Development of Machine Learning Technology for Product Design Space

Abstract

In design of make-to-order products, fast estimate of design parameters corresponding to customer requests through interactive communication between customer and designers is required. To realize fast estimate design, we are constructing analysis led customer co-design system, which can evaluate the product performance and specification in short time, utilizing a machine learning technology. As a vehicle product of customer co-design system, we try to apply for estimate design of centrifugal compressor. Engineer considers estimate design of product specification corresponding to customer demands. We developed a specification predicting technology to provide a design specification faster time for engineers. The design specification predicting program shows the result of rotation speed and number of impeller in 32 seconds.