Prediction and optimization kerf width in laser beam machining of titanium alloy using genetic algorithm tuned adaptive neuro-fuzzy inference system
Min Ji,Muthuramalingam Thangaraj,Saravanakumar Devaraj,Ryszard Machnik,Nikolaos E. Karkalos,Panagiotis Karmiris-Obratański
DOI: https://doi.org/10.1007/s00170-024-13681-x
IF: 3.563
2024-05-08
The International Journal of Advanced Manufacturing Technology
Abstract:In the power diode laser beam machining (DLBM) process, the kerf width ( K W ) and surface roughness (SR) are important factors in evaluating the cutting quality of the machined specimens. Apart from determining the influence of process parameters on these factors, it is also very important to adopt multi-response optimization approaches for them, in order to achieve better processing of specimens, especially for hard-to-cut materials. In this investigation, adaptive neuro-fuzzy inference system (ANFIS) and genetic algorithm tuned ANFIS (GA-ANFIS) were used to predict the K W on a titanium alloy workpiece during DLBM. Five machining process factors, namely power diode, standoff distance, feed rate, duty cycle, and frequency, were used for the development of the model due to their correlation with K W . As in some cases, traditional soft computing methods cannot achieve high accuracy; in this investigation, an endeavor was made to introduce the GA-assisted ANFIS technique to predict kerf width while machining grooves in a titanium alloy workpiece using the DLBM process based on experimental results of a total of 50 combinations of the process parameters. It was observed that FIS was tuned well using the ANN in the ANFIS model with an R 2 value of 0.99 for the training data but only 0.94 value for the testing dataset. The predicting performance of the GA-ANFIS model was better with less value for error parameters (MSE, RMSE, MAE) and a higher R 2 value of 0.98 across different folds. Comparison with other state-of-the-art models further indicated the superiority of the GA-ANFIS predictive model, as its performance was superior in terms of all metrics. Finally, the optimal process parameters for minimum K W and SR, from gray relational–based (GRB) multi-response optimization (MRO) approach, were found as 20 W (level 2) for laser power, 22 mm (level 5) for standoff distance, 300 mm/min (level 5) for feed rate, 85% (level 5) for duty cycle, and 18 kHz (level 3) for frequency.
engineering, manufacturing,automation & control systems