Multiple objectives optimization of injection-moulding process for dashboard using soft computing and particle swarm...

Area of Science:

• Manufacturing Engineering
• Materials Science
• Computational Science

Background:

• Injection moulding is a key process for plastic product manufacturing.
• Product defects such as sink marks, shrinkage, and warpage can significantly impact quality.
• Optimizing process parameters is crucial for defect reduction and performance enhancement.

Purpose of the Study:

• To investigate the impact of key injection moulding parameters on plastic product defects.
• To develop and validate a computational model for predicting and mitigating these defects.
• To identify optimal process settings for minimizing defects and improving product quality.

Main Methods:

• Utilized a full factorial design of experiments to analyze process parameters (cooling time, mould temperature, melt temperature, pressure holding time).
• Employed soft computing methods, including finite element (FE) analysis, integrated with CAD models for defect quantification.
• Explored various machine learning models (Decision Tree, MLP, LSTM, GRU) for process modelling and defect prediction.
• Applied Multi-Objective Particle Swarm Optimization (MOPSO) to extract optimal process parameters.

Main Results:

• The study successfully quantified shrinkage, warpage, and sink marks using FE simulations.
• Machine learning models demonstrated effectiveness in predicting injection moulding defects.
• MOPSO identified 18 optimal parameter sets, presented via a Pareto Front, balancing multiple objectives.

Conclusions:

• The integrated approach of FE simulation, machine learning, and MOPSO effectively addresses plastic product defects in injection moulding.
• Optimal process parameters derived from this method can significantly improve product quality and reduce manufacturing waste.
• This research provides a robust framework for optimizing injection moulding processes in plastic product manufacturing.