Research On Improvement Of Taguchi Method Based On Data Envelopment Theory

With the upgrading of industry and the increasingly fierce market competition,how to control the quality of products has become the focus of enterprises,and Taguchi method is a simple and practical method to solve the problem of product quality.The core idea of Taguchi method is to solve the quality problem through parametric optimization design rather than directly inspecting the product quality.Taguchi method applies signal-to-noise ratio,orthogonal experimental design and loss function to parameter design,which saves a lot of time and cost.After years of theoretical research and practical application,researchers summed up a series of advantages and disadvantages of the conventional Taguchi method,which pointed out that the general formula of signal-to-noise ratio has shortcomings in dealing with three different characteristic response parameters,resulting in the lack of some basic information of the response parameters after processing.In this paper,a new multi-response parameter fusion model is proposed,in which the response parameters with different quality characteristics are fused into a composite response variable to improve the existing Taguchi method.Based on the quality characteristics of response parameters,a new standardized processing formula for response parameters was established to replace the original signal-to-noise ratio formula.On the basis of the standardized response parameters,the less the variability,the better is the principle,the variation coefficient is used to analyze the main cause,and the response parameter fusion results of different quality characteristics under different parameter level combinations are obtained respectively.The ratio analysis of the fusion results of response parameters with different quality characteristics was carried out,and the optimization of the horizontal combination of parameters was carried out according to the principle that the smaller the ratio,the better.Among them,excel will be used as the test analysis and optimization tool in the whole test analysis process to obtain the optimal parameter level combination and main response parameters.The proposed response parameter fusion model is an improvement on the basis of data envelopment analysis.Different from data envelopment analysis which can only deal with LB data and SB data,this model can deal with all the response parameters of quality characteristics,that is,LB parameter,SB response parameter and NB response parameter.The optimal level combination of design parameters and optimal level combination of machining parameters can be obtained by applying the model to product design and machining process.After the improved Taguchi method and other multi-response Taguchi methods were applied to parameter optimization respectively,the comparison of the results showed that the main cause analysis result of the improved parameter optimization was more comprehensive and effective than the original one.By means of parameter optimization based on compound response variables,the essence of quality can be further recognized and the application of quality design in product design and processing can be easily understood in this paper.