Optimization of X-axis servo drive performance using PSO fuzzy control technique for double-axis dicing saw

Area of Science:

• Semiconductor Manufacturing
• Control Systems Engineering
• Artificial Intelligence in Engineering

Background:

• Dicing saws are crucial for integrated circuit (IC) processing, requiring high accuracy to prevent wafer damage.
• Traditional proportional integral (PI) controllers struggle with the precision needed for high-speed dicing operations.
• Existing methods lack optimal parameter tuning for enhanced servo response in dicing saws.

Purpose of the Study:

• To enhance the X-axis servo response accuracy of a dicing saw using advanced control strategies.
• To introduce and evaluate a novel fuzzy controller approach for dicing saw position loop control.
• To compare the efficacy of Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) for optimizing fuzzy controller parameters.

Main Methods:

• Simulation of an ADT-8230 dual-axis abrasive wheel dicing saw's X-axis servo response.
• Implementation of a fuzzy controller with a two-input, two-output rule-based system for on-line PI parameter correction.
• Optimization of fuzzy controller parameters using heuristic algorithms, specifically PSO for quantization and GA for proportionality factors.
• Validation through MATLAB/Simulink simulations using real servo data.

Main Results:

• The PSO-fuzzy controller demonstrated a significant reduction in position control error by 11.8%.
• Tracking performance was improved by 26% compared to traditional methods.
• Torque pulsation was reduced by 23% using the proposed PSO-fuzzy control strategy.
• Comparative analysis confirmed the superiority of the PSO-fuzzy approach over the GA-fuzzy controller.

Conclusions:

• The PSO-optimized fuzzy controller offers a superior method for improving dicing saw servo accuracy in IC processing.
• The developed control strategy effectively mitigates errors and enhances performance critical for wafer dicing.
• Future research should explore more advanced optimization algorithms to further refine servo accuracy.