Two approaches for estimation of production time: robust regression analysis and neural network

Lisjak, Dragutin and Ćosić, Predrag and Milčić, Diana (2011) Two approaches for estimation of production time: robust regression analysis and neural network. = Two approaches for estimation of production time: robust regression analysis and neural network. In: International Conference on Flexible Automation and Intelligent Manufacturing (21 ; 2011), 26.-29.06.2011., Taichung, Taiwan.

Full text not available from this repository.
Official URL: https://www.bib.irb.hr/538480

Abstract

A robust regression analysis model as a possible approach to time/cost estimation is used for the estimation of requested results based on the previous stochastic results and experiments. The requests for classification consideration of the product shape and process sequencing are important conditions for designing a general model for the estimation of production times. In fact, it means development of a technological knowledge base. As a result of our analysis, we created eight regression equations with the obtained index of determination, with the most important independent variables different for 2D and 3D model respectively. The observed level of subjectivity, constraints and errors were the reasons to use neural networks as the second approach to estimate production times. According to the presented results, we can conclude that the assumption on the use of a neural network for the production time estimation in relation to a robust regression analysis model is justified. For all experimental models the applied backpropagation neural network gives better values of key performance indexes (R, R2, RMSE, NRMSE).

Item Type: Conference or Workshop Item (Lecture)
Keywords (Croatian): production time, robust regression analysis, neural network
Subjects: TECHNICAL SCIENCE > Mechanical Engineering
Divisions: 700 Department of Industrial Engineering > 710 Chair of Production Design
700 Department of Industrial Engineering > 720 Chair of Production Control
Indexed in Web of Science: No
Indexed in Current Contents: No
Date Deposited: 28 Jun 2016 14:53
Last Modified: 08 Feb 2018 14:02
URI: http://repozitorij.fsb.hr/id/eprint/5985

Actions (login required)

View Item View Item

Nema podataka za dohvacanje citata