基于置信规则库最优决策结构的装备体系保障性评估方法研究

The equipment system of system (SoS) supportability is essential to the fulfillment of SoS mission and tasks. SoS supportability assessment result can help identify the supportability gap which could further be utilized for optimizing supportability resources. The supportability assessment result could also be used as the foundation to maintain the technical status as well as to make SoS utility plans. The SoS supportability assessment has shown strong uncertainty which is expressed in two characteristics, multi-source information and SoS behavior. So far, current approaches for SoS supportability assessment have shown disadvantages, namely, vulnerable to subjective influence, high cost, randomness, poor interpretability, and most importantly, unable to address the uncertainty characteristic. With this, the Belief Rule Base (BRB) is applied. Targeted on the uncertainty characteristic of the SoS supportability assessment, the BRB structure and parameter joint optimization is conducted based on the power set recognition framework, which produces the best decision structure for BRB. Based on the multi-source information, namely the high-dimensional variables, multi-source testing information and failure-time information, the supportability assessment approach based on the BRB best decision structure and multi-source information is proposed. Oriented from the SoS emergence and evolving behaviors, the supportability assessment approach based on the BRB best decision structure and SoS behaviors is proposed.

装备体系保障是关系到装备体系能否完成作战使命任务的关键问题，保障性评估结果有助于识别保障性水平差距，是合理配置装备体系保障资源的重要依据；对保持技术状态完好，提升综合效能具有重要的支撑作用。装备体系保障性评估具有强不确定性，主要体现在具有多源信息和体系行为特征。当前研究方法存在易受利益主体影响、成本昂贵、评估结果随机性强、可解释性差等不足，尤其是不能很好地建模装备体系保障性评估问题中的强不确定性。针对保障性评估中的强不确定性需求，拟采用置信规则库对装备体系进行保障性评估。在幂集辨识框架基础上开展置信规则库的结构与参数联合优化，获得置信规则库最优决策结构；针对保障性评估中需要处理多源信息(包括高维特征变量、多源测试信息、失效时间信息等)的特征，开展基于置信规则库和多源信息的保障性评估方法研究；针对保障性评估中需要处理体系涌现与演化特征，开展基于置信规则库和体系行为的保障性评估方法研究。

装备体系保障是关系到装备体系能否完成作战使命任务的关键问题，保障性评估结果有助于识别保障性水平差距，是合理配置装备体系保障资源的重要依据；对保持技术状态完好，提升综合效能具有重要的支撑作用。.本课题主要开展四个方面的主要内容：（1）体系保障性评估建模方面。基于体系结构框架理论（DoDAF）分别自上而下和自下而上进行分析，建立装备体系保障性的评估指标体系，并将提出的方面在某实例中进行应用。（2）面向体系保障性评估的置信规则库结构与参数联合联合优化。以赤池信息准则（AIC）作为综合优化准则，上层以AIC作为优化目标，以规则数量作为决策变量，下层以建模精度作为优化目标，分别提出并集与交集假设下的BRB结构与参数联合优化方法。（3）面向多源信息的体系保障性评估。为应对作战体系保障能力评估中存在的异构、缺失信息和少数据的特征，集中提出了相应的BRB传播方法。提出了面向多源信息与专家知识的保障性评估方法。该方法中还采用了多算子推荐策略，在优化过程中自动选择最佳算子，以达到最佳优化效果。（4）向体系行为的体系保障性评估。分别进行了考虑相关性和冗余性的评估方法、基于聚类的体系保障性评估方法、面向体系动态特征的体系评估方法。.在本青年科学基金项目的资助下，以置信规则库（Belief Rule Base，BRB）为主要建模手段，已经完成课题申请阶段设想的全部研究内容，项目周期内合计发表SCI期刊论文25篇，其中第一作者8篇（包括《IEEE Transactions on Fuzzy Systems》、《IEEE Transactions on Systems, Man, Cybernetics: Systems》、《Information Sciences》、《Knowledge-Based System》、《International Journal of Approximate Reasoning》、《IEEE Journal of Biomedical and Health Informatics》等），通讯作者3篇，合作论文14篇，EI期刊论文5篇；同时在本课题的支持下，综合以上研究内容，以第一作者于科学出版社出版学术专著一部：《装备技术体系设计与评估》；以第一作者于国防工业出版社出版译著：《大数据与智能计算》。