Study On Process Quality Diagnosis Technology In Multi-Type&Small-batch Condition

This paper is based on the major prior theme project of Zhejiang Province "key technology research of digital design and manufacture of service-oriented architecture and application in discrete manufacturing enterprises". In multi-type and small-batch condition, less quality data sample can be obtained and process quality is difficult to diagnosis. For solving these problems and optimizing quality diagnosis method, study on quality diagnosis technology was developed. The study is concerned about the following aspects:1. Considering the lack of the collected quality data sample and complexity of process quality data, the mixed patterns recognition of control chart based on wavelet analysis and SVM was suggested. At the same time, in order to improve the effect of control chart patterns recognition, PSO was introduced into SVM. The three layers control chart pattern recognition model framework and the basic process was designed. Though constructing reasonable simulation sample for training and testing, the effectiveness of the model was proved.2. In order to provide reference for the quality process adjustment, the parameters estimation method of control chart patterns was suggested, and parmeters estimation model was designed to estimate four parameters of three abnormal patterns. The feasibility of the model was showed by using the simulation.3. Though learning from the knowledge base and explain mechanism of expert system, relevant database was structured, and unusual reason diagnosis model was based on PSO-SVM was designed. At the same time, transformational rule between the model data and content that can be identified by user was designed.4. Based on the analysis of the present situation in SJ company, the model of quality control chart patterns recognition, parameters estimation and abnormal reason diagnosis was applied to diagnose and control quality of SJ company. The model was proved feasibility.