Case-Based Reasoning for Stiffness Changes Detection in Structures

Abstract

This chapter presents an expert monitoring algorithm approach to detect, locate and quantify stiffness variations in structures. The algorithm is based on pattern recognition and artificial intelligence techniques that emulate knowledge based on human reasoning. The expert system (ES) uses time-frequency information about dynamics of structure, which is processed by using discrete wavelet transform (DWT), self-organizing maps (SOM), case-based reasoning (CBR) and principal component analysis (PCA). In addition, two applications are considered in order to evaluate the effectiveness of vibration analysis methodology and CBR in damage detection. The first application (Camacho 2010) uses the environmental excitation to detect and quantify damage in a Mechanical UBC ASCE Benchmark. The second one (Sandoval 2010) uses a predesigned signal to detect geometric damages on a gas pipeline. In both cases, a finite element model (FEM) is used to simulate different damages scenarios, which correspond to stiffness variations in different location.