关键词: 铣削加工, 经验公式, 极端梯度提升, 遗传算法

Abstract: To address the limitations of traditional surface roughness prediction methods,including insufficient information fusion and reliance on manually tuned or poorly optimized model parameters,a prediction method integrating Genetic Algorithm (GA),Empirical Formula (EF),and Extreme Gradient Boosting (XGBoost),termed GA-EF-XGBoost is proposed. The method calculates the first component of surface roughness using an empirical formula based on milling parameters and obtains the second component by applying the XGBoost model to vibration signals. These components are then integrated using a genetic algorithm to produce a comprehensive surface roughness prediction. Experimental results show that the GA-EF-XGBoost model achieves a prediction accuracy of 93.39 %,significantly outperforming traditional machine learning models and other approaches. By combining milling parameters and real-time vibration signals,this method enhances surface roughness prediction accuracy through a synergy of empirical and data-driven approaches,demonstrating potential for practical applications.

Key words: milling process, Empirical Formula(EF), Extreme Gradient Boosting(XGBoost), Genetic Algorithm(GA)