Performance analysis of simplified Fuzzy ARTMAP and Probabilistic Neural Networks for identifying structural damage growth

doi:10.1016/j.asoc.2016.12.020

Applied Soft Computing

Volume 52, March 2017, Pages 53-63

https://doi.org/10.1016/j.asoc.2016.12.020 Get rights and content

Highlights

•
An unprecedented comparative performance analysis between Fuzzy ARTMAP and Probabilistic Neural networks applied to identifying structural damage.
•
The fuzzy ARTMAP method is better suited to the problem of damage growth in regard to success and training/testing times.
•
The suitable damage prognosis method can lead to an increase in safety and reduce maintenance costs in a variety of applications.
•
It contributes to helping researchers choose more effective approaches for structural damage growth.

Abstract

Structural Health Monitoring (SHM) has been advancing worldwide as evidenced by recent papers and practical applications. SHM systems increase safety and reduce maintenance costs in a variety of applications contributing significantly to prevent structural failures. Studies focusing on the structural damage identification have been proposed in literature by using methods based on Probabilistic Neural Networks (PNN). Although Fuzzy based methods are recurrent in SHM approaches, the Simplified Fuzzy ARTMAP Network (SFAN) has not been explored for analyzing structural damage. Furthermore, there is no evidence of comparative studies of these two methods when applied to the problem of progressing structural damage. Thus, this paper presents a comparative analysis of these two methods in the context of identifying structural damage growth. The comparison is made based on the factors of success rate and training/testing times. Additionally, this approach has carried out the performance analysis of suitable parameters setup for SFAN: choice parameter, training rate and vigilance parameter. As a practical case study, both methods were applied to a unidirectional composite plate containing four PZT (Lead Zirconate Titanate) patches where the damage growth scenarios were also simulated by loosening bolts for three different levels. In addition, the repaired structural condition was also considered by retightening bolts. The results have shown that both methods are suitable for the problem of damage growth, particularly to supporting decisions about the structural damage assessment. In short, the comparative analysis has shown that the SFAN method is better suited to the problem of damage growth especially in regard to training and testing times. Thus, this comparative study contributes to helping researchers and industrialist in choosing more effective approaches for structural damage growth.

Introduction

The world commercial aircrafts fleet has increased on average 1.8% during 2010 reaching more than 25,000 new aircrafts in operation [1]. Furthermore, there are many old aircraft that are still operating in the global air space [2]. Likewise, advanced composite materials such as used in new projects of modern aircraft have become an interesting field of development and research for both academia and industry. For example, Airbus designed the A350XWB using 53% of its structure made up of composite materials becoming a landmark in the history of commercial aircraft due to transcending the barrier of the 50% in using composite materials [3]. However, all those aforementioned structures are also susceptible to structural failure especially after a long time in operation. Hence, Structural Health Monitoring (SHM) and damage detection have been advancing worldwide as evidenced by recent papers and practical applications. In particular, Nondestructive Evaluation (NDE) methods allow for detecting a variety of structural damage such as delamination, cracks, rivets or screws loosened and others. Also, new techniques and methods for structural failures prognosis and identification reduce cost and enable more efficient maintenance [4].

Within this context, Electromechanical Impedance (EMI) based methods have received attention. The method uses small and lightweight piezoelectric wafers as active sensor attached to the structure to be monitored. Piezoelectric/piezocomposite sensor system has been used as a useful tool for assessment of structural damage applied to different kinds of structures. However, methods based on EMI when applied to composite structures are challenging due to increased damping in the impedance signatures (resonant/antiresonant peaks) [5]. PZT (Lead Zirconate Titanate) patches cover a small sensing area, around 0.4 m (radius), when applied to aluminum structures [6]. Yet the coverage area of the PZT may be drastically reduced for structure made up of composite by using higher frequencies (EMI-based methods) [5]. Thus, new approaches exploring EMI methods have been proposed for damage prognosis in composite structures to overcome the aforementioned issues [7], [8], [9].

