The Research Of Robust And Tolerance Design In The Injection Molding Based On Gaussian Process

Along with the development of scientific society,the requirement of product are tend to more preciseness,higher percent of pass and lower cost, artificially technology of production technology turns to intelligence technology gradually. Recently, computer simulation technology is highly applied in quality prediction and the adjustment of process parameters before real injection.However, many noise parameters affect the performance of the product besides process parameters, and robust design which can reduce those factors’s influence to improve the percent of pass becomes more important. On the other hand, the cost of product also becomes one of the most important factors the enterprises considered, the tolerance of process parameters not only impact the product but also the cost of system. In allusion to the point, this dissertation takes Gaussian Process as an surrogate model, which affords the theory and method direction for realizing the robust and tolerance design between process parameters. First of all,an adaptive canonical correlation analysis method based on Gaussian Process is accepted, and the method takes two data sheets of technological parameters and part’s quality indexes as object of study, canonical vectors are extracted from the two data sheets meanwhile so as to realize the Gaussian association analysis between process parameters and design objectives. Then according to the correlation degree, the process parameters can be classified as major parameters and minor parameters, which simplifies the multiple parameter model computational complexity reasonably and improves the design efficiency.Secondly, a robust optimization method is developed by those major parameters, while fluctuation quantity is added into both objective function and constraint conditions, so that the robust combination of process parameters can be obtained by optimizing both mean and variance. And it is proved that the proposed method is very effective on the model precision and finding robust solutions according to a mathematics example.Moreover, the relationship between process parameters tolerance limit and quality loss cost, manufacturing cost and rejection cost is analyzed to develop the loss of quality mathematical expectation model, and then set the optimum setting of tolerance of the control factors by lambert W function to guarantee the total loss is minimum, which can reduce the cost.At last,based on Gaussian Process model, the robust and tolerance design approach is employed to optimize processing parameters. As an application, the injection molding process of an auto guide groove assembly part is investigated to study the process parameters and short shot defect. The result indicates that, compared with the traditional deterministic optimization, the proposed method can avoided unstable quality indicators due to ignore the process parameters fluctuation; on the other hand, calculating the tolerance reasonably can also reduce the cost causing by optimal process parameters.The result indicates that the proposed method can effectively improve the product qualified rate and decrease the cost.