摘要
为提高多传感器组合导航系统对各导航传感器的在线故障检测能力,提出了一种基于调频高斯小波变换的导航传感器故障诊断改进算法。该算法在分析调频高斯小波特性的基础上,采用高斯小波变换计算出观测量的小波系数后,然后利用带遗忘因子的数据平滑算法对小波系数进行平滑,通过判断平滑值来诊断导航工作正常与否。其优点是仅利用传感器的观测量来直接检测导航传感器故障,适当选择小波变换的拉伸因子和数据的衰减因子可以对方差突变等软故障进行有效的在线检测,并解决了误检问题。仿真结果证明了该算法的有效性。
An improved MGWT（modulated Gaussian wavelet transformation）-based fault diagnosis algorithm for navigation sensors was presented to improve the online fault-detecting capability of each sensor in multi-sensor integrated navigation system. Based on the characteristics analysis of modulated Gaussian wavelet（MGW）, the algorithm could calculate wavelet coefficients of the observing variables using Gaussian wavelet transfer（GWT）, then smoothed them by the data smoothing algorithm with forgotton-gene, and decided whether the navigation works properly. The advantage of this algorithm is that the effectiveness of each sensor can be directly detected using only the observing variables of sensors. Some soft malfunction such as abnormality of covariance can be detected online, and the false detecting can be avoided by proper selecting stretching and data-attenuation coefficient. Simulation result proves the effectiveness of this method.
出处
《中国惯性技术学报》
EI
CSCD
2007年第1期112-115,共4页
Journal of Chinese Inertial Technology
基金
国防预研项目(51309060305)
关键词
组合导航
故障诊断
算法
调频高斯小波变换
遗忘因子
integrated navigation
fault diagnosis
algorithms
modulated Gaussian wavelet transformation
forgotton-gene
