Structural Novelty Detection Based on Adaptive Consensus Data Fusion Algorithm and Wavelet Analysis

摘要

This paper proposes a new structural novelty detection method in the case of vast measurement data having uncertainties. Considering the effects of measurement accuracy and environmental variations on measurement variance, precise analytical methods of adaptive confidence distance and measurement variance are presented on the basis of statistical theory, and thus an adaptive consensus data fusion algorithm has been firstly developed to deal with the large volume of data involving considerable uncertainties. The proposed adaptive fusion algorithm can adaptively choose sensors whose data will be subsequently fused. The algorithm is then incorporated with wavelet analysis for the purpose of structural novelty detection. Two numerical examples are carried out to validate the efficiency and adaptability of the proposed method. The obtained results have been compared with those from other existing methods, which demonstrate the high efficiency of the proposed method in data processing considering uncertainties and unsatisfied performance of some sensors, as well as its accuracy in structural novelty detection. The proposed method also shows some robustness to noise.