Denny Suci Prastiya
Bakti Indonesia University

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Time Prediction Using Artificial Neural Networks for the Permutation Flow Shop Problem: Predict-Then-Optimize Framework Denny Suci Prastiya
Performa: Media Ilmiah Teknik Industri Vol. 25 No. 2 (2026): Performa: Media Ilmiah Teknik Industri
Publisher : Industrial Engineering Study Program, Faculty of Engineering, Universitas Sebelas Maret (UNS)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.20961/performa.v25i2.3246

Abstract

This study introduces the implementation of predict-then-optimize (PTO) as a revolution in operational research (OR) in the industrial engineering field. In this contribution, we proposed an artificial neural network (ANN) model with a 4-20-4 multi-layer perceptron (MLP) using training and learning gradient descent with momentum under the backpropagation process as a prediction framework. Meanwhile, for the optimized model, we proposed mixed integer linear programming (MILP) for the permutation flow shop scheduling problem PFSP case using the NEH algorithm. Subsequently, we tested momentum, gradient, and learning rate parameters to validate the best performance of MSE in the ANN model. The best solution with the highest accuracy is used to simulate a scheduling prediction. Moreover, in the scheduling case, we compare the general flow shop FIFO model based on existing conditions with the proposed PFSP model, using the results of ANNs that consist of 4 machines and 4 jobs, based on the ANNs' prediction simulation. Furthermore, we proposed a PFSP model that can re-sequence jobs using the NEH algorithm with aims for minimize makespan. The results show that in the PTO framework, adaptive prediction and optimal conditions can achieve the best decision solution. Ultimately, this study successfully proves, in both practical and theoretical aspects, that the two biggest topics in the current era, that is, machine learning (ML) and operational research (OR), can work together under the PTO framework model.