Development cost estimation of civil aircraft based on combination model of GM (1, N) and MLP neural network

Abstract

Scientific prediction and estimation for the development cost of civil aircraft, not only conducive to strengthening the control of cost, is also the key to ensure the success of the project. Considering the complex influence factors of civil aircraft cost with the scarce sample data, a combination model is adopted. Firstly, constructing a multi factor GM(1,N) model to predict the development cost of civil aircraft based on the collection of cost affecting characteristic sequence. Secondly, MLP neural network algorithm is used to optimize and revise the forecasting cost. Making full use of the grey GM(1,N) model with few data and the effective use of simulation advantages of MLP neural network. Finally, a number of domestic and foreign civil aircrafts as an example to verify the combination model, the results show that the combination forecasting method has satisfactory and stable prediction accuracy, and it can effectively be used to estimate the civil aircraft development cost.