面向表面质量的镍基高温合金铣削参数多目标优化研究
Research on multi-objective optimization of milling parameters for nickel based high-temperature alloys facing surface quality
田应权 1尹瑞雪 1易望远 1欧丽1
作者信息
- 1. 贵州大学 机械工程学院,贵阳 550025
- 折叠
摘要
针对镍基高温合金材料在铣削过程中存在表面加工质量低的问题,提出一种基于神经网络及NSGA-Ⅱ算法的工艺参数多目标优化方法.采用不同工艺参数进行数控铣削镍基高温合金Inconel 718 加工并获取数据集,以表面粗糙度为输出,不同工艺参数组合为输入,利用麻雀搜索算法建立SSA-BP 神经网络模型用于预测Inconel 718 铣削表面粗糙度;以最大材料去除率、最小表面粗糙度为优化目标,构建NSGA-Ⅱ工艺参数多目标优化主体模型,调用构建好的预测模型作为主体模型的目标函数并优化求解得到Pareto最优解集.使用TOPSIS法对Pareto最优解集进行最优解决策,得出最佳的工艺参数组合.优化结果表明:该方法不仅可用于高温合金材料数控铣削表面粗糙度预测,还可用于工艺参数优化,为进一步提高数控铣削材料加工质量和效率提供参考.
Abstract
A multi-objective optimization method for process parameters based on neural network and NSGA-Ⅱ algorithm is proposed to address the problem of low surface processing quality in the milling process of nickel based high-temperature alloy materials.First,different process parameters are employed for CNC milling of nickel based high-temperature alloy Inconel718 and a dataset is obtained.The surface roughness is used as the output and different process parameter combinations as the input.The sparrow search algorithm is employed to establish an SSA-BP neural network model for predicting the surface roughness of Inconel718 during milling;Subsequently,with the maximum material removal rate and minimum surface roughness as optimization objectives,a multi-objective optimization main model for NSGA Ⅱ process parameters is built.The constructed prediction network model is called the objective function of the main model and optimized to obtain the Pareto optimal solution set.TOPSIS method is employed to make optimal solution decisions on the Pareto optimal solution set and obtain the optimal combination of process parameters.Our optimization results indicate this method can be used for predicting surface roughness in CNC milling of high-temperature alloy materials and optimizing process parameters,further improving the processing quality and efficiency of CNC milling materials.
关键词
数控铣削/表面粗糙度/质量优化/难加工金属材料/神经网络Key words
CNC milling/surface roughness/quality optimization/difficult to machine metal materials/neural network引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据