Manufacturing process multivariate quality diagnostic classifier based on improved information entropy

Abstract

The invention provides a manufacturing process multivariate quality diagnostic classifier based on improved information entropy. The manufacturing process multivariate quality diagnostic classifier isused for collecting original data of quality characteristics in the manufacturing process, carrying out data preprocessing, applying a hybrid algorithms to perform process analysis on multivariate quality characteristics of key processes, discriminating stability and whether an abnormal phenomenon occurs or not according to data recorded in a control chart, adopting an information entropy methodto find out a process abnormity source, and introducing an adjustment factor, a chi-square value, a weight ratio between the two and an Euclidean distance stability discrimination rule so that the classification result is enabled to be more accurate. The manufacturing process multivariate quality diagnostic classifier is strict in process capability coefficient conditions, accurate in determination state, low in algorithm complexity and fast in processing time, combines multivariate quality, misjudgement factors and major constituent factors, is higher in applicability, standard in parameter processing and complete in data processing, reduces the misjudgement probability, solves the problems of data offset and unit non-uniformity, has higher accuracy degree than that of a support vector machine, and can realize the abnormity diagnosis technology.

Description

Technical field [0001] The invention relates to the technical field of quality diagnosis in the processing and manufacturing process of mechanical products, and in particular to a multivariate quality diagnosis classifier for manufacturing processes that improves information entropy. Background technique [0002] Modern manufacturing processes are highly correlated with multiple variables. Process monitoring of this type of production process is called multiple quality control (MQC) or multiple statistical process control (MSPC). The process of finding the cause of out-of-control is called MSPC diagnosis or abnormal identification. There are mainly two types of methods: one is statistical decomposition technology; the other is technology based on machine learning. Mainstream decomposition techniques include principal component analysis (PCA), feature space comparison method, MTY method, step-down method, and multi-directional kernel principal component analysis method. However,...

AI Technical Summary

Problems solved by technology

However, these methods usually involve complex statistical procedures, which are not conducive to the application

Images