RELIABILITY EVALUATION MODEL BASED ON DATA FUSION FOR AIRCRAFT ENGINES

Reliability evaluation for aircraft engines is difficult because of the scarcity of failure data. But aircraft engine data are available from a variety of sources. Data fusion has the function of maximizing the amount of valu- able information extracted from disparate data sources to obtain the comprehensive reliability knowledge. Consid- ering the degradation failure and the catastrophic failure simultaneously, which are competing risks and can affect the reliability, a reliability evaluation model based on data fusion for aircraft engines is developed, Above the characteristics of the proposed model, reliability evaluation is more feasible than that by only utilizing failure data alone, and is also more accurate than that by only considering single failure mode. Example shows the effective- ness of the proposed model.

基于时间序列分析的分布式拒绝服务攻击检测

该文分析了分布式拒绝服务(DDoS)攻击的特点,提出了一种基于流连接密度(FCD)时间序列分析的DDoS攻击检测方法,该方法通过拟合FCD时间序列的自适应自回归模型,获得能够在多维空间描述当前流量状态的AAR模型参数向量序列,然后使用经过样本训练的支持向量机(SVM)分类器进行攻击识别;充分考虑了报警的时间间隔及分布情况,提出一种报警可信度评估算法对SVM分类结果进行二次处理,以消除网络流量噪声及分类错误所带来的影响.实验结果显示,该检测方法能够有效检测DDoS攻击,可信度评估算法能够明显减少误报,降低误报率,显著提高检测质量.

基于监控日志的航班查询系统可靠性评估方法

针对运行阶段的系统软件可靠性评估理论和评估方法的研究缺失,提出一种基于系统监控日志的软件可靠性评估方法。以中国航信的航班查询系统为可靠性评估研究对象,首先解析该系统的监控日志,提出基于监控日志的失效数据提取方法。其次,基于航班查询系统的特点和可靠性数据的分析结果,构建了航班查询系统可靠性模型。最后,实验验证了该可靠性模型是可行的,并利用此模型评估了航班查询系统的可靠性,形成了系统软件可靠性评估过程方法。