A Validity Study of Taguchi Method on a Single Data Collection and Comparison to Firefly Algorithm in Turning Process

• Main Author: Dwi Saputro, Ervin
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/34565
• Institution: Institut Teknologi Bandung

The Taguchi method uses the Signal to Noise Ratio (SNR) analysis to obtain optimum value that requires repeated data acquisition in one combination of control factors. However, most current use of the Taguchi method does not follow the concept, although there is another option using the standard analysis that does not require data repetition. The weakness of the Taguchi method is the possibility of the optimum value obtained that is not at the global optimum value. This is different from the Firefly Algorithm method that is able to achieve global optimum value. In this research, the validity study of Taguchi method for single data collection was conducted by comparing the optimization results obtained against standard analysis and validation experiment. The validity is performed based on four comparator parameters, i.e. combination of control factors, optimum value prediction, error of the optimum value prediction, and rank of influence of control factors. This research was conducted using three cases of turning process. In this research, there was also a comparison between Taguchi Method and Firefly Algorithm on three cases of turning process. The comparison was done based on the value of each control factors and the optimum value generated by both methods. Based on the research that has been done, it can be concluded that optimization using Taguchi Method on single data collection is valid. Firefly Algorithm was able to predict response value more minimum than Taguchi Method on three cases of turning process observed.