A Method Using Clustering and SVDD for Quality Detection

With the development of industry 4.0, intelligent detection methods for product quality have received widespread attention. Traditional quality control methods are usually based on statistical process control. Process data must meet the requirements of independent and identical distribution, which limits its application in industry. According to the characteristics of current product quality data of manufacturing industry, a product quality detection method based on clustering hypersphere model was proposed. First, the data set is divided into
$k$
subsets by
$k$
-means clustering. Then describe the data of each subset separately to obtain closed hyperspheres containing most normal data, and use their radius as the control limit. Finally, through the comprehensive detection of the test samples in each hypersphere, the quality of the product is identified. By selecting the
$k$
value, a more flexible detection boundary can be obtained and the control limit is more reasonable. It is an effective method to ensure product quality stability and realize intelligent manufacturing for the manufacturing industry.