Currently, there is a vast literature about SHM approaches focusing on Artificial Neural Network (ANN) along with EMI methods, applied to structural damage identification [10], [11], [12]. Lately, new methods based on Probabilistic Neural Network (PNN) have been proposed [13] for predicting structural damage. Therein, damage was simulated drilling holes to the glass fiber plate in different positions. According to [13], methods based on PNN have a faster training procedure and are easier than backpropagation. Also, their damage classifier is guaranteed to converge as the size of the representative training set increases. In [14], authors developed a method based on the EMI for detecting and locating damage applied to a carbon reinforced fiber composites. The localization process was based on PNN, where inputs were derived from the principal components obtained from damage metrics. A method based on EMI along with PNN and Fuzzy cluster analysis for identification, location, and classification of structural damage was proposed in [15]. Therein, damage was simulated by loosening a rivet in an aluminum panel.

Methods based on Fuzzy systems have been proposed in literature for damage identification in structures [15], [16]. However, methods based on Fuzzy ARTMAP Network (FAN) applied to structural damage identification have not been explored in literature. On the other hand, FAN based methods have been applied successfully in different fields such as electrical power system forecasting [17], [18], computer network intrusion detection [19], image recognition systems [20] and medical/biomedical applications [21]. The exception is [22], where the authors proposed identification of flaws in aircraft structures using an ARTMAP-Fuzzy-Wavelet artificial neural network. Signals obtained from a numerical model (finite elements) of an aluminum beam were used as input for FAN. Failure was simulated in wear levels 1, 5, 10, 15, 20, 25 and 30%. Those results showed a success of 100% for failure identification, thereby promising an excellent tool for identifying the progress of structural damage.

The purpose of the research presented here is to evaluate the performance of the Probabilistic Neural Network (PNN) and Simplified Fuzzy ARTMAP Network (SFAN) applied to the progression of structural damage. The comparison was made based on several factors: success rate and training/testing times. Furthermore, this approach has carried out the performance analysis for the parameters setup of SFAN: choice parameter, training rate and vigilance parameter. In order to evaluate the performance, both methods were applied to a unidirectional composite plate with four PZT patches attached. The damage growth scenarios were simulated by loosening bolts to three different levels. Additionally, the repaired structural condition was also considered by retightening bolts. The experimental setup was carried out based on the EMI technique, in the time domain, along with Euclidean distance. A performance analysis for the comparative study showed that both methods can be useful for identifying the growth of structural damage making it an excellent approach for SHM applications.

The remainder of the paper is organized as follows: a brief review of EMI technique applied to SHM application; the main concepts about Fuzzy ARTMAP; and Probabilistic neural networks. Next, test scenarios and experimental case study based on the EMI method are shown followed by the results for both SFAN and PNN. Finally, the paper concludes highlighting advantages and remarks of the proposed approach.

Section snippets

Electromechanical impedance

The technique based on the Electromechanical Impedance (EMI) applied to SHM was introduced by [23] and improved by several other authors [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36]. EMI methods are considered an important technique of NDE which is based on the Frequency Response Function (FRF). The EMI based methods have received a lot of attention lately by using small and lightweight piezoelectric wafer active sensor attached to the structure to be monitored.

Experimental case study

Experimental tests were conducted by using the acquisition system proposed by [27]. Their system was developed on the LabVIEW^® platform and makes use of a National Instruments Data Acquisition (DAQ) device, model USB-6259 (Fig. 4). In this system, the DAQ provides the signals x[n] and y[n] in discrete forms. Furthermore, the DAC (digital to analog converter) and the ADC (analog to digital convertor), which are integrated into the DAQ device, are synchronized ensuring for each sample n generated

Results and discussion

In this section an EMI-based experimental setup is used as a case study to evaluate the performance of identifying the progress of structural damage by using both ED/SFAN and ED/PNN. All analysis is carried out in a real composite structure shown in Fig. 5.

