Skip to main content
. 2020 Jul 3;20(13):3721. doi: 10.3390/s20133721

Table 3.

Results and characteristics offered by the proposed work and previous methods.

Work Proposed Methods Damage Level Accuracy (%)
[9] 1. Feature extraction is performed by using Homogeneity analysis
2. Gaussian probability density function is employed as classifier.
HBRB, 1- and 2BRB 99
[10] 1. Features extraction is performed by using MUSIC technique
2. Bayes method is employed as classifier.
1- and 2BRB 100
[12] 1. Features extraction is performed by using Wavelet and Hilbert transforms.
2. Linear discriminant technique is employed as classifier.
1- and 2BRB 100
[23] 1. Feature extraction is performed by using Fractal dimension
2. Fuzzy logic is employed as classifier.
HBRB, 1- and 2BRB 95
[26] 1. Features extraction is performed by using extended Kalman filter
2. MUSIC technique is employed as classifier.
HBRB and 1BRB 100
[43] 1. Wavelet transform is used to transform the measured signals to images.
2. A CNN is employed as features estimator and classifier.
3BRB 99
[70] 1. Features extraction is performed by using Wavelet transform.
2. Correlation Pearson is employed as classifier.
HBRB, 1- and 2BRB 95
[71] 1. Feature extraction is performed by using Hilbert transform.
2. Gaussian probability density function is employed as classifier.
HBRB, 1- and 1½BRB 99
Proposed work 1. Short time Fourier transform is used to transform the measured signals to images.
2. A CNN is employed as features estimator and classifier.
HBRB, 1- and 2BRB 100