Evaluatıon Of The Rısks In Metal Sector And Inspectıon Decısıon Support System

Thesis Type: Doctorate

Institution Of The Thesis: Gazi University, Turkey

Approval Date: 2019

Thesis Language: Turkish

Student: Cemal Can Ayanoğlu

Supervisor: MUSTAFA KURT


Occupational health and safety studies performed at modern-day workplaces are based on the identification of hazards and taking preventive measures with regard to the risks which may occur. Therefore, evaluation of the risks with effective methods has become as a necessary requirement in order to identify the priorities for taking measures. In metal sector, which is one of the basic constituents of our economy, serious occupational accidents may be observed due to the unforeseen risks. Thus, assessment of hazards at metal sector workplaces and evaluation of the risks by the effective methods will be an important achievement. The purpose of this thesis is to develop a risk evaluation method based on objective criteria that would be able to determine overlooked risks for metal sector and to emerge inspection decision support system that will enable the efficient usage of the resources. In this context, by evaluation of the reports of occupational accidents occurred at metal sector workplaces, an accident data has been prepared. After applying reductions and inferences over data through multivariate statistical analysis methods, the final data set has been obtained with 192 accident instances and 39 variables. Subsequently, in the basis of the artificial neutral networks with machine learning algorithm, which shows the best performance in the data set, accident prediction model has been created. Analysis of the model has showed an accuracy of 90%. Afterwards, forecasting model has been integrated with Fine Kinney method in order to create the quantitative risk evaluation procedure with objective decision criteria. Also, inspection decision support system has been offered in the basis of the developed forecasting model. Finally, developed methods have been applied at sample workplace data. The results obtained from application support the purposes of the study.