OXIDATION RATE PREDICTION OF MO-SI-B ALLOY USING MACHINE LEARNING WITH RANDOM FOREST REGRESSOR METHOD

• Main Author: Yuliasari, Adisya
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/67913
• Institution: Institut Teknologi Bandung

Mo-Si-B alloy is a high-temperature material that can be one of the candidates for the constituent material of turbine jet engines because of its good oxidation resistance. Research related to the high-temperature oxidation behavior of Mo-Si-B alloys has been carried out experimentally which requires time and costs in its development. Another method is needed to accelerate the development of Mo-Si-B alloy as the constituent material for turbine jet engines. To tackle that challenge, this study focuses on predicting the oxidation rate of Mo-Si-B alloys using machine learning which will be a reference in conducting Mo-Si-B alloy oxidation test through experiment. This machine learning uses a random forest regressor algorithm with the lolopy learners library and also optimized with feature engineering. The results of this study obtained the accuracy of machine learning models using random forest regressors with R-squared and MAE of 0.912 and 1.02. The prediction of the parbolic rate of Mo-Si-B oxidation using the model is also close to the actual trend. There is decrease in the oxidation rate as the composition of B and Si increases. A new feature that has an influence on the oxidation rate of Mo-Si-B alloys is also obtained, namely mixing entropy.