Predicting equipment failure time and downtime using data mining techniques in Khuzestan Steel Company furnaces

نوع: Type: Thesis

مقطع: Segment: masters

عنوان: Title: Predicting equipment failure time and downtime using data mining techniques in Khuzestan Steel Company furnaces

ارائه دهنده: Provider: Akram Shaabani nia

اساتید راهنما: Supervisors: Dr.Hamidreza Dezfoulian

اساتید مشاور: Advisory Professors:

اساتید ممتحن یا داور: Examining professors or referees: Dr.Nafiseh Soleimani-Dr.Vahid KhodaKarami

زمان و تاریخ ارائه: Time and date of presentation: 2025

مکان ارائه: Place of presentation: 62

چکیده: Abstract: One of the most important assets of any organization, especially in industries, is the existence of abundant recorded data. Organizations can achieve very valuable goals by analyzing data. One of the most important tools for analyzing this data is the science of data mining. The goal of data mining is to obtain useful knowledge and extract hidden patterns in existing data. In the steel industry, by predicting failures in a timely manner, sudden failures can be prevented and unnecessary production stops can be minimized. In this study, the recorded stoppage data for the years 1397 to 1401 of six steelmaking furnaces of the Khuzestan Steel Company was examined. Appropriate features were selected and after identifying the data, data preprocessing was performed. For data preprocessing, outliers and inconsistent records were removed and new features were created by data transformation. In the next step, the features that affected the stoppage rate were selected. The features that were identified as the most important factors were selected for modeling. The modeling stage was carried out in two parts: classification and clustering. In the classification stage, the algorithms SVM, D.T, KNN, ANN Adaboost, Bagging, which are the most widely used algorithms in the field of prediction, were used. To evaluate the models, accuracy, sensitivity and precision criteria were used, and the best techniques were determined with the target feature of the stopping time KNN, SVM and Adaboost and the stopping distance feature Adaboost, Ann, Svm. In the clustering section, using selected features from the three algorithms Kimmins, Twostep and Cohen, the data were divided into clusters with similar features and the appropriate number of clusters in each method was determined according to the silhouette index. With this technique, equipment can be clustered based on similarities.

فایل: ّFile: Download فایل