Background: In most of the deep-cavity shell plastic products injection moulds, such as test tubes, syringes, and other medical devices, the slender core structure is essential. The uneven pressure of the polymer melt acting on the core during the filling process easily causes core deformation, resulting in core shift, the wall thickness inequality of the parts, and a series of problems, which then affect the final shape and mechanical properties of the plastic parts. The type of deep cavity component is prone to core shift defects during injection molding, resulting in uneven wall thickness of the syringe sleeve and inability to achieve perfect fit with the syringe core rod. Reducing the core shift of the component is the key issue of this article. Background: In most of the deep-cavity shell plastic products injection moulds, such as test tubes, syringes, and other medical devices, the slender core structure is essential. The uneven pressure of the polymer melt acting on the core during the filling process easily causes core deformation, resulting in core shift, the wall thickness inequality of the parts, and a series of problems, which then affect the final shape and mechanical properties of the plastic parts. The type of deep cavity component is prone to core shift defects during injection molding, resulting in uneven wall thickness of the syringe sleeve and inability to achieve perfect fit with the syringe core rod. Reducing the core shift of the component is the key issue of this article. Objective: This study aims to combine injection moulding CAE with neural networks and optimization algorithms to reduce core shift, improve molding quality and dimensional accuracy, extend mold life, establish a regression model for predicting core shift, shorten product development time, and improve production efficiency. Methods: Propose an injection moulding process parameter optimization method based on Adaptive Neural Network Fuzzy Inference System (ANFIS) and Rireworks Algorithm (FWA). Taking a certain medical syringe as the research object, the Taguchi experiment was designed with mold temperature, melt temperature, injection time, injection pressure, packing pressure, and packing time as optimization variables. A simulation was conducted using Moldflow software, and the signal-tonoise ratio was used to analyze the significance and trend of the influence of process parameters on the plastic core shift. A predictive regression model is constructed based on ANFIS to express the nonlinear function relationship between process parameters and core shift. FWA is used to conduct global optimization based on the ANFIS model to determine the optimal combination of injection molding process parameters. Results: Through simulation verification, the errors between the FIS prediction model and the ANFIS prediction model were compared with the actual simulation values. The ANFIS prediction model had the highest accuracy, and the core shift of plastic parts based on the ANFIS-FWA combination optimization method decreased by 27.6% compared to before optimization. It was effective in improving the molding quality of plastic parts. result: Through simulation verification, the errors between the FIS prediction model and the ANFIS prediction model were compared with the actual simulation values. The ANFIS prediction model had the highest accuracy, and the core shift of plastic parts based on the ANFIS-FWA combination optimization method decreased by 27.6% compared to before optimization. It was effective to improve the molding quality of plastic parts. Conclusion: The core shift prediction model established based on ANFIS can accurately predict the core shift of the product, greatly reducing the lengthy simulation time of the computer and shortening the product development cycle. The ANFIS-FWA coupling method can find the best combination of process parameters among multiple injection molding process parameters, achieve the improvement of core shift defects, and provide new ideas for optimizing the core shift of the same type of product.
Read full abstract
Dec 1, 2024
Enno Fischer 
Open Access
Just Published
