scholarly journals A Novel Assisted Artificial Neural Network Modeling Approach for Improved Accuracy Using Small Datasets: Application in Residual Strength Evaluation of Panels with Multiple Site Damage Cracks

2020 ◽  
Vol 10 (22) ◽  
pp. 8255
Author(s):  
Ala Hijazi ◽  
Sameer Al-Dahidi ◽  
Safwan Altarazi
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Residual Strength ◽  
Prediction Accuracy ◽  
Training Dataset ◽  
Multiple Site ◽  
Input Output ◽  
Accuracy Improvement ◽  
Multiple Site Damage ◽  
Data Points

An artificial neural network (ANN) extracts knowledge from a training dataset and uses this acquired knowledge to forecast outputs for any new set of inputs. When the input/output relations are complex and highly non-linear, the ANN needs a relatively large training dataset (hundreds of data points) to capture these relations adequately. This paper introduces a novel assisted-ANN modeling approach that enables the development of ANNs using small datasets, while maintaining high prediction accuracy. This approach uses parameters that are obtained using the known input/output relations (partial or full relations). These so called assistance parameters are included as ANN inputs in addition to the traditional direct independent inputs. The proposed assisted approach is applied for predicting the residual strength of panels with multiple site damage (MSD) cracks. Different assistance levels (four levels) and different training dataset sizes (from 75 down to 22 data points) are investigated, and the results are compared to the traditional approach. The results show that the assisted approach helps in achieving high predictions’ accuracy (<3% average error). The relative accuracy improvement is higher (up to 46%) for ANN learning algorithms that give lower prediction accuracy. Also, the relative accuracy improvement becomes more significant (up to 38%) for smaller dataset sizes.

scholarly journals Prediction of Bead Geometry with Changing Welding Speed Using Artificial Neural Network

Materials ◽  
2021 ◽  
Vol 14 (6) ◽  
pp. 1494
Author(s):  
Ran Li ◽  
Manshu Dong ◽  
Hongming Gao
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Welding Speed ◽  
Gas Metal Arc Welding ◽  
Welding Process ◽  
Training Dataset ◽  
Welding Parameters ◽  
Bead Geometry ◽  
Bead Width ◽  
Artificial Neural

Bead size and shape are important considerations for industry design and quality detection. It is hard to deduce an appropriate mathematical model for predicting the bead geometry in a continually changing welding process due to the complex interrelationship between different welding parameters and the actual bead. In this paper, an artificial neural network model for predicting the bead geometry with changing welding speed was developed. The experiment was performed by a welding robot in gas metal arc welding process. The welding speed was stochastically changed during the welding process. By transient response tests, it was indicated that the changing welding speed had a spatial influence on bead geometry, which ranged from 10 mm backward to 22 mm forward with certain welding parameters. For this study, the input parameters of model were the spatial welding speed sequence, and the output parameters were bead width and reinforcement. The bead geometry was recognized by polynomial fitting of the profile coordinates, as measured by a structured laser light sensor. The results showed that the model with the structure of 33-6-2 had achieved high accuracy in both the training dataset and test dataset, which were 99% and 96%, respectively.

Modified Swift Criteria for Residual Strength of Multiple Site Damage Structure

2009 ◽  
Vol 417-418 ◽  
pp. 881-884 ◽  
Author(s):  
Jian Yu Zhang ◽  
Rui Bao ◽  
Bin Jun Fei
Keyword(s):  
Residual Strength ◽  
Yield Criterion ◽  
Fatigue Cracks ◽  
Tensile Load ◽  
Maximum Load ◽  
Alloy Sheet ◽  
Multiple Cracks ◽  
Multiple Site ◽  
Multiple Site Damage ◽  
Aluminum Panels

As more aircrafts reach or exceed their design life, it is becoming very important to research multiple cracks damage, especially the multiple site damage (MSD) in order to re-evaluate their service life and damage tolerance/durability performance. The existing of MSD may remarkably reduce the residual strength of an aerospace structural component than those with a singe lead crack. This study investigated the residual strength of aluminum alloy sheet with MSD through three types of aluminum specimens test. Aluminum panels with bare collinear constant diameter holes were chosen as specimens. After some constant amplitude tension-tension load cycles, the MSD were found in these specimens since there were multiple fatigue cracks emanating from the saw cuts of holes. The residual strength was recorded as the maximum load when every specimen was subjected to monotonically increasing tensile load until failure occurred. In different failure prediction criteria that were often used in engineering in order to evaluate the accuracy of these criteria, Swift criterion (ligament yield) criterion got more accurate prediction results than other criteria. Although Swift criterion was more accurate than some other criteria, its error was still big for some specimens. Two modified approaches were proposed in order to get more accurate and appropriate failure criterion for MSD structure.

scholarly journals Application of Wavelet Neural Network in Building Settlement Prediction

2020 ◽  
Vol 198 ◽  
pp. 03014
Author(s):  
Ruijie Zhang
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Prediction Accuracy ◽  
Building Design ◽  
Wavelet Neural Network ◽  
Deformation Monitoring ◽  
Polynomial Method ◽  
Observation Data ◽  
Bp Artificial Neural Network ◽  
Artificial Neural

Deformation monitoring, as a key link of information construction, runs through the entire process of the building design period, construction period and operation period[1]. At present, more mature static prediction methods include hyperbolic method, power polynomial method and Asaoka method. But these methods have many problems and shortcomings. In this paper, based on the characteristics of building foundation settlement and the methods widely discussed in this field, a wavelet neural network model with self-learning, self-organization and good nonlinear approximation ability is applied to the prediction problem of building settlement[2]. Using comparative analysis and induction method. The 20-phase monitoring data representing the deformation monitoring points of different settlement states of the line tunnel, using the observation data sequence of the first 15 phases respectively to take the cumulative settlement and interval settlement as training samples, through the BP artificial neural network and the improved wavelet neural network, for the last five periods Predict the observed settlement.Through the comparison, it is found that whether the interval settlement or the cumulative settlement is used, the prediction results of the wavelet neural network are basically better than the prediction results of the BP artificial neural network, and the number of trainings is greatly reduced. The adaptive prediction of the wavelet neural network. The ability is particularly obvious, and the prediction accuracy is significantly improved. Therefore, it can be shown that the wavelet neural network is indeed used in the settlement monitoring and forecast of buildings, which can obtain higher prediction accuracy and better prediction effect, and is a prediction method with great development potential.

