Methodology Research Considering Both Location And Dispersion Effects In Multiresponse Optimization

This thesis concerns on the evaluation between the location and dispersion effects in multiresponse optimization when the correlation of multiple responses is considered,which can help designers find the optimal settings for their specific problems in the design of product and process.The literature review is conducted at the beginning to illustrate the current research status in balancing the location and dispersion effects of multiple responses.Then,we propose a loss function-based approach to tackle the relative importance of the location and dispersion effects simultaneously in the optimization process,while considering the responses' correlations at the same time.Instead of fixing the relationship between the location and dispersion effects,our approach works out more compromise solutions versus their relative importance,which allows for a flexible choice when designers balancing the location and dispersion effects for multiple responses.Examples are used to furtherly illustrate the advantages of the proposed approach The advantages can be concluded as followings:(i)the correlation of multiple responses are considered in the optimization.(ii)the mean and standard deviation of responses are directly used to reflect their location and dispersion effects respectively.(iii)the relative importance between the location and dispersion effects is reflected in an effective and flexible way.(iv)two loss indexes are defined to offer useful information for designers when balancing the location and dispersion effects.