Optimization Of MKPCA Model And Design Of Monitoring Software For Injection Process

Injection Molding Machine (IMM) is one of the most important equipment in plastic industry. The deviation between the process variables tracks and the set time tracks, differences of raw materials purity and size and so forth will make quality of product in batches show a certain inconsistency. If large deviation occurs without eliminated, one batch or more batches will have decline in quality. So, monitor of batch process to check out the abnormal can help operators to take remedial measures to eliminate the error or stop the batches production in order to reduce the waste of raw materials and the occupation for equipment.Multivariate statistical methods for injection process have two parts:multiway kernel principal component analysis(MKPCA) and multiway principal component analysis(MPCA). First, collect many batches of data in normal condition, then use which to set up model in normal condition. Monitor test data in the use of the two models separately. Comparing the results of the two models, it is apparent to say that MKPCA is more excellent than MPCA for nonlinear process monitor. But MKPCA also has its disadvantages such as adapt normal data without fault data, adapt empirical way to determine parameters and types of kernel function.In view of the shortcomings of MKPCA above, this paper proposes optimized MKPCA to monitor injection process. In this method, make the type of kernel function, parameters and the number of principal components as decision-making variables. The goal of optimization is to ensure the greatest accuracy of online monitor. Use the optimized MKPCA model to monitor online process and make a comparison between MKPCA and MPCA. The optimized MKPCA can monitor the occurrence of fault with high sensitivity.Design the software of injection process monitor that is programmed in C# and database is Oracle 9i. Design the structure and specific function of this software based on the principle of reliability, openness and scalability. The main functions of this software are document management, model creation and import, model management, monitoring model set, show of monitoring results and statistical function and so on. This software has advantages such as user-friendly, integrity of function, easy operation. Practical operation shows that this software is stable and reliable.