The Line Feeding Problem (LFP) involves the delivery of components to the production area. Previous models minimise the delivery costs and optimally assign each component to a line feeding mode between line stocking, kitting, and sequencing but cannot provide easily comprehensible guidelines. We use the Classification And Regression Tree (CART) algorithm to develop, in a supervised way, a decision tree based on problems that are solved with a Mixed Integer Programming (MIP) model for training purposes. Based on selected attributes of the components and the manufacturing environment, the decision tree suggests a line feeding mode for every component. For a synthetically determined training and evaluation data set, we find that the classification tree can predict the line feeding mode with an average classification accuracy of 78.49%. After the decision tree is implemented and a line feeding mode is selected for each component, an infeasible solution might occur. We develop a repair approach that solves this problem with an average cost deviation from the optimal solution of 0.38%.

A supervised machine learning approach for the optimisation of the assembly line feeding mode selection

Zangaro F.
Writing – Review & Editing
;
Battini D.
Conceptualization
2020

Abstract

The Line Feeding Problem (LFP) involves the delivery of components to the production area. Previous models minimise the delivery costs and optimally assign each component to a line feeding mode between line stocking, kitting, and sequencing but cannot provide easily comprehensible guidelines. We use the Classification And Regression Tree (CART) algorithm to develop, in a supervised way, a decision tree based on problems that are solved with a Mixed Integer Programming (MIP) model for training purposes. Based on selected attributes of the components and the manufacturing environment, the decision tree suggests a line feeding mode for every component. For a synthetically determined training and evaluation data set, we find that the classification tree can predict the line feeding mode with an average classification accuracy of 78.49%. After the decision tree is implemented and a line feeding mode is selected for each component, an infeasible solution might occur. We develop a repair approach that solves this problem with an average cost deviation from the optimal solution of 0.38%.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11577/3379387
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 6
  • ???jsp.display-item.citation.isi??? 6
social impact