摘要
研究飞行发动机的数据异常实时检测问题。针对现代飞机安装的飞机通信寻址与报告系统只能实时向地面传输数据,不能对飞行数据进行分析判断,为了有选择的将异常数据发送到地面,提出能对数据进行实时分析判断的异常识别机制。为了保证检测的实时性和区别异常检测类型,利用改进递归小波变换和小波变换模极大值原理对飞行数据中的异常值进行实时检测,获取异常值的类型、位置、对应的时间,通过MATLAB完成对飞行数据的仿真。实验结果表明,改进递归小波变换和小波变换模极大值方法的飞行数据异常检测机制能准确地对数据进行分析和判断。
Aircraft communications addressing and reporting system can transmit the data in real-time, but cannot judge the flight data. In order to have a choice to send the anomaly data to the ground, this paper proposed an anomalous recognition mechanism which can analyze the data in real-time. In order to ensure the real-time detection and differentiate anomaly detection type, we detected outliers of flight data in real time based on improved recursive wavelet transform and wavelet transform modulus maxima method, and obtained the type, location and the corresponding time of outliers. We completed the simulation of flight data through the MATLAB. The results show that the anomalous detection system based on the improvement of recursive wavelet transform and wavelet transform modulus maxima method can be used to analyze the data distinctly.
出处
《计算机仿真》
CSCD
北大核心
2016年第9期118-122,共5页
Computer Simulation
基金
民航局重大专项(MHRD20130112)
国家重点基础研究发展计划(973计划)项目(2014CB744904)
关键词
飞行数据
异常检测
实时
小波变换
Flight data
Anomalous detection
Real time
Wavelet transform
