Key Quality Feature Identification and Quality Prediction in Complex Manufacturing Processes

There are many factors involved in the complex manufacturing production process (CMP). It is necessary to identify key processes and features in the CMP. A CE-ND-based algorithm is proposed to provide an effective data-based scheme to identify key features in CMP. Nonparametric Copula Entropy (CE) is applied to estimate the correlation between features and establish a correlation relational network; network deconvolution (ND) is applied to eliminate the influence of transfer effects in the relational network. Abundant experiments in the analysis of industrial production process data are conducted, which indicate that our proposed method outperforms other methods.