Research On HCI Recognition Technology Of BIW Welding Error

The vehicle body manufacturing level not only reflects the level of a country’s industrial manufacturing, but also directly affect the quality of the vehicle. BIW is one of the most important parts of the vehicle, its assembly quality directly determines the vehicle’s many properties, such as noise, sealing and power and so on. Therefore, the accurate diagnosis of car body in white assembly and welding errors that occurred during the rapid is essential to improve the quality of automotive products. However, in the current practical production process, the combining between body assembly processes and theory that be used for assembly errors diagnosis is inadequate. The disjointed makes diagnosis relies heavily on subjective judgments of field engineers, and reliability of results of the diagnostic is not high.This issue stems from the actual project, and be part of the industry’s well-known project "2mm Project". The purpose is to be able to effectively solve the problem of auto-chip error identification BIW. Paper based on statistical analysis algorithms such as cluster analysis、principal component analysis and combined with the process of the body assembly processes and assembly tree structure make manual intervention. This method successfully achieved human-computer interaction in BIW assembly errors identified. Maximize the combination of theory, methodology and experience. improve the efficiency and accuracy of the diagnosis BIW chip error.Aiming at the characteristics of BIW welding error and based on cluster analysis theory paper,designed the clustering algorithm for regional recognition errors and using this algorithm diagnosis BIW welding error. And elaborated interactive identify technical based cluster analysis using diagnosis of a car BIW welding error. Proposed method for removing spurious correlation points that appear in the clustering process and analyze the reasons for these points generated. Combine the attributes and characteristics of the error-chip design characteristics involves clustering algorithm, put forward innovatively use of main relevant points to find the error source. Using principal component analysis to find the error area that have multiple error direction. At last using this method successfully diagnosis of a commercial vehicle in the back and roof error used with clustering analysis theory.