Multivariate Quality Control Chart Based On Component Data

With the rapid development of market economy,the competition between enterprises is becoming more and more intense,if enterprises want to develop faster and better,they need to ensure the quality of products and the quality of products is the most important factor for the success of enterprises.Therefore,enterprises need to adopt certain technical means to effectively monitor the production process of products to ensure the high quality and reduction of products.Statistical Process Control(SPC)is a kind of control technology to ensure the stability of the product quality process.It monitors every stage of the production process by using control chart.If abnormal process occurs,the control chart sends an alarm signal immediately,diagnoses the abnormal process,eliminates abnormal factors in the process,and ensures that the process is in a controlled state.In the actual production enterprises,the criteria for evaluating product quality are not only limited to one feature,such as the configuration process of a cosmetic water.The factors affecting the product quality are the content of different components.Therefore,in view of this multi-dimensional data,multidimensional evaluation is needed,and multi-quality control charts are used to monitor multiple quality characteristics.But the common multivariate quality control charts are used to monitor multiple features of common data,but when there are certain limitations on these quality features,that is,component data,it is obviously unreasonable to use them directly to monitor component data.In order to monitor the deviation of the covariance matrix,which is a special structure of component data,this paper first presents a special result processing method to deal with the limitations of component data: transformation and interpolation of component data,three common transformation methods to transform component data into common European spatial data,and interpolation method based on simplex spatial component data.Secondly,four common multivariate quality control charts are compared to monitor the sensitivity of common data,and simulation experiments are carried out to visualize the monitoring results.On this basis,a multivariate quality control chart based on component data is proposed.The component data is processed and converted into common data in European space.The control chart is monitored according to the monitoring steps.A new control chart is proposed to monitor the covariance matrix offset of component data,namely,the principal component analysis control chart based on component data,simulated experiments and monitored results.Visual display shows that this new method is better than other multielement quality control charts,and easy to operate,saving monitoring time and cost.Finally,the real data of cosmetics configuration process is applied to the control chart proposed in this paper for example analysis,and the monitoring results are visually displayed.