Rearch On Auto Body Manufacturing Quality Inspection Based On Statistical Process Control

An auto body is the frame of the vehicle. The quality of auto body manufacturingwill influence quality and functionality of the vehicle. The report of D.J.Power’s globalautomotive product quality investigation and assessment of key issues （Initial QualitySurvey） in1997shows that40%of automotive product quality problems are related withthe quality of auto body manufacturing. So the quality of auto body manufacturing hasattracted a great attention of industry and academia. Here, we study the quality of autobody manufacturing through the view of statistical process control.Firstly, according to statistical process control theory, there are abnormal variationand normal variation in the manufacturing process. Quality control charts and processcapability indices are usually used to monitor the abnormal variation and normalvariation. But they can not effectively monitor detection points of auto body, somean-variance single control charts the multi-process performance analysis chart（MPPAC） has been proposed.Secondly, traditional X-R chart are used to control the dimensional variation ofauto body. However, these charts that need two control charts are obviously not easy toread consuming and not sensitive to detect small changes in manufacturing process. Sothe four mean-variance single control charts of WL, WLC, T, EWMA-T are introducedand also are compared with X-R control chart. The result shows the fourmean-variance single control charts perform better than X-R control chart. Finally,the application of the mean-variance single control charts in auto body manufacturingquality control demonstrates that they not only reduce the number of control charts, savethe time of drawing control charts and improve the readability of the control charts, butalso alarm more time when the manufacturing process out of control, reducing the bodyloss of manufacturing quality.Lastly, because multi-process performance analysis charts ignore sampling errors,consequently the result will be unreliable and the continuous quality improvement willbe influenced. Hence, we propose a modified multi-process performance analysischart（MMPPAC）, which is devoleped by process capability indice C_pmk and generalizedinference approach.C_pmk in MMPPAC is surperior over C_pk in MPPAC, and thegeneralized inference approach in MMPPAC converts the estimated index values to thelower confidence bounds, so that the MMPPAC can perform more reliablely. Finally, Acase study of auto body manufacturing quality control demonstrates is presented todemonstrate MMPPAC performs more reliabely than MPPAC.