Bridge diagnosis system by using nonlinear independent component analysis

The aim of this paper is to structure a daily diagnosis system for bridge monitoring and maintenance. Technology of wireless sensor, signal processing and structure analysis are used in this diagnosis system. The vibration data are collected through wireless sensor network by exerting some external forces such as running vehicles. Then use nonlinear independent component analysis (nonlinear ICA) and spectral analysis to analyze the data for extracting character frequency. In the past, linear ICA such as FastICA is used to do the signal processing step. But simple linear ICA algorithms work efficiently only in linear mixing environments. Whereas a nonlinear ICA model, which is more complicated, would be more practical for bridge diagnosis system. In this paper, we firstly use post nonlinear (PNL) method to change the data from nonlinear to linear, after that do linear separation by FastICA. Through the processed data this diagnosis technology can be used to understand the phenomena like corrosion and crack and evaluate the health condition of a bridge. We apply this system to do experiments at Nakajima Bridge in Yahata, Kitakyushu, Japan and successfully extract the character frequency of a bridge.