Computer Aided Error Diagnosis On Multi-stage Manufacturing Process

In 1920s, the concept of Quality Control was put forward. Since then, it has been developed and improved along with practices of corporation management, and has become an independent subject. Among all methods of Quality Control, Statistical Process Control (SPC) is most widely used in current enterprises, which has improved the level of corporation management considerably. Through sample collecting and statistical analysis of key quality parameters and key working procedures, with the aim of improving quality level, SPC can analyze producing process data related to quality of products.Commonly used SPC can only statistically analyze important indexes of certain working procedures one by one, for example, drawing Shewhart charts for every one. Faced with systems of multi-variables, multi-stages, traditional SPC is not competent. For quality control in multi-stage process, it is just necessary to carry out the study on analytical and evaluating standards in multi-variable, multi-stage manufacturing process. Reducing fluctuating and finding out fluctuating resources in manufacturing process is the base of conducting quality control and improvement in multi-variable, multi-stage process.Multi-variable quality control is a method of controlling several quality characters simultaneously. Despite of the existing numerous multi-variable control theories, multi-variable quality control charts are still preferred because of their trenchancy and intuitionistic characters. T~2 chart designed by H.Hotelling will be applied here to accomplish quality control of multi-variable process. Principle Component Analysis, as an important multi-variable statistical method, is used here to transform lots of indexes into less comprehensive indexes, while facing the problem of complicated sample data and reticula relationship among them.The T~2 -PCA method is put forward by combining T~2 char and PCA, which is used for the single parameter charts and principle component charts, assisting diagnosis the fault exiting in the stage.In a multi-stage process, the final variation is accumulated from previous stages. Modeling and controlling faults spread is essential for improvement of production quality. Two-quality Diagnosis Theory is applied two distinguish faults resource.A computer-aided quality control system is designed to deal with data from multi-variable, multi-stage process and output results.