Conclusions

This work evaluated the performance of two methods: Simplified Fuzzy ARTMAP Network (SFAN) and Probabilistic Neural Network (PNN) applied to identifying structural damage growth by exploiting the electromechanical impedance technique. In order to test both methods, experiments on a unidirectional composite plate with four PZT patches attached were performed. The best performance for damage identification based on the SFAN, in terms of success rate was obtained when using the choice parameter

Funding

The authors would like to thank the CNPq, Brazilian Research Agency (Grant 248665/2013-8), and the University of Michigan through the Kelly Johnson Collegiate Chair fund.

Acknowledgments

The authors are very grateful to the reviewers for their careful and meticulous reading as well as their contributions which helped to improve the manuscript. The authors also acknowledge the useful help and the support during the experiments provided by Dr Jared Hobeck.

References (43)

S. Na et al.
Resonant frequency range utilized electro-mechanical impedance method for damage detection performance enhancement on composite structures
Compos. Struct.
(2012)
C. Bois et al.
Monitoring a delamination in a laminated composite beam using in-situ measurements and parametric identification
J. Sound Vib.
(2007)
S. Na et al.
Steel wire electromechanical impedance method using a piezoelectric material for composite structures with complex surfaces
Compos. Struct.
(2013)
A. Saxena et al.
Evolving an artificial neural network classifier for condition monitoring of rotating mechanical systems
Appl. Soft Comput.
(2007)
J. Min et al.
Impedance-based structural health monitoring incorporating neural network technique for identification of damage type and severity
Eng. Struct.
(2012)
S. Na et al.
Neural network approach for damage area location prediction of a composite plate using electromechanical impedance technique
Compos. Sci. Technol.
(2013)
P. Selva et al.
Smart monitoring of aeronautical composites plates based on electromechanical impedance measurements and artificial neural networks
Eng. Struct.
(2013)
M.L.M. Lopes et al.
Electric load forecasting using a fuzzy ART&ARTMAP neural network
Appl. Soft Comput.
(2005)
X. Lu et al.
Distributed HS-ARTMAP and its forecasting model for electricity load
Appl. Soft Comput.
(2015)
J. Vieira Filho et al.
Time-Domain analysis of piezoelectric impedance-Based structural health monitoring using multilevel wavelet decomposition
Mech. Syst. Sig. Process.
(2011)

C.G. Gonsalez et al.

Structural damage detection in an aeronautical panel using analysis of variance

Mech. Syst. Sig. Process.

(2015)

D.K. Morrow et al.

World airliner census

(2011)

C. Boller

Ways and options for aircraft structural health management

Smart Mater. Struct.

(2001)

T. Chady

Airbus Versus Boeing − Composite Materials: The Sky’s the Limit…

(2013)

K. Worden et al.

Overview of intelligent fault detection in system and structures

Struct. Health Monit.

(2004)

G. Park et al.

Overview of piezoelectric impedance-based health monitoring and path forward

Shock Vib. Digest

(2003)

M. Gresil et al.

Predictive modeling of electromechanical impedance spectroscopy for composite materials

Struct. Health Monit.

(2012)

V. Lopes et al.

Impedance-Based structural healthy with artificial neural networks

J. Intell. Mater. Syst. Struct.

(2000)

L.V. Palomino et al.

Probabilistic neural network and fuzzy cluster analysis methods applied to impedance-based SHM for damage classification

Shock Vib.

(2014)

S. Silva et al.

Structural damage detection by fuzzy clustering

Mech. Syst. Sig. Process.

(2008)

N. Araújo et al.

Performance evaluation of the fuzzy ARTMAP for network intrusion detection

Cited by (31)

