Tool geometries optimization based on machine learning for aviation parts towards green manufacturing

Abstract

Tool geometries optimization is of great importance to reduce cutting energy consumption and tool wear of aeronautical parts under green manufacturing background. Considering this, a machine learning prediction and optimization method is proposed. Firstly, a prediction method, e.g. a tuned Random Forest Regression (RFR) is built up to model the relationship between tool geometries and cutting energy consumption as well as tool wear. Secondly, based on the prediction model, an optimization function of tool geometries is formulated, and seven-spot ladybird algorithm (SLO) is adopted to solve the function. Then, the turning process of aviation shaft parts, which is made of Aluminum alloy (AA) 7075 is viewed as the research object, and Finite Element Method (FEM) is introduced to undertake the simulation machining and generate training and testing data for prediction and optimization. Results show that, compared with previous geometries, the optimized tool geometries could reduce 30.47% and 29.95% of cutting energy consumption and tool wear, demonstrating the effectiveness of the proposed method to explore energy saving potentials in aviation parts manufacturing.