Comparative Analysis of Machine Learning Methods for Predicting Robotized Incremental Metal Sheet Forming Force

Area of Science:

• Manufacturing Engineering
• Materials Science
• Mechanical Engineering

Background:

• Single Point Incremental Forming (SPIF) is a flexible sheet metal forming process.
• Accurate prediction of forming forces is crucial for process control and failure prevention.
• Friction significantly influences forming forces in SPIF, necessitating advanced control strategies.

Purpose of the Study:

• To develop a method for extracting information from SPIF process parameters.
• To propose an innovative solution for actively controlling friction forces using modulated vibrations.
• To investigate the influence of material properties, process parameters, and sheet thickness on maximum forming force.

Main Methods:

• Utilized Artificial Neural Networks (ANN) and various machine learning (ML) algorithms for force prediction.
• Employed modulated vibrations to actively control friction between the tool and sheet metal, replacing traditional lubricants.
• Generated sampling data assuming normal distribution for input function variability.

Main Results:

• Developed efficient force prediction models using ANN and Gaussian Process Regression (GPR).
• Predicted forces showed good agreement with experimental results.
• The proposed vibration-based friction control method demonstrated effectiveness.

Conclusions:

• ANN and GPR are highly effective methods for developing robust forming force prediction models in SPIF.
• The developed models offer high performance and balance bias-variance for industrial applicability.
• Vibration-induced friction modulation presents a viable, environmentally friendly alternative to conventional lubrication in SPIF.