Multi objective optimization methods for damage assessment of composite laminates: A review
2024, Composite Structures
The goal of this paper is to present a review of multi-objective optimization (MOO) techniques used in the damage assessment of composite laminates, with regard to damage identification, damage detection, damage quantification, and damage prediction models or approaches. Prior recognition of damage is crucial to ensure the performance and integrity of a structure of high structural responsibility. It is only then catastrophic structural failure or massive property loss and even in human casualties, in extreme situations, can be avoided or at least reduced. As such damage assessment is required, especially using advanced optimization and multi-objective optimization techniques, to prevent the failure of structural laminated composites. According to researchers, AI-based optimization algorithms are preferable to programmed inspections, conventional methods (like ultrasonic and magnetic damage or acoustic damage assessment methods), model or non-model (feature-based methods), and both in the sense that they are quick, universal, and less expensive methods for damage assessment. Unfortunately, there have only been a few or a small number of research investigations, particularly for damage prediction.
State-of-the-art review on advancements of data mining in structural health monitoring
2022, Measurement: Journal of the International Measurement Confederation
To date, data mining (DM) techniques, i.e. artificial intelligence, machine learning, and statistical methods have been utilized in a remarkable number of structural health monitoring (SHM) applications. Nevertheless, there is no classification of these approaches to know the most used techniques in SHM. For this purpose, an intensive review is carried out to classify the aforementioned techniques. In doing so, a brief background, models, functions, and classification of DM techniques are presented. To this end, wide range of researches are collected in order to demonstrate the development of DM techniques, detect the most popular DM techniques, and compare the applicability of existing DM techniques in SHM. Eventually, it is concluded that the application of artificial intelligence has the highest demand rate in SHM while the most popular algorithms including artificial neural network, genetic algorithm, fuzzy logic, and principal component analysis are utilized for damage detection of civil structures.
A new approach to online training for the Fuzzy ARTMAP artificial neural network
2021, Applied Soft Computing
The evolution of internet resources has led to an increase in the flow of data and, consequently, the need for classification or forecasting models that support online learning. The Fuzzy ARTMAP neural network has been used in the most areas of knowledge; however, few have explored real-time applications that require continuous training. In this work, a Fuzzy ARTMAP neural network with continuous training is proposed. This new network can acquire knowledge via classification or prediction. Modifications made to the architecture and learning algorithm enable online learning from the first sample of data and perform the classification or forecasting at any time during training. To validate the proposed model, three experiments were performed, one for forecasting and two for classification. Each experiment used benchmark databases and compared its final results with the results of the original Fuzzy ARTMAP neural network. The results demonstrate the ability of the proposed model to acquire knowledge from the first data samples in a stable and efficient way. Thus, this study contributes to the evolution of the Fuzzy ARTMAP neural network and introduces the continuous training method as an effective alternative to real-time applications.
Integrated electromechanical impedance technique with convolutional neural network for concrete structural damage quantification under varied temperatures
2021, Mechanical Systems and Signal Processing
Citation Excerpt :
Therefore, accurate quantification of structural damage severity under varied temperatures still necessitates more effective and intelligent approaches, on account of the complexity in solving such multivariate problems. Belonging to the second category, machine learning methods, such as ANN, probabilistic neural network (PNN) or Convolutional Neural Network (CNN) are regarded as a powerful tool to solve complex issue of damage classification by learning the representations of monitoring data [17,26,27]. Previous studies have examined the superiority of neural networks (NNs) approach for estimation of structural damage type and severity due to the versatility in dealing with various types of inputs and outputs and a quick diagnosis capability after training of the NN was completed [28].
The electromechanical impedance/admittance (inverse of impedance, EMI/EMA) technique has demonstrated its high sensitivity to structural incipient damages, however, it is also vulnerable to ambient temperature fluctuation that may adversely cause false alarms in real-life monitoring applications. This paper proposed an innovative approach integrating the EMA technique with Convolutional Neural Network (CNN) to quantify concrete structural damage severity under varied temperatures. In the approach, EMA signals were first split into multiple sub-range responses, and the corresponding statistical indices namely Pearson Correlation Coefficient (PCC), were calculated and utilized as input of one-dimensional CNN. In order to meet high accuracy for training the CNN model, an Orthogonal Matching Pursuit (OMP) algorithm based EMA data reconstruction technique was employed to generate sufficient data for the establishment of CNN data library. Effectiveness of the proposed connectionist approach was verified through crossover experiments of diagnosing crack damages on a standardized concrete cube subjected to varied temperatures. Both temperature recognition and damage quantification under elevated temperatures were accomplished via performing the EMA signal based CNN procedure. Experimental results revealed that the proposed approach was of high accuracy for temperature recognition and damage quantification, which potentially facilitates the in-situ monitoring of engineering structures under varied temperature conditions.
An anthropomorphic fuzzy model for the time-spatial assessment of sandstone seepage damage
2020, Automation in Construction
Citation Excerpt :
The most likely moisture content can be obtained using fuzzy calculation. The anthropomorphic fuzzy model can realize the depth location of seepage damage and moisture content real-time assessment [9,11]. AE data is acquired in real time through sensor arrays.
Sandstone moisture is a significant cause of structural instability and efficient seepage damage identification is an inevitable problem for complicated construction engineering. To overcome the shortcomings of traditional rock moisture measurements in terms of real-time application, prior knowledge-dependence, etc., an anthropomorphic fuzzy model is proposed. In this model, an adaptive fuzzy identification algorithm and an anthropomorphic causal reasoning method are employed to conduct real-time moisture content assessment and damage depth level location, respectively. The experiment carried on a self-developed three-layer system shows the consistency of this model compared with measurement data and finite element simulation results. The historical acoustic emission (AE) data is fully used in this model, and time-spatial assessment of sandstone seepage damage can be achieved independent of complex construction engineering knowledge. In follow-up research, the probability information between AE parameters and seepage damage would be mined through more experiments and numerical simulations.
Damage localization in irregular shape structures using intelligent FE model updating approach with a new hybrid objective function and social swarm algorithm
2019, Applied Soft Computing Journal
Health monitoring of structures and damage diagnosis are important research disciplines under investigation worldwide. Soft computing techniques are usually used to solve the uncertain complex inverse problem of revealing structural damage. In the current research, FE model updating (FEMU) paradigm is embraced for solving the damage tracking problem in three dimensional irregular shape structures. By taking into account the complexity of problem, the pivotal point is to efficiently educe damage through well-evolved objective function. Therefore, a novel objective function merging the modal characteristics of modal strain energy (MSTEN) and mode shape curvature (MSC) is established. Posteriorly, to solve the FEMU problem, a hybrid algorithm combining the particle swarm optimization with a new social version of the sine–cosine optimization algorithm (SPSOSCA) is proposed. The SPSOSCA is considered to take advantage of two enhanced search mechanisms to overcome the overall problem complexity. The proposed paradigm is evaluated using many damage scenarios even under noise conditions and the total outcome reveals outstanding performance with fair computational time.

