Purpose – The purpose of this paper is to establish a combined model for selecting key indexes of complex equipment, and then improve the cost forecasting precision of the model. The problem how to choose the key elements of complex products has always been concerned on many fields, such as cost assessment, investment decision making, etc. Design/methodology/approach – Using Grey System Theory to establish a cost estimation model of complicated equipment is more reasonable under the few data and poor information. Therefore, this paper constructs cost index’s system of complex equipment, and then quantitative and qualitative analysis methods are utilized to calculate the grey entropy between the characteristic parameter and the behavior parameters. Further, establish the grey relational clustering matrix of the behavior sequences by using the grey relative incidence analysis. Finally, the authors select key indicators according to the grey degree. Findings – The experiment demonstrates that the cost key parameters of complex equipment can be successfully screened out by the proposed approach, and the cost estimation accuracy of complicated products is improved. Practical implications – The method proposed in this paper could be utilized to solve some practical problems, particularly the selection of cost critical parameters for complex products with few samples and poor information. Taking the cost key indexes of civil aircraft as an example, the results verified the validity of the GICM model. Originality/value – In this paper, the authors develop the method of GICM model. Taking the data of civil aircraft as an example, the authors screen the key indicators of complex products successfully, and improve the prediction accuracy of the GM (1, N) model by using the selected parameters, which provides a reference for some firms.

Purpose The purpose of this paper is to study a new approach by combining a multilayer perceptron neural network (MLPNN) algorithm with a GM(1, N) model in order to estimate the development cost of a new type of aircraft. Design/methodology/approach First, data about developing costs and their influencing factors were collected for several types of Boeing and Airbus aircraft. Second, a GM(1, N) model was constructed to simulate development costs for a civil aircraft. Then, an MLPNN algorithm was added to optimize and revise the simulative and forecasting values. Finally, a combined approach, using both a GM(1, N) model and an MLPNN algorithm was adopted to forecast development costs for new civil aircraft. Findings The results show that the proposed approach could do the work of cost estimation for new types of aircraft. Rather than using a single model, the combined approach could improve simulative and forecasting accuracy. Practical implications Scientific cost estimation could improve management efficiency and promote the success of a new type of civil aircraft development. Considering that China’s civil aircraft research and development is at its very beginning stages, only very limited data could be collected. The development costs for civil aircraft are affected by a series of factors. The approach outlined by this paper could be applied to development cost estimations in China’s civil aircraft industry. Originality/value The paper has succeeded by constructing a cost estimation index system and proposing a novel combined cost estimation approach comprised of a GM(1, N) model and an MLPNN. It has undoubtedly contributed to improving the accuracy of cost estimations.