Diagram Kendali Fuzzy Multiatribut

Wibawati, Wibawati (2020) Diagram Kendali Fuzzy Multiatribut. Doctoral thesis, Institut Teknologi Sepuluh Nopember.

Text 06211460010005_Disertation.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Diagram kendali merupakan teknik yang secara luas digunakan untuk pengendalian proses dalam bidang manufaktur. Hasil pengukuran karakteristik kualitas di bidang manufaktur tidak selalu dinyatakan dalam angka. Skala linguistik banyak digunakan dalam menggambarkan kualitas suatu produk, seperti baik ataupun cacat. Pada monitoring kualitas dengan skala linguistik sering melibatkan unsur subjektifitas dari inspektor. Hal ini mengakibatkan adanya ambiguitas terhadap hasil monitoring. Untuk mengatasi permasalahan tersebut, dikembangkan diagram kendali Modified Fuzzy Multinomial (MFM) serta diagram kendali atribut berbasis distribusi Bivariat Poisson yang disebut diagram kendali Fuzzy Bivariat Poisson (FBP). MFM merupakan pengembangan dari diagram kendali Fuzzy Multinomial (FM) yang melibatkan jumlah karakteristk kualitas dalam penentuan batas kendali serta dilakukan dengan berbagai transformasi fuzzy dan berbagai skenario fungsi keanggotaan. Diagram kendali FBP dikembangkan berdasarkan estimasi parameter fuzzy dengan berbagai skenario nilai korelasi antar karakteristik kualitas (ρ) dan keketatan pemeriksaan (γ_level ). Evaluasi kinerja kedua diagram kendali dilakukan melalui studi simulasi berdasarkan nilai Average Run Length (ARL). ARL diagram kendali MFM dibandingkan dengan diagram kendali Fuzzy Multinomial (FM), dan diagram kendali FBP dibandingkan dengan diagram kendali Bivariat Poisson (BP) konvensional. Berdasarkan nilai ARL, diperoleh bahwa diagram kendali MFM lebih sensitif dibandingkan dengan diagram kendali FM. Diagram kendali FBP lebih sensitif dalam mendeteksi pengamatan out-of-control dibandingkan dengan diagram kendali BP konvensional. Kinerja diagram kendali FBP semakin baik seiring dengan meningkatnya korelasi antar karakteristik kualitas (ρ) dan keketatan pemeriksaan (γ_level ). Kinerja terbaik diagram FBP dicapai untuk ρ dan tingkat γ_level sebesar 0,8. Penerapan diagram MFM pada studi kasus produksi botol minuman tipe “RC” menunjukkan bahwa untuk keempat transformasi fuzzy proses produksi belum terkendali secara statistik. Keunggulan diagram FBP dibandingkan diagram BP konvensional juga diperlihatkan dalam kasus pemonitoran proses produksi botol minuman tipe “NT”.

================================================================================================================================================

A control chart is a widely used technique to control processes in manufacturing. In manufacturing field, some quality characteristic are measured in categorical (linguistic) scale instead of numerical scale. The linguistic scale is also often used to judge the quality of a product, either good or defective. The quality monitoring on a linguistic scale often influenced by the subjectivity of the inspector. It caused ambiguity in the monitoring results. In order to solve this problem, we proposed new atributte control charts called Modified Fuzzy Multinomial (MFM) and Fuzzy Bivariate Poisson (FBP). MFM is a development of a Fuzzy Multinomial (FM) control chart that involves the number of quality characteristics in determining control limits and is carried out with various fuzzy transformations and various membership function scenarios. The FBP control chart is developed based on the estimation of fuzzy parameters with various scenarios of correlation values between quality characteristics (ρ) and tightness of inspection (γ_level ). The performance evaluation of the two control charts is done through a simulation study based on the Average Run Length (ARL) value. ARL MFM control chart is compared with Fuzzy Multinomial (FM) control chart, and FBP control chart is compared with Bivariate Poisson (BP) control chart. Based on the ARL value, the findings show that the MFM control chart is more sensitive than the FM control chart and FBP control chart is more sensitive in detecting out-of-control observations compared to conventional BP control chart. The sensitivity of the FBP chart improves with an increasing correlation between the quality characteristics (ρ). The best performance of the FBP chart is achieved for the γ_level and ρ equal to 0.8. Moreover, an illustration example of MFM chart shows that the production process of the beverage bottle type “RC” by using four fuzzy transformation has not been controlled statistically. On other hand, an excellent performance of FBP chart compared to conventional BP chart is also evident in the monitoring of the beverage bottle type “NT” production process.

Item Type:	Thesis (Doctoral)
Uncontrolled Keywords:	diagram kendali atribut, modified fuzzy Multinomial, fuzzy bivariat Poisson, γ_level, ARL. atributte control chart, modified fuzzy Multinomial,fuzzy bivariat Poisson, γ_level,ARL.
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Mathematics and Science > Statistics > 49001-(S3) PhD Thesis
Depositing User:	Wibawati Wibawati
Date Deposited:	24 Aug 2020 02:21
Last Modified:	24 Aug 2020 02:21
URI:	http://repository.its.ac.id/id/eprint/79385

Actions (login required)

View Item