Predicting the Yield Rate of DRAM Modules by Support Vector Regression

Abstract

Dynamic random access memory (DRAM) module is one of the principal components of electronic equipment, which impacts the quality, performance and price of the final products singinifcantly. Typically, DRAM module is composed of DRAM ICs (integrated circuit). DRAM ICs with higher quality can be used to produce DRAM modules with higher quality. Generlly speaking, high quality DRAM ICs are more costly. Due the the cost down and material saving reason, some DRAM module manufacturers purchase batches DRAM ICs containing defective units, and then have the batch tested in order to select DRAM ICs for production of DRAM modules. Thus, this kind of DRAM module is suitable only for products not intended for work in harsh environments being sold in lower price markets. Due to the lower quality of the DRAM ICs, the actual quality of the DRAM module is not easily predicted. Predicting the yield rate of the DRAM module is thus an important issue for DRAM module manufacturers who purchase DRAM ICs with lower quality at lower prices. This study used support vector regression (SVR) to predict the yield rate of the DRAM modules produced using defective DRAM ICs. SVR is a very capable method and has been successfully applied across many fields. However, the parameters and input features differ depending on the application. Thus, a scatter search (SS) approach is proposed to obtain the suitable parameters for the SVR and to select the beneficial subset of features which result in a better prediction of the DRAM module yield rate. The experimental results showed that the performance is better than that of traditional stepwise regression analysis.