一种面向多传感器系统的故障检测与隔离算法

A Fault Detection and Isolation Algorithm for Multi-Sensor System

摘要: 故障检测与隔离技术是自确认传感器的核心研究内容之一。本文针对多传感器系统的异常状态监测过程中的故障检测与隔离准确率较低的问题，提出一种基于主成分分析(PCA)自适应重构技术的故障检测与隔离算法。该方法利用多传感器系统的正常数据建立PCA模型，并通过SPE统计量对多传感器系统的多路信号进行实时监测。一旦故障发生，SPE统计量的数值将显著提高，从而实现对传感器故障的检测；为了进一步对故障传感器进行隔离，采用SPE统计量自适应重构技术，实现对故障传感器的定位。仿真实验表明，本文提出的多传感器系统的故障与隔离算法能够有效地实现多传感器系统的故障检测与隔离，具有较高的准确率。

Abstract: Fault detection and isolation technology is one of the core research contents of self-identified sensors. Aiming at the problem of low fault detection and isolation accuracy in the abnormal state monitoring process of multi-sensor system, this paper proposes a fault detection and separation algorithm based on principal component analysis (PCA) adaptive reconstruction technology. The method uses the normal data of the multi-sensor system to establish the PCA model, and monitors the multi-channel signals of the sensor system in real time through the SPE statistics. Once the fault occurs, the value of the SPE statistic will be significantly improved to achieve the detection of the sensor fault. In order to further isolate the fault sensor, the SPE statistic adaptive reconstruction technology is used to locate the fault sensor. Simulation experiments show that the fault and isolation algorithm of multi-sensor system proposed in this paper can effectively realize fault detection and isolation of multi-sensor system, and has high accuracy.