Identification and grouping the risks of construction industrial units by Delphi method and self-organized artificial neural network (Case study: Construction of traverse factory)

One of the main pillars of employment and business development in each country are the industrial units of that country. Success in building industrial units depends on mechanisms and other factors such as identifying and assessing their risk. Therefore, the model presented in the research is based on a neural network to classify and evaluate the risks of building a traverse plant with a view to technology localization. To identify the risks by Delphi method with the help of experts and university professors in 3 levels of 40 risks, 21 risks were selected. Then, according to the opinions of experts, professors and executives of the construction of the traverse factory, risk factors on the probability of occurrence and severity of risk impact were scored from 0 to 10. The risks were then grouped using Mohlonen's self-organized neural network in the MATLAB software program. The results of the neural network output show that risk factors are classified into 5 groups, the most important of which is the lack of similar experience of the contractor, which is the highest level of radar importance. A noteworthy point in this study is the importance of the risks in group 4, of which 17 risks (over 80% of the risks studied) are in this group. The risks in group 4 are between 60 to 80% in terms of importance, which can be greatly reduced by planning and studying this issue in the construction of new industrial units in the country.