传感器故障检测与隔离算法研究

摘要:

传感器数据采集和分析是桥梁健康系统对桥梁状态评估的基础。由于传感器采集的数据格式复杂、信息量大,如不能有效的对传感器故障进行自动检测和隔离,将影响评估的准确性,产生错误预警信息。本文提出一种基于几何后非线性 ICA(Geometric Post Nonlinear ICA,gp ICA)的传感器故障检测与隔离算法。该算法通过引入几何后非线性(PNL)混合模型,将非线性数据线性化,再利用快速独立元分析(FastICA)对故障进行检测。通过计算监测数据对监控统计量的贡献度,基于贡献度分析法实现对故障传感器的隔离。利用MATLAB软件进行数值模拟,实现了模拟故障传感器的检测和隔离。该算法相比传统的线性ICA故障检测具有更高的故障检测率,更适用于桥梁健康监测系统的故障检测与隔离。。

Abstract:

Sensor data collection and analysis are the basis for bridge health monitoring system to evaluate the bridge condition. Due to the complex format of the data collected by the sensors and the large amount of information, the failure of sensors could not be automatically detected and isolated effectively, the accuracy of the evaluation will be affected and false warning information will be generated. This paper proposes a sensor fault detection and isolation algorithm based on Geometric Post Nonlinear ICA (gp ICA). The algorithm introduces a post-geometric nonlinear (PNL) hybrid model to linearize the nonlinear data, and then uses Fast Independent Element Analysis (FastICA) to detect faults. By calculating the contribution of the monitoring data to the monitoring statistics, the faulty sensor is isolated based on the contribution analysis method. Using MATLAB software for numerical simulation, the detection and isolation of simulated fault sensors are realized. This algorithm has a higher fault detection rate than traditional linear ICA fault detection, and is more suitable for fault detection and isolation of bridge health monitoring systems.