Research On Auto Body Welding Assembly Dimension Deviation Inspection Based On MEWMA Chart

With the development of the automotive industry and the continuousimprovement of people’s needs, people have become increasingly demanding on thebody quality. Body welding assembly quality is an important factor which affects thequality of the body. The topic of this paper comes from the Natural ScienceFoundation of China(71271078). Aiming at the features of many points to bemeasured, great workload of testing and dimensional shift is slow and small, thethesis has studied monitoring technique for dimension deviation of body weldingassembly based on multivariate exponentially weighted moving average control chart,and has proposed a new multivariate exponentially weighted moving average controlchart which is more suitable for actual production process. The main research workand innovative research results are as follows:（1）Analyzed three computing methods of average run length which is theperformance evaluation index of unconventional control chart based on MEWMAcontrol chart. According to the analysis results of the statistic performance ofHotelling’s T2chart, MEWMA chart and MCUSUM chart, we know the MCUSUMchart and MEWMA chart are more sensitive in detecting small deviation.（2）Aiming at the correlation of deviation parameter δ, upper control limit Hand weight parameter r, their relation model and the relational expression of δ andoptimum r are established. Based on the qulity data of body welding assembly, theoptimized MEWMA chart enjoys better sensibility to small procedure shift comparedwith Hotelling’s T2control chart and MEWMA chart with other values of r.（3）Aiming at the strong restrain among elements in smoothing parametermatrix of MEWMA chart and the special choice of smoothing parameter matrix, thisthesis loosens the restrain among elements in smoothing parameter matrix andintroduces a new kind of MEWMA control chart-GEWMA control chart, which iscompared with MEWMA chart and FEWMA chart to do the performance analysis.The results show that GEWMA control chart is more sensitive than MEWMA chartand FEWMA chart while monitoring small deviation.（4）Aiming at choosing the complexity and irregularity of smoothing parametermatrix of GEWMA chart, the method of using Particle Swarm Optimization to do theoptimization selection is raised. According to the comparison among the MEWMAchart, FEWMA chart and GEWMA chart based on the same shift in mean vector, it is found that GEWMA chart has better performance on small deviation. According tosimulating and actual example of welding assembly process, it is found that applyingGEWMA control chart is faster to detect the shift of producing process. Finally, underthe condition that the parameters are estimated and the existence of small deviationbetween estimated parameters and actual parameters, the thesis gives sensitivityanalysis of GEWMA chart for small deviation. The result shows that the GEWMAchart has good stability for small deviation.