Sensor Fault Diagnosis for Structural Health Monitoring Based on Statistical Hypothesis Test and Missing Variable Approach

Abstract

Using structural monitoring data collected from a sensor network to assess the health condition of a monitored structure relies on the accurate operation of the sensors and therefore could be affected by various sensor faults. This paper presents a sensor-fault detection and isolation approach with application to structural health monitoring. Principal-component analysis (PCA) is first applied to model the fault-free history monitoring data to generate uncorrelated residuals, which can be seen as the projection of the additional measurement noise into the residual subspace of the PCA transform. Then, under the assumption that the measurement noise is Gaussian distributed, a statistical hypothesis test model is established for the subsequent sensor-fault detection procedure, after that two fault detectors are deduced through the rejection of the null hypothesis. Next, the missing variable approach is used to establish an isolation index to identify the specific faulty sensor. A benchmark structure developed for bridge health monitoring is adopted to validate and demonstrate the performance of the proposed method, and the analysis results indicate that the method is effective in detecting and isolating both bias and drift sensor faults.