Kerf geometry prediction and optimization in laser cutting of basalt fiber reinforced polymer composites using decision tree and coati optimization algorithm

Abstract

Abstract

This study introduces a Coati Optimization Algorithm (COA) integrated with a Decision Tree (DT) algorithm to predict and optimize kerf geometry in Basalt Fiber Reinforced Polymer (BFRP) composites during laser cutting. Initially, an optimized DT model was developed using COA to forecast kerf width and kerf taper based on five control factors: assist gas pressure, cutting speed, lamp current, pulse width, and pulse frequency. COA was utilized as an internal optimizer to fine-tune the DT's hyperparameters. The model's accuracy was assessed and compared to a basic DT using various statistical measures, demonstrating superior performance despite the composites' heterogeneous and anisotropic nature. Validation involved additional experiments to generate out-of-sample data. The COA was then used to find optimal process control settings, resulting in consistent kerf taper (0.275°) and kerf width (0.297 mm) across all trials, indicating the algorithm's high stability and reliable convergence to optimal solutions regardless of initial conditions or variable factors.