Abstract: A cost estimation model based on cost cluster is presented with big samples in early product design phase because precise cost estimation in early adptive or derivative product design phase is very difficult. Euclidean distance between samples and estimating product are calculated to express their similarity. Difference matrix, distance matrix and similarity matrix are built base on euclidean distance sets between samples and indeterminate estimation product, then most correlated sample sets are selected to get interval of estimated cost based on cost cluster analysis. An example of brake disc shows invalidity of the cost cluster model. The result of experiment proved that cost cluster model was able to separate relative order of qulity of the samples.

Key words: Cost cluster, Cost interval, Euclidean distance, Gray fuzzy cluster