A hybrid conceptual cost estimating model using ANN and GA for power plant projects

Providing an accurate completion cost estimate helps managers in deciding whether to undertake the project due to cash in hand. Hence, MAPNA Group Co. as an Iranian leading general contractor of power plant projects is not an exception too. Cost prediction in these projects is of great importance, whereas it can assist managers to keep their overall budget under control. Literature has been reviewed and influencing variables are explored. Thereafter, an artificial neural network model is developed and combined with genetic algorithm to select the best network architecture. According to the literature reviewed, almost all of the performed studies have selected the optimum network architecture through a process of trial and error, which makes the present method worthy of implementation. The best network architecture is capable of predicting projects' cost of accuracy equal to 94.71%. A sensitivity analysis is then performed to test the significance degree of model input variables.