Modeling and optimization of surface roughness in keyway milling using ANN, genetic algorithm, and particle swarm optimization
Modeling and optimization of surface roughness in keyway milling using ANN, genetic algorithm, and particle swarm optimization
复制标题
DOI：
10.1007/s00170-017-1417-4
复制
发表时间：
2019-02-01
影响因子：
3.4
通讯作者：
Mondal, Subhas Chandra
中科院分区：
文献类型：
--
作者：
Ghosh, Gourhari;Mandal, Prosun;Mondal, Subhas Chandra
本文重点发展了表面粗糙度的综合研究，用于湿条件下 C40 钢键槽铣削操作的建模和切削参数优化。主轴转速、进给量和切削深度被视为输入参数，表面粗糙度（R-a）被选择为输出参数。表面粗糙度模型是通过人工神经网络（ANN）和响应面法（RSM）开发的。执行方差分析以确定过程参数对响应的影响。分别使用基于Levenberg-Marquardt（LM）和梯度下降（GDX）方法的反向传播算法来训练神经网络，并对两种方法获得的结果进行比较。发现用LM算法训练的网络给出了更好的结果。 ANN 模型（由 LM 算法训练）与遗传算法 (GA) 相结合，RS 模型进一步与 GA 和粒子群优化 (PSO) 相结合，以优化切削条件，从而实现最小的表面粗糙度。结果发现RSM与PSO结合得到了更好的结果，并通过验证测试验证了结果。 RSM-PSO 技术的预测 R-a 值与实验 R-a 值之间观察到良好的一致性。
This paper emphasizes on the development of a combined study of surface roughness for modeling and optimization of cutting parameters for keyway milling operation of C40 steel under wet condition. Spindle speed, feed, and depth of cut are considered as input parameters and surface roughness (R-a) is selected as output parameter. Surface roughness model is developed by both artificial neural networks (ANN) and response surface methodology (RSM). ANOVA analysis is performed to determine the effect of process parameters on the response. Back-propagation algorithm based on Levenberg-Marquardt (LM) and gradient descent (GDX) methods is used separately to train the neural network and results obtained from the two methods are compared. It is found that network trained by the LM algorithm gives better result. ANN model (trained by the LM algorithm) is coupled with genetic algorithm (GA) and RS model is further interfaced with the GA and particle swarm optimization (PSO) to optimize the cutting conditions that lead to minimum surface roughness. It is found that RSM coupled with PSO gives better result and the result is validated by confirmation test. Good agreement is observed between the predicted R-a value and experimental R-a value for RSM-PSO technique.