Research And Application Of SVDD-CRL Control Chart Based On Single Class Multidimensional Data Under The Background Of Industrial Internet Of Things

With the advent of Industry 4.0 and the rapid development of Industrial Internet of Things,the manufacturing industry gradually tends to be intelligent and automatic.The development and innovation of Internet of Things and technology has opened up more application fields.On this basis,people's living standards continue to improve,the requirements of product quality are also gradually improved,the importance of product quality control is becoming increasingly prominent.In order to ensure the quality of products and reduce production losses,the production process of products needs process monitoring.Statistical process control(SPC)is the most commonly tool for process monitoring.Traditional control charts,such as unitary process control chart and multivariate process control chart,have been widely used in system process monitoring,but their monitoring performance is limited in many aspects,such as data distribution,data dimension and data sample size.Under the background of the continuous development of Industrial Internet of Things,the production technology and product quality of enterprises have been different in the past.The product quality characteristics have changed from a single variable to a comprehensive characteristic of high dimension,unknown data distribution and bilateral sample differences.Although the existing control chart technology has been studied in various feature areas,the current control chart needs to be able to monitor the above feature data at the same time.Support vector data description(SVDD)brings a new idea to control chart.SVDD is a single class classification algorithm based on statistical learning theory and support vector machine.It can not only be applied to the situation of unknown sample distribution and high-dimensional data,but also support the training of single class data.In this paper,the characteristics of the data of enterprise production process monitoring are analyzed in detail under the background of Industrial Internet of Things.Based on these data characteristics,combined with SVDD algorithm and CRL control chart,SVDD-CRL control chart is proposed.Through Montakaro simulation method,the monitoring effect under different distribution,variable dimension and training sample size was simulated,and SVDD-CRL control chart was applied to white wine production data.After performance evaluation with k-chart,it is proved that the proposed control chart has better monitoring performance.