Multilayer Perceptron Neural Network for Damage Identification Based on Dynamic Analysis

Detection of damages with low levels has been one of the most critical challenges. As a result, many damage detection methods cannot detect damages or cracks with a level lower than 10%. On-surface damages as low-level damages are challenging to localize. A new technique is proposed to eliminate this challenge based on wavelet transformation of the difference in damaged and intact mode shapes. In this way, a finite element model is developed for obtaining governing equations of thin beams. The developed finite element model provides the mode shape signals. Then, the signals are decomposed by wavelet transform. The findings of this study show that in both numerical and experimental investigations, the proposed method is very efficient since the proposed method can detect on-surface damages having a level below 10%.

show abstract

Multilayer Perceptron Neural Network for Damage Identification Based on Dynamic Analysis

Cited by 3 publications

References 13 publications

Damage Identification in Thin Steel Beams Containing a Horizontal Crack Using the Artificial Neural Networks

Damage Identification in Thin Steel Beams Containing a Horizontal Crack Using the Artificial Neural Networks

Damage Detection in Rectangular Laminated Composite Plate Structures using a Combination of Wavelet Transforms and Artificial Neural Networks

Damage detection of beam-like structures using a combination of wavelet transform and subtraction of intact and damaged mode shapes

Contact Info

Product

Resources

About