摘要
针对水下目标跟踪和石油地震勘探等领域中传感器故障检测的必要性和复杂性,提出了一种带乘性噪声系统的传感器故障模型。并基于状态估计,给出了故障发生前后新息序列正交性变化的结论及相关证明,由此建立了故障检测指标。该方法摆脱了基于状态估计的故障检测方法中对新息需满足正态分布特性的束缚,同时显著地提高了故障诊断的精度,也在一定程度上实现了对滤波器工作性能的监测及观测数据可靠性的判断。仿真结果表明:该方法能够及时有效的识别出故障发生点。
In view of the necessity and complexity of sensor fault detection as to the underwater target tracking and seismic exploration for oil,we propose a new sensor fault model based on systems with multiplicative noise.Before and after the fault happen we prove the statistical properties of innovation,based on which a new fault indicator is build.Different from the traditional methods restricted with the Gaussian distribution characteristics of the innovation,the new fault detection method is more flexible,which can not only improve the accuracy of fault detection markedly but achieve better performance in testing the filter and determining the reliability of the observational data.Simulation results show that the method can efficiently identify the fault point.
出处
《中国海洋大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第4期112-116,共5页
Periodical of Ocean University of China
基金
山东省自然科学基金项目(ZR2010DQ003)资助
关键词
故障检测
乘性噪声
状态估计
新息
非高斯
正交性
fault detection
multiplicative noise
state estimation
innovation
non-Gaussian
orthogonality
