Abstract: To carry out the dynamic multi-objective optimization of milling parameters under considering the varying performance of machine tool,a strategy of dynamic multi-objective optimization of milling parameters was developed based on digital twin.The Gradient Boosting Regression Tree (GBRT) algorithm  was used to construct the nonlinear mapping relationship between processing parameters and processing results.The Dynamic Non-Dominated Sorting Genetic Algorithm (DNSGA-Ⅱ-A) was adopted to dynamically optimize the milling parameters considering tool wear.Based on obtained pareto optimal solution,a decision analysis model was established by combining Analytic Hierarchy Process (AHP) and Technique For Order Preference By Similarity To An Ideal Solution (TOPSIS),and the visual analysis ranking was realized as well.The established dynamic multi-objective optimization strategy of milling parameters could provide the optimal milling parameter selection scheme in accordance with the characteristics of the current machine tool,which ensured the processing quality and efficiency.

Key words: digital twin, milling parameters, gradient boosting regression tree, dynamic multi-objective optimization

摘要： 为解决机床性能动态变化过程中的铣削参数动态多目标优化问题,提出一种基于数字孪生的铣削参数动态多目标优化策略。首先采用梯度提升回归树算法构建加工参数与加工结果间的非线性映射关系;然后基于动态非支配排序遗传算法进行铣削参数动态寻优;最后在Pareto最优解的基础上,结合层次分析法和理想解相似度顺序偏好法建立决策分析模型并进行可视化分析排序。该策略能够针对机床整个运行时段提供符合当前机床特性的最优铣削参数取值方案,从而保证加工质量和加工效率。

关键词: 数字孪生, 铣削参数, 梯度提升回归树, 动态多目标优化