Research On The SPC Control Chart Application In Many Varieties And Small Batches Production

SPC as a key technology of quality controlling plays an important role in the production process in a company. With the development of economic globalization and the intensification of market competition, the competition among enterprises is growing more and more intense and the quality of products has been on unprecedented attention. In recent years, market demands are gradually various, so the production mode of many varieties and small batches production is becoming the mainstream production mode of society. In the environment of many varieties and small batches production, there is a problem of insufficient data. So, in the environment of many varieties and small batches production, research on the way to establish SPC control charts is very necessary requirements is the focus of this study. The main research content is as the following:(1) Through cluster analysis method, the quality characteristics of the data are classified in order to better establish quality control chart.In a semiconductor package process, as it involves many varieties and small batches, how to carry out quality controlling of the thrust of gold ball bonding this feature, is the key and difficult problem in the whole process of quality controlling. In this study, two clustering methods of hierarchical clustering and K-means clustering, for the analysis and classification of 27 kinds of products, effectively solve the problem of insufficient data. Finally, based on these two clustering methods, X-bar and S control chart is designed for thrust of such products. The control charts prove that the two cluster analysis methods are effective.(2) Through one-way ANOVA, the quality data is classified in order to better establish quality control chart.In gold ball bonding process of a semiconductor package, the thrust is quality controlling. In this study, F and t test in one-way ANOVA, for the analysis and classification of 27 kinds of products, effectively solve the problem of insufficient data. Finally, based on these two tests, X-bar and S control chart is designed for thrust of such products. The control charts prove that the one-way ANOVA method is effective.(3) Through PCA method, the problem of missing part in the forging process is under consideration. And control charts prove the effectiveness of the designed classifier.In the multi-stage forging process of small batches of parts, the parts missing is an important issue of quality controlling. Due to the parts lacking, machine is in more complex conditions. It is necessary to design a control chart to classify these conditions. This study is following:segmenting the collecting data in the method of combining time domain and frequency domain; using PCA method for data conversion in order to reduce the dimension of data; studying the characteristic of conversion data and determining the rules of classification; according to the principle of minimum classification error, designing the classifier; estimating the performance of the classifier to take advantage of classification error probabilities determined; at last, designing a multi-step classification of X-bar control chart. The actual data of the enterprise verifies the above-mentioned study, and the results show performance classification good, classification results correct.Finally, in the summary and generalization of the full paper, further research is proposed.