Decision-Making System Based on a Fuzzy Hierarchical Analysis Process and an Artificial Neural Network for Flow Shop Machine Scheduling Model Under Uncertainty

Luis Fernando Villanueva-Jimenez,Valentin Calzada-Ledesma,Juan De Anda-Suarez,Jose Antonio Vazquez-Lopez,Javier Yanez-Mendiola

doi:10.1109/access.2021.3099342

Luis Fernando Villanueva-Jimenez, Valentin Calzada-Ledesma + Show 3 more

Open Access

https://doi.org/10.1109/access.2021.3099342

Copy DOI

Abstract

The management of the uncertainty existing in any production system is fundamental to define machine scheduling models that allow programming production instances attached to the real world. In this research, a generalized decision-making system is developed for the management of uncertainty existing in flow shop machine scheduling models. The system assessment the uncertainty existing in internal and external factors that influence the decision-making process of production programming experts, and that is decisive in a final machine scheduling. The system is based on the combination of the Fuzzy Hierarchical Analysis Process, a membership analysis, and an Artificial Neural Network (ANN). The system allows to concentrate the experience of experts in machine scheduling and generalize their knowledge. The efficiency of the system is verified with a Fuzzy Hierarchical Analysis Process Model, the “ANN toolbox” preloaded in MATLAB and variety of structures of an Artificial Neural Network. The results are validated in an industrial application and the system is contrasted against an expert. The results show the efficiency of the system as it defines and predicts the final machine scheduling of production instances; the joint assessment of variables that add uncertainty to the production system allowed to reduce delays in product deliveries.

Highlights

Machine scheduling is responsible for organizing, choosing and scheduling the efficient use of resources in such a way that products or services are produced within a reasonable agreement with customer demand
In Mexico, most companies in the "leather-footwear" manufacturing sector are characterized by having a flow shop machine scheduling model, where their main needs are related to the properties of completion times which seek to avoid late delivery of orders
A dominant trend has been identified in the formulation of stochastic and deterministic solution models and procedures, to focus on the machine scheduling models [3]; these processes do not consider the inherent uncertainty of production systems

Summary

INTRODUCTION

Machine scheduling is responsible for organizing, choosing and scheduling the efficient use of resources in such a way that products or services are produced within a reasonable agreement with customer demand. In Mexico, most companies in the "leather-footwear" manufacturing sector are characterized by having a flow shop machine scheduling model, where their main needs are related to the properties of completion times which seek to avoid late delivery of orders In this sector, much of the production programming is based on the experience of experts who have to deal with a complex decision-making process involving external and internal alterations that affect effective machine scheduling; these alterations are considered as the uncertainty existing in the production system. The complexity involved in machines scheduling, in a real-world context, involves the development of algorithms that converge to optimal solutions, and the development of systems that support the complex decision-making process experienced by the production programming expert This has motivated the development of this research.

CONCEPTUAL BASIS

STAGE I

STAGE II

STAGE IV

RESULTS

DISCUSSION AND CONCLUSIONS