Abstract: By extracting the feature parameters and setting up the relation model between technical parameters and structure parameters with artificial neural network, the new feature parameters were mapped. Then this new feature parameters were regarded as the attribution of the new problem, Lance and William distance were chose as similarity measure to realize the similar case search, so the physical structure similarity searching was turned into parameter vector similarity searching and the new feature parameters were also used to modify the similar case obtained. The working principle, feature parameter extracting, ANN model and Lance and William similarity measure were discussed. The workflow and modules of the prototype for bucket elevator deign were also presented to test the feasibility of the new approach, and this research can provide technical support for rapid and intelligent design of complex products.