基于不确定性计算的鲁棒风险评估关键技术研究

项目名称： 基于不确定性计算的鲁棒风险评估关键技术研究

项目编号： No.61502068

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 其他

项目作者： 曲衍鹏

作者单位： 大连海事大学

项目金额： 21万元

中文摘要： 大规模风险信息中存在的不确定性因素会导致不可靠、不合理的评估结论。针对风险信息自身缺陷对评估结论所造成的多种影响，本项目创新性的通过研究模糊决策系统和模糊粗糙理论在信息处理和决策方面的作用，在统一的框架下实现了两类鲁棒风险评估方法。第一类是由模糊粗糙不确定信息引导的基于快速模糊决策系统的风险评估方法。这一方法通过降维数据、特征不确定信息以及快速模糊决策系统，来解决模糊决策系统在风险评估过程中的效率鲁棒性的科学问题。第二类是采用模糊决策系统特征评价机制优化的模糊粗糙风险评估方法。这一研究通过基于快速模糊决策系统的特性评价机制、抗噪模糊粗糙近邻算法以及灵活不确定性分类算法，提高了相关风险评估方法的抗噪能力以及综合处理多种不确定性因素的能力，解决了风险评估方法的抗噪鲁棒性和灵活鲁棒性的科学问题。本项目进一步的拓宽了不确定性计算技术的应用范围，具有重要的理论与实际应用价值。

中文关键词： 糢糊决策系统；模糊粗糙集；特征选择；特征加权；风险评估

英文摘要： In the large-scale risk information, the existence of the uncertainty factors can result in certain unreliable and irrational assessment conclusions. Aiming at the multiple impacts caused by the self-defects of the risk information, this project inventively implements two robust risk assessment (RA) approaches under a uniform framework, through the investigation of the roles played by the fuzzy decision system and fuzzy-rough theory in information processing and decision. The first part is the fast fuzzy decision system-based RA method guided by the fuzzy rough uncertainty information. This technology employs the reduced datasets, the features’ uncertainty information and the fast fuzzy decision system, to address the scientific problem with regard to the efficiency robustness of the fuzzy decision system in the RA procedure. The second part is the fuzzy-rough RA approach optimized by the fuzzy decision system-based feature evaluation mechanism. By using the fast fuzzy decision system based feature evaluation mechanism, the anti-noise fuzzy-rough nearest-neighbour algorithm and the flexible uncertainty classification method, this research enhances the associated RA methods’ capabilities of resisting noise and comprehensively handling the multi-uncertainty factors, respectively. And it further provides a solution to the scientific problem of the anti-noise robustness and flexibility robustness of the RA methods. This project ulteriorly broadens the application scope of the uncertainty computing technologies, and has great value in theory and practical significance.

英文关键词： Fuzzy Decision Systems;Fuzzy-rough Sets;Feature Selection;Feature Weighting;Risk Assessment