Tekstil boyahane işletmelerinde makine öğrenmesi ile fire tahmini
Loading...
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Trakya Üniversitesi Fen Bilimleri Enstitüsü
Access Rights
info:eu-repo/semantics/openAccess
Abstract
Rekabet şartlarının uluslararası boyutta olduğu günümüz şartlarında işletmelerin pazar paylarını koruyabilmeleri için maliyetlerini azaltmaları gerekmektedir. Bu sebeple işletmelerin müşteri siparişlerini yeterli seviyede üretmeleri zorunlu hale gelmiştir. Ayrıca üretim sırasında gerçekleşen imalat hataları da kalite kontrol işlemi sırasında tespit edilmektedir. Bu ürünler fire olarak kabul edilmektedir. Hem sipariş fazlası üretimler hem de imalat hatalı ürünler daha düşük fiyatlara satılmaktadır. Bu durum, işletmelerin zarar etmelerine ve fazladan stok alanlarının işgal edilmesine sebep olmaktadır. Tekstil boyahane işletmeleri emek yoğun ve kimyasal ürünlerin tüketiminin fazla olduğu bir sektördür. Fazla üretim için harcanan enerji ve su kaynağı ekolojik sorunlara da yol açmaktadır. Gerçekleştirilen bu tez çalışmasında, bir işletmenin geçmiş üretim verileri incelenerek sipariş için gerekli en az üretim miktarının tespiti yapılmıştır. Bu tespit yapılırken makine öğrenmesi algoritmalarından faydalanılmıştır. Bu algoritmalardan sınıflandırma temelli çalışan özellikle J48, Naive Bayes, KNN, SVM ve Multilayer Perceptron algoritmaları WEKA uygulamasında kullanılmıştır. Algoritmalardan çıkan sonuçlar karşılaştırılarak en başarılı algoritma tespit edilmiştir.
In today's conditions, where competition conditions are international, businesses need to reduce their costs in order to maintain their market share. For this reason, it has become mandatory for businesses to produce customer orders at a sufficient level. In addition, manufacturing defects that occur during production are also detected during the quality control process. These products are considered as waste. Both over-order productions and manufacturing-defective products are sold at lower prices. This situation causes businesses to make losses and occupy extra stock areas. Textile dyeing establishments are a labor-intensive industry with a high consumption of chemical products. The energy and water resources spent for overproduction also cause ecological problems. In this thesis study, the past production data of an enterprise was examined and the minimum production amount required for the order was determined. While making this determination, machine learning algorithms were used. Among these algorithms, J48, Naive Bayes, KNN, SVM and Multilayer Perceptron algorithms, which are classificationbased, have been used in WEKA application. The results of the algorithms were compared and the most successful algorithm was determined.
In today's conditions, where competition conditions are international, businesses need to reduce their costs in order to maintain their market share. For this reason, it has become mandatory for businesses to produce customer orders at a sufficient level. In addition, manufacturing defects that occur during production are also detected during the quality control process. These products are considered as waste. Both over-order productions and manufacturing-defective products are sold at lower prices. This situation causes businesses to make losses and occupy extra stock areas. Textile dyeing establishments are a labor-intensive industry with a high consumption of chemical products. The energy and water resources spent for overproduction also cause ecological problems. In this thesis study, the past production data of an enterprise was examined and the minimum production amount required for the order was determined. While making this determination, machine learning algorithms were used. Among these algorithms, J48, Naive Bayes, KNN, SVM and Multilayer Perceptron algorithms, which are classificationbased, have been used in WEKA application. The results of the algorithms were compared and the most successful algorithm was determined.
Description
Keywords
Makine öğrenmesi, Tekstil, Boyahane, Kalite kontrol, Fire tahmini, Machine learning, Textile, Dyeing, Quality control, Wastage prediction
