Early cost estimation of strip-steel coiler using BP neural network

Traditionally, cost estimation was performed after the design process, though most opportunities of cost reduction have already passed. It is advantageous to be able to estimate the product costs early in the product development cycle. In this paper, previous work in product cost estimation is discussed first, and an overview of the BP neural network approach for early cost estimation is given. Based on cost-related features, the cost estimation model using a backpropagation neural network is proposed. A case study of strip-steel coiler is discussed. The cost-related features of the product design are extracted and quantified according to their cost effects. The construction, training and validation of the neural network are also described. This makes it possible to estimate the product´s cost without a full knowledge of the manufacturing and assembly process plans, and encourage designers to design according to cost.