Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Maden Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2018
Öğrenci: MERVE MEVSİM
Danışman: CELAL KARPUZ
Özet:In Turkey, underground coal mine accidents are commonly faced problems and cause loss of lives and money. In order to prevent these accidents, the hazards and risks specific to each mine should be assessed by both ordering the risk scores and building a statistical model for the future trend of accidents. In this study, past accident records of Üzülmez District of Turkish Hard Coal Enterprises is utilized. The risk scoring model is built by using Riskex Risk Score Calculator based on Fine-Kinney risk score equation. In accident forecasting, multiple linear regression and time series analysis techniques are utilized by the aid of the Minitab 17 software. The data taken from Turkish Hard Coal Enterprises the Department of Labor Health, Safety and Education includes accident type, location, work shift, job, affected body part, age and experience for each recorded accident. Firstly, the scores of the risk of those seven category is found by utilizing Fine-Kinney method and the risks of each accident type, accident location, shift, job, body part affected after having an accident, age group and experience duration are ordered in order to find the relative seriousness of them. Secondly, expected number of accidents in future is forecasted with the aid of multiple linear regression and time series analysis. By performing multiple linear regression it is aimed to observe how the variables like raw coal production, total gallery advance, total number of workers, explosive consumption, and timber consumption effect number of accidents. By time series analysis, the number of accidents expected in future is determined which is thought to provide benefits to occupational health and safety managers in monitoring the performance of safety precautions. Time series model is found to be more reliable and practicable than multiple linear regression which uses just the monthly number of accidents as random variable. Research findings revealed that moving average time series model is the best fit model when the evaluation is made in terms of calculated accuracy values. However, quadratic trend model should be preferred when long term forecast horizon is needed.