节点文献

基于融合量测预报器的联邦滤波故障检测方法

New Fault Detection Method for Federated Filter Based on Fusion Measurement Predictor

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 邱恺吴训忠陈天如魏瑞轩张宗麟

【Author】 Qiu Kai, Wu Xunzhong, Chen Tianru, Wei Ruixuan, Zhang Zonglin(Institute of Engineering, Air Force Engineering University, Xi’An 710038)

【机构】 空军工程大学工程学院三系

【摘要】 基于融合状态递推器对各个子传感器的量测预报,本文提出了一种新的联邦滤波器传感器故障检测方法。该方法基于融合状态递推器对各个子传感器的量测预报,利用高精度的融合量测预报与各传感器量测之间的偏差构造x~2统计量,进行故障检测。该方法的重要特征是实现先检测后滤波,可以在各子滤波器启动之前确定对应传感器量测的有效性。仿真结果表明,与传统的检测方法相比,不但具有较高的故障检测灵敏度,同时降低了复杂性和提高了运算效率。

【Abstract】 A new fault detection method using a fusion state propagator to predict the measurement of each local sensor is presented to detect sensor failures for the federated filter. Failure detection is based on the Chi-square statistic hypothesis test and performed by checking the consistency between the outputs of the local sensors and that of the fusion measurement predictor, which is served as the common failure detection reference. The key to this method is that the local filters can be started up only after no failures are detected in their corresponding sensors. By this method, high fault detection sensitivity can be derived as well as a low computation load. In addition, the complexity yielded by the traditional fault detection algorithm (such as state Chi-square test) can be reduced. Simulation shows this method is effective and useful.

  • 【会议录名称】 第二十四届中国控制会议论文集(下册)
  • 【会议名称】第二十四届中国控制会议
  • 【会议时间】2005-07
  • 【会议地点】中国广州
  • 【分类号】TP18
  • 【主办单位】中国自动化学会控制理论专业委员会
节点文献中: