A hybrid approach based on locally linear neuro-fuzzy modeling and TOPSIS to determine the quality grade of gas well-drilling projects

Evaluation of a project and its contractors has considerable importance in gas well-drilling projects due to their high investments and worth. In this paper, the quality of some gas well-drilling projects is analyzed in order to evaluate and grade project tasks. A neuro-fuzzy network is utilized to learn the grading process and generate models. To select among these models, a ranking method, namely technique for order of preference by similarity to ideal solution (TOPSIS) is employed. During seven gas well-drilling projects, 77 tasks are studied based on quality practitioners׳ points of view. After generating the primary models, three indices namely, root mean square error (RMSE), mean absolute percentage error (MAPE), and a newly introduced Q-index are selected to prioritize 31 models in optimistic, pessimistic, and average modes.