Pornthip Tabkosai.. A web-based decision support system for cost identification using deep learning in plastic injection industry. Doctoral Degree(Industrial Engineering). King Mongkut's University of Technology North Bangkok. Central Library. : King Mongkut's University of Technology North Bangkok, 2023.
A web-based decision support system for cost identification using deep learning in plastic injection industry
Abstract:
The plastic injection moulding industry has been growing and expanding rapidly.
Cost identification is important to business operations. The complexity of the plastic
parts and manufacturing data resulted in a long data waiting time and inaccurate cost
identification. The objective of this research is to develop a Web-Based Decision
Support System (WB-DSS) for the prediction cost identification that compares a hybrid
deep learning model between a Convolutional Neural Network (CNN) and Artificial
Neural Network (ANN) with Tunicate Swarm Algorithm (TSA) and Artificial Neural
Network (ANN) for bulk price analysis of multi-price for multi-volume to enhance the
efficiency in cost identification of complex geometry parts in the plastic injection
industry. CNN-ANN can improve the accuracy from a dense voxel of CNN that can
disentangle the difficulty of cost identification. The methodology consists of
3D-voxelization adopted to CNN and the feature-based of complex geometry parts to
feature parameters using the learning ability of ANN to achieve better accuracy.
TSA-ANN can achieve a faster convergence rate for optimal solutions and higher
accuracy. The methodology entails ANN, which applies feature-based extraction of
complex geometry parts to develop a cost identification approach. TSA is used to
construct the initial weight into the learning model of ANN, which can generate faster-
to-convergent optimal solutions. Then, the best model has been developed for bulk
price analysis of multi-price for multi-volumes. The result shows that the new hybrid
deep learning CNN-ANN has better accuracy than TSA-ANN and is applied in bulk
price analysis. These results can predict cost evaluation at about 98.65% accuracy for
parts costs, 95.17% accuracy for mould costs, and 96.83% accuracy for a bulk price.
The contribution of this research is based upon a new hybrid deep learning using CNNN
with ANN on WB-DSS that is practical and accurate in performing cost identification
bulk price for decision-making in the plastic injection industry which is unavailable in
the literature.