Bridge performance assessment based on an adaptive neuro-fuzzy inference system with wavelet filter for the GPS measurements

Abstract

This study describes the performance assessment of the Huangpu Bridge in Guangzhou, China based on long-term monitoring in real-time by the kinematic global positioning system (RTK-GPS) technique. Wavelet transformde-noising is applied to filter the GPS measurements, while the adaptive neuro-fuzzy inference system (ANFIS) time series output-only model is used to predict the deformations of GPS-bridge monitoring points. In addition, GPS and accelerometer monitoring systems are used to evaluate the bridge oscillation performance. The conclusions drawn from investigating the numerical results show that: (1)the wavelet de-noising of the GPS measurements of the different recording points on the bridge is a suitable tool to efficiently eliminate the signal noise and extract the different deformation components such as: semi-static and dynamic displacements; (2) the ANFIS method with two multi-input single output model is revealed to powerfully predict GPS movement measurements and assess the bridge deformations; and (3) The installed structural health monitoring system and the applied ANFIS movement prediction performance model are solely sufficient to assure bridge safety based on the analyses of the different filtered movement components.