# Structural Damage Identification Using Improved RBF Neural Networks in Frequency Domain

TL;DR: The novel improved RBF network is shown to be a good damage identification strategy for multiple member structures compared to conventional RBF and existing hybrid methods in terms of accuracy and computational effort.

Abstract: This paper presents a novel two stage improved Radial basis function (RBF) neural network for the damage identification of multimember structures in the frequency domain. The improvement of the proposed RBF network is carried out in two stages, viz. (i) first stage damage prediction by conventional RBF network trained with effective input-output patterns and (ii) in the second stage, minimization of the prediction error below the predefined error tolerance (3%) by training the network with patterns from reduced search space located after the first stage prediction. The network effective input patterns are fractional frequency change ratios (FFCs) and damage signature indices (DSIs), and the corresponding output patterns are stiffness values or damage severity of the structure at different damage levels. A Latin hypercube search (LHS) technique is used for finding the effective input-output patterns from the search space to improve the training efficiency. The numerical simulation of structural damage iden...

...read more

##### Citations

18 citations

14 citations

8 citations

8 citations

### Cites background or methods from "Structural Damage Identification Us..."

...The Euclidean distance is calculated from xm ci k k(2) (Machavaram and Shankar, 2012)....

[...]

...where Wni is the weight vector in the output layer for the nth output node and i is the radial basis function of the ith node (Machavaram and Shankar, 2012)....

[...]

...The output from the i th Gaussian neuron from an input xm is calculated using i ¼ exp xm cik k2 2i ð1Þ The Euclidean distance is calculated from xm cik k2 (Machavaram and Shankar, 2012)....

[...]

...The output of the hidden neuron is computed as: yn¼ X i Wni i ð2Þ where Wni is the weight vector in the output layer for the nth output node and i is the radial basis function of the ith node (Machavaram and Shankar, 2012)....

[...]

8 citations

##### References

6,318 citations

6,213 citations

^{1}

4,272 citations

### "Structural Damage Identification Us..." refers background or methods in this paper

...They are an active research topic and have been applied to the solution of many problems (Moody and Darken 1989)....

[...]

...Conventional RBF neural networks are trained in three stages (Moody and Darken 1989; Reddy and Ganguli 2003; Zang et al. 2007)....

[...]

...…study is exact interpolation type where the number of centroids (J) and the size of training patterns (m) are equal. c) The two training parameters (Moody and Darken 1989): the centroids represented by cj and the width or spread represented by σj (Section 2) are selected from the LHS clusters and…...

[...]

...c) The two training parameters (Moody and Darken 1989): the centroids represented by cj and the width or spread represented by σj (Section 2) are selected from the LHS clusters and random selection by trial and error basis respectively to train the network....

[...]

2,777 citations

### "Structural Damage Identification Us..." refers methods in this paper

...Doebling et al. (1996) presented a review on detection, location, and characterization of the structural damage via techniques that examine changes in measured structural vibration response....

[...]

584 citations

### "Structural Damage Identification Us..." refers background or methods in this paper

...The size of an LHS design matrix thus grows much more slowly with k than more conventional experimental designs (e.g., factorial, fractional factorial or Central Composite Designs (Kleijnen 2008))....

[...]

..., factorial, fractional factorial or Central Composite Designs (Kleijnen 2008))....

[...]