Application of Low-Cost Microcontroller-Based Data Acquisition in Multi-objective Machine Learning Optimization for Injection Molding: Aiming at Energy Consumption and Product Quality

In the traditional injection molding industry, enhancing production efficiency and product quality remains a persistent challenge. As environmental, social, and governance (ESG) standards rise, businesses must balance economic benefits with environmental responsibilities, particularly concerning energy consumption, carbon emissions, and resource utilization efficiency. Although computer-aided engineering (CAE) technologies have achieved significant improvements in process optimization, there is still a gap in real-time monitoring and optimization of machine energy consumption.

This study utilizes a flat plate as the platform, combined with low-cost microcontrollers (Arduino) for energy consumption data acquisition, covering equipment such as mold temperature machines, hydraulic systems of injection molding machines, and screw heaters. Through single-factor experiments and heat maps, the study analyzes the impact of various process parameters on energy consumption. The experiments employ the Central Composite Design (CCD) to reduce both time and costs. A neural network model, built using TensorFlow and Keras, predicts the energy consumption during the injection molding process. The results are then applied to a Non-dominated Sorting Genetic Algorithm (NSGA-II) to optimize both energy consumption and product quality simultaneously.

The experimental results show that the Pareto optimal solutions obtained through the NSGA-II model indicate a conflict between product quality and energy consumption, particularly in low-energy regions where higher product quality standards are challenging to achieve. By adjusting mold temperature and material temperature to increase the gate solidification time, allowing holding pressure to act longer on the un-solidified product, the model compensates for potential decreases in weight and increases in shrinkage due to higher mold and material temperatures. This adjustment ultimately stabilizes the product length.

This experiment demonstrates how a low-cost data acquisition device can be developed to effectively reduce energy consumption by 12.7% while maintaining product quality. It provides a cost-effective experimental evaluation scheme for small and medium-sized enterprises and offers a reference for the injection molding industry's transition towards green manufacturing.