A case study of processing abnormal GNSS monitoring data of a cable-stayed bridge in Vietnam

Abstract

Global Navigation Satellite System (GNSS) plays an important role in the real-time monitoring of superstructures, especially for long-span cable-stayed bridges. Currently, GNSS has been applied to integrate a Structural Health Monitoring system (SHMs) of many long-span cable-stayed bridges worldwide through its advantages in observing a large displacement of structures and monitoring the global deformation of bridges. However, some studies of actual GNSS monitoring data of a cable-stayed bridge showed that there are a lot of abnormal data occurrences such as missing data or several abnormal data. This paper investigates the application of some methods for processing the GNSS abnormal data acquired from an actual cable-stayed bridge in Vietnam. Firstly, a long-term monitoring dataset of an actual bridge was acquired for the study experiment. A clean-short dataset was used to investigate the accuracy and applicability of some methods in the interpolation of abnormal data, for example, linear formula, Moving Average, Artificial Neural Network, and Hampel Identifier. Some criteria are used to assess the difference between the interpolated data and actual data such as Root Mean Square Error (RMSE) or correlated coefficients with temperature data. Otherwise, the applied methods were then assessed their abilities and effectiveness in real applications, for instance, creating an automatic interpolated program of abnormal data.