View all citing articles on Scopus

View full text

Performance analysis of simplified Fuzzy ARTMAP and Probabilistic Neural Networks for identifying structural damage growth

Highlights

Abstract

Introduction

Section snippets

Electromechanical impedance

Experimental case study

Results and discussion

Conclusions

Funding

Acknowledgments

Compos. Struct.

J. Sound Vib.

Compos. Struct.

Appl. Soft Comput.

Eng. Struct.

Compos. Sci. Technol.

Eng. Struct.

Appl. Soft Comput.

Appl. Soft Comput.

Mech. Syst. Sig. Process.

Mech. Syst. Sig. Process.

World airliner census

Ways and options for aircraft structural health management

Smart Mater. Struct.

Airbus Versus Boeing − Composite Materials: The Sky’s the Limit…

Overview of intelligent fault detection in system and structures

Struct. Health Monit.

Overview of piezoelectric impedance-based health monitoring and path forward

Shock Vib. Digest

Predictive modeling of electromechanical impedance spectroscopy for composite materials

Struct. Health Monit.

Impedance-Based structural healthy with artificial neural networks

J. Intell. Mater. Syst. Struct.

Probabilistic neural network and fuzzy cluster analysis methods applied to impedance-based SHM for damage classification

Shock Vib.

Structural damage detection by fuzzy clustering

Mech. Syst. Sig. Process.

Performance evaluation of the fuzzy ARTMAP for network intrusion detection