Multi-Objective Optimisation of Injection Moulding Process for Dashboard Using Genetic Algorithm and Type-2 Fuzzy Neural Network

Abstract

This study examines the use of injection moulding to evaluate mechanical properties in plastic products, such as shear and residual stresses. Key process variables like melt temperature, mould temperature, hold pressure duration, and pure hold duration are meticulously chosen for study. A full factorial experiment design is utilised to determine the best settings. These variables notably influence the end product’s physical and mechanical properties. Computational techniques, like the finite element method, are used to analyse behaviours based on varied input parameters. A CAD model of a dashboard part is incorporated into a finite element analysis to measure shear and residual stresses. Four specific parameters from the injection moulding process are subjected to an in-depth experimental design. It is worth noting that the injection moulding process does not incorporate a type-2 fuzzy neural network (T2FNN). However, in this particular investigation, T2FNN was employed to replicate the mechanical stress model associated with dashboard injection moulding. Its purpose was to estimate shear and residual stress levels. Additionally, the multi-objective genetic algorithm (MOGA) was utilised to extract the most optimal parameters for the injection moulding process, aiming to minimise shear and residual stress and thereby increase the resistance of the final product. The proposed model was developed and implemented using MATLAB software. A Pareto front was derived from the MOGA by employing the T2FNN within the process, identifying fourteen optimal solutions.