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

Ala Hijazi,Sameer Al-Dahidi,Safwan Altarazi

doi:10.3390/app10228255

Ala Hijazi, Sameer Al-Dahidi + Show 1 more

Open Access

https://doi.org/10.3390/app10228255

Copy DOI

Journal: Applied Sciences	Publication Date: Nov 20, 2020
Citations: 10	License type: CC BY 4.0

Affiliation: German Jordanian University

Abstract

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.

Highlights

Data-driven methods are increasingly being used in a wide variety of scientific fields.These methods extract knowledge and insights from datasets, which are typically large, and use this acquired knowledge to forecast new outputs
The main objective of this study is to introduce a novel assisted-Artificial neural network (ANN) modeling approach that can further improve the accuracy of the ANN predictions, and at the same time this approach enables the ANN to hold accurate even when relatively small datasets are used for training the network
A novel assisted-ANN modeling is proposed and it is used for estimating the residual strength of aluminum panels with multiple site damage (MSD) cracks

Summary

Introduction

Data-driven methods are increasingly being used in a wide variety of scientific fields These methods extract knowledge and insights from datasets, which are typically large, and use this acquired knowledge to forecast new outputs. A variety of analytical and computational approaches can theoretically be used for estimating the residual strength of panels with MSD cracks. These approaches include analytical, semi-analytical (empirically corrected), computational and data-driven, and the accuracy of these approaches varies significantly [27]. According to LEFM, failure (or unstable crack extension) will occur when the value of the crack-tip SIF reaches a critical value This limiting value of the SIF is called the fracture toughness (KC ). For a panel with a major crack and collinear adjacent MSD cracks subjected to tension, as illustrated in Figure 1, the Mode-I stress intensity factor (KI ) for the lead crack is:

Objectives

Results

Conclusion