关键词: 状态空间模型, 混联多级制造过程, 关键质量特性, 最小二乘回归方法

Abstract: To improve the existing methods of identifying the key quality characteristics in multistage manufacturing process, an identification method for key quality characteristics in multi-stage manufacturing process was put forward by combining the Partial Least Squares Regression (PLSR) method with the state space model. According to the features of serial-parallel multistage manufacturing processes, the state space method was introduced to construct quality relationship model of serial-parallel multistage manufacturing process, and the PLSR method was used to solve the multi-collinearity problem in the quality characteristics, conduct model analysis and identify the key quality characteristics. The cigarette production process was taken as an example to introduce the application of the method. The result showed that this method not only could identify the key quality characteristics in multistage manufacturing process, but also established the interrelationship model of influence of output quality of various process on the final product quality, which reflected the structures of the multistage manufacturing process and causal relationships among quality characteristics at all process levels and provided the basis for quality analysis and control in multistage manufacturing process.

Key words: state space model, serial-parallel multistage manufacturing process, key quality characteristics, partial least squares regression