ARTIFICIAL NEURAL NETWORK (ANN)-BASED PREDICTIVE TOOL FOR ESTIMATING LIGHTNING DAMAGE IN COMPOSITES

2021 ◽  
Author(s):  
DEVIN NIELSEN ◽  
TYLER LOTT ◽  
SOM DUTTA ◽  
JUHYEONG LEE
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Network Architecture ◽  
Back Propagation ◽  
Training Dataset ◽  
Predictive Tool ◽  
Ann Models ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Hidden Layer

In this study, three artificial neural network (ANN) models are developed with back propagation (BP) optimization algorithms to predict various lightning damage modes in carbon/epoxy laminates. The proposed ANN models use three input variables associated with lightning waveform parameters (i.e., the peak current amplitude, rising time, and decaying time) to predict fiber damage, matrix damage, and through-thickness damage in the composites. The data used for training and testing the networks was actual lightning damage data collected from peer-reviewed published literature. Various BP training algorithms and network architecture configurations (i.e., data splitting, the number of neurons in a hidden layer, and the number of hidden layers) have been tested to improve the performance of the neural networks. Among the various BP algorithms considered, the Bayesian regularization back propagation (BRBP) showed the overall best performance in lightning damage prediction. When using the BRBP algorithm, as expected, the greater the fraction of the collected data that is allocated to the training dataset, the better the network is trained. In addition, the optimal ANN architecture was found to have a single hidden layer with 20 neurons. The ANN models proposed in this work may prove useful in preliminary assessments of lightning damage and reduce the number of expensive experimental lightning tests.

scholarly journals Residual Strength Prediction of Aluminum Panels with Multiple Site Damage Using Artificial Neural Networks

Materials ◽  
2020 ◽  
Vol 13 (22) ◽  
pp. 5216 ◽  
Author(s):  
Ala Hijazi ◽  
Sameer Al-Dahidi ◽  
Safwan Altarazi
Keyword(s):  
Neural Networks ◽  
Artificial Neural Networks ◽  
Residual Strength ◽  
Lap Joints ◽  
Multiple Site ◽  
Ann Model ◽  
Model Predictions ◽  
Multiple Site Damage ◽  
Artificial Neural ◽  
Aluminum Panels

Multiple site damage (MSD) cracks are small fatigue cracks that may accumulate at the sides of highly loaded holes in aging aircraft structures. The presence of MSD cracks can drastically reduce the residual strength of fuselage panels. In this paper, artificial neural networks (ANN) modeling is used for predicting the residual strength of aluminum panels with MSD cracks. Experimental data that include 147 unique configurations of aluminum panels with MSD cracks are used. The experimental dataset includes three different aluminum alloys (2024-T3, 2524-T3, and 7075-T6), four different test panel configurations (unstiffened, stiffened, stiffened with a broken middle stiffener, and bolted lap-joints), many different panel widths and thicknesses, and the sizes of the lead and MSD cracks. The results presented in this paper demonstrate that a single ANN model can predict the residual strength for all materials and configurations with high accuracy. Specifically, the overall mean absolute error for the ANN model predictions is 3.82%. Furthermore, the ANN model residual strength predictions are compared to those obtained using the most accurate semi-analytical and computational approaches from the literature. The ANN model predictions are found to be at the same accuracy level of these approaches, and they even outperform the other approaches for many configurations.

A Simple Heat Transfer Correlation for Three Inlet Configurations Using Artificial Neural Network in the Complex Transition Flow Regime

2003 ◽  
Author(s):  
A. J. Ghajar ◽  
L. M. Tam ◽  
S. C. Tam
Keyword(s):  
Neural Network ◽  
Heat Transfer ◽  
Artificial Neural Network ◽  
Least Squares Method ◽  
Transition Flow ◽  
Transfer Coefficients ◽  
Transition Flow Regime ◽  
Bell Mouth ◽  
Data Points ◽  
Artificial Neural

Local forced and mixed heat transfer coefficients were measured by Ghajar and Tam (1994) along a stainless steel horizontal circular tube fitted with reentrant, square-edged, and bell-mouth inlets under uniform wall heat flux condition. For the experiments the Reynolds, Prandtl, and Grashof numbers varied from about 280 to 49000, 4 to 158, and 1000 to 2.5×105, respectively. The heat transfer transition regions were established by observing the change in the heat transfer behavior. The data in the transition region were correlated by using the traditional least squares method. The correlation predicted the transitional data with an average absolute deviation of about 8%. However, 30% of the data were predicted with 10 to 20% deviation. The reason is due to the abrupt change in the heat transfer characteristic and its intermittent behavior. Since the value of heat transfer coefficient has a direct impact on the size of the heat exchanger, a more accurate correlation has been developed using the artificial neural network (ANN). A total of 1290 data points (441 for reentrant, 416 for square-edged, and 433 for bell mouth) were used. The accuracy of the new correlation is excellent with the majority of the data points predicted with less than 10% deviation.

Sign in / Sign up

Export Citation Format

Share Document