Intelligent prediction and optimization of nanofluid-assisted turning using hybrid generative deep learning

Area of Science:

• Sustainable Manufacturing
• Materials Science
• Artificial Intelligence

Background:

• Conventional cutting fluids pose environmental challenges.
• Nanofluid-assisted machining offers a greener alternative but faces prediction and optimization difficulties due to process nonlinearities and limited data.

Purpose of the Study:

• To develop an innovative hybrid generative deep-learning model for intelligent prediction and optimization of nanofluid-assisted turning.
• To enhance the sustainability and efficiency of the machining process for EN31 steel.

Main Methods:

• Integration of Response Surface Methodology (RSM), Artificial Neural Networks (ANN), and Generative Adversarial Network (GAN) into a hybrid model.
• Experimental validation of MWCNT-based nanofluid performance against traditional coolants.
• Utilizing a hybrid ANN-RSM strategy for parameter optimization based on desirability functions.

Main Results:

• MWCNT nanofluid reduced surface roughness by 12.83% and cutting forces by up to 23.4%.
• The GAN-ANN hybrid model demonstrated superior predictive robustness (R² ≈ 0.97-0.99) under limited data conditions compared to RSM and ANN alone.
• Optimized parameters predicted cutting force (205 N), surface roughness (0.45 μm), and material removal rate (16.6 m³/min) with <5% experimental error.

Conclusions:

• The proposed hybrid deep-learning framework provides a scalable, interpretable, and environmentally friendly decision support system for intelligent machining.
• This approach effectively addresses the challenges of prediction and optimization in nanofluid-assisted processes.
• The study highlights the potential of advanced AI techniques in advancing sustainable manufacturing practices.