基于盲源分离与信息不完备的结构响应分析与安全性评定

项目名称： 基于盲源分离与信息不完备的结构响应分析与安全性评定

项目编号： No.50878057

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 武器工业

项目作者： 姜绍飞

作者单位： 福州大学

项目金额： 36万元

中文摘要： 本项目采用盲源分离、统计与智能信息处理等技术，从信号分析与噪声剔除技术、不完备数据处理与非线性特征选择与提取技术、智能损伤识别与可靠性评定技术三个方面展开技术攻关。采用盲源分离技术进行信噪分离,研究、开发了2种信号处理技术和2种消噪新方法，并申报3项国家发明专利;采用统计、优化等理论进行线性/非线性特征选择与提取技术研究，研究了动力特征指纹，提出了将粒子群优化与核主组分分析的非线性特征提取新技术；系统地研究了不完备数据下的信息处理方法，结合数据融合概念与实测环境因素的影响，研究提出了基于一致性数据补齐方法和自适应数据融合方法，并申报了2项国家发明专利；运用智能信息处理技术处理海量、不确定数据的优势，将数据融合、神经网络与支持向量机技术作为分类器，并结合非线性特征选择与提取技术，研发了多种智能损伤检测技术，并将子结构技术运用到可靠度评价之中，提出了新的基于子结构技术的大型结构可靠性评定新方法.项目圆满完成了立项任务并取得了预期成果，解决了影响结构健康监测的部分技术"瓶颈"问题，而且申报了5项国家发明专利,为以后开发相关的信号处理仪器、健康监测设备提供技术支持.

中文关键词： 信号处理;盲源分离;信息不完备;智能信息处理;安全评定

英文摘要： Three techniques, i.e., Blind Sources Separation (BSS), statistical and intelligent information processing (IIP) techniques are employed in this project. They are used to investigate in three aspects, namely, signal processing and de-noising, nonlinear feature extraction, and intelligent damage detection and reliability assessment in the case of incomplete data. Two signal processing analysis methods and two de-noising techniques are proposed and 3 invention patents are applied using BBS and wavelet analysis techniques. Statistical and optimization theory are used to extract and select features. After the dynamic indices are studied, a novel nonlinear feature extraction technique by integrating the particle swarm optimization (PSO) and kernel principal component analysis (KPCA) is presented. The data processing methods are systematically studied in the case of incomplete data, consequently a consensus data supplement method and an adaptive data-fusion algorithm are proposed. Two invention patents are also applied. Owing to the virtues of IIP technique, a few intelligent damage detection techniques are developed based on the integrating the above-mentioned nonlinear feature extraction methods and intelligent classifiers, i.e. neural network, support vector machine and data fusion. On the basis of the sub-structure technique, a novel structural reliability assessment method is presented for large-scale structures finally. The project has completely finished the tasks of the contract and the expected achievements are also achieved in the proposal. This investigation can not only partly solve technical bottle-neck problems influencing on structural health monitoring, but also apply 5 invention patents; furthermore, it can provide technical supports on the development of signal processing and structural health monitoring instruments in the future.

英文关键词： signal processing; blind sources separation; incomplete information; intelligent information processing; safety assessment