Learning material defect patterns by separating mixtures of independent component analyzers from NDT sonic signals

Advances in Computational Intelligence

2011

Self Cite

Abstract. This paper presents a new algorithm for nonlinear prediction based on independent component analysis mixture modelling (ICAMM). The data are considered from several mutuallyexclusive classes which are generated by different ICA models. This strategy allows linear local projections that can be adapted to partial segments of a data set while maintaining generalization (capability for nonlinear modelling) given the mixture of several ICAs. The resulting algorithm is a general purpose technique that could be applied to time series prediction, to recover missing data in images, etc. The performance of the proposed method is demonstrated by simulations in comparison with several classical linear and nonlinear methods.

Section: Resultsmentioning

confidence: 99%

Nonlinear Prediction Based on Independent Component Analysis Mixture Modelling

Advances in Computational Intelligence

2011

Self Cite

“…We selected the Mixca algorithm for the preprocessing step. This algorithm implements non-parametric source density estimation, which has allowed its application to different applications such as NDT [2][14] and biosignal processing [15]. Fig.…”

Section: Performance Analysis Of Background Estimation Methodsmentioning

confidence: 99%

“…Non-destructive testing (NDT) has been supported by computational intelligent methods such as neural networks [1] and independent component analysis (ICA) [2]. The use of ICA for Ground Penetrating Radar (GPR) signal processing has been recently studied in [3], [4], [5], [6].…”

Section: Introductionmentioning

confidence: 99%

Application of Independent Component Analysis for Evaluation of Ashlar Masonry Walls

Advances in Computational Intelligence

2011

Self Cite

Abstract. This paper presents a novel application of Independent Component Analysis (ICA) to the evaluation of ashlar masonry walls inspected with Ground Penetrating Radar (GPR). ICA is used as preprocessor to eliminate the background from the backscattered signals and thus signal-to-noise ratio of the GPR signals is enhanced. Several experiments were made on scale models of historic ashlar masonry walls. These models were loaded with different weights, and the corresponding BScans were obtained. ICA shows the best performance to enhance the quality of the B-Scans compared with classical methods used in GPR signal processing.

“…These waves propagate inside the material structure and its response is measured by signals that contain the reflections produced by the material grain microstructure plus the echoes caused by the in-homogeneitiesinhomogeneities inside the material [1,2]. Signal processing allows extraction of information for characterization of the propagation medium and for detecting material in-homogeneitiesinhomogeneities [3]. In this paper, we present results in detection of defects in the wall scale models by impact-echo, ultrasounds and GPR separately (each one in a separate sectionSection 2, 3 and 4) and enhanced results (Section 5) obtained by fusing all three results.…”

Section: Introductionmentioning

confidence: 99%

Assessment of Historic Structures Based on GPR, Ultrasound, and Impact-Echo Data Fusion

et al. 2013

KEM

Self Cite

Abstract. This paper presents a method for assessment of historic structures based on the fusion of data from ground-penetrating radar (GPR), ultrasound, and impact-echo testing. The method consists of the following steps: measuring, feature extraction, fusion, representation, and evaluation. The employed techniques for an application in scale models of historical walls are described. Thus, experimental deployment; signal feature processing; fusion operators (including order statistics digital filters); 2D non-destructive testing images, and figures of merits of the fused results are explained in detail. The deformation of different imperfections in the material structure related to the application of weight load increments applied on the wall is analyzed by using different kinds of fusion configurations.