优化算法提供的技术支持，实现了多状态系统可靠性理论的创新和提升，保障了大规模多性能多态系统安全可靠、经济高效地运行。.本项目在《IISE Transactions》、《IEEE Transactions on Reliability》、《Reliability Engineering and System Safety》等国内外高水平期刊上发表论文23篇；授权国家发明专利6项；项目负责人受邀在国内外学术会议上作报告4次；项目组成员参加国际会议并作报告3次；邀请国际知名学者进行学术交流3人次；培养博、硕士研究生6名。