基于神经网络与遗传算法的刀具磨损检测与控制

发布时间：2018-06-10 08:30

本文选题：刀具磨损 + 检测　；参考：《光学精密工程》2015年05期

【摘要】：针对切削参数对刀具磨损状况和使用寿命的影响,研究了基于神经网络和遗传算法的刀具磨损检测与控制方法。采用多因素正交试验设计方法进行了马氏体不锈钢平面的铣削实验,通过万能工具显微镜测量后刀面的磨损量得到训练样本。借助BP神经网络的非线性映射能力,通过有限的训练样本建立了关于切削速度、每齿进给量、背吃刀量和切削时间的刀具磨损预测模型。实验显示该神经网络预测模型的预测误差不超过5.4%。最后构建了使刀具磨损量为最小的切削参数优化模型,根据每一代的刀具磨损量定义个体的适应度评价函数,提出了切削参数优化模型的遗传算法求解技术。与Taguchi法相比,基于遗传算法的优化方法所获得的最优切削参数减小了6.734%的刀具磨损量。实验显示:提出的刀具磨损检测与控制技术提高了刀具磨损量的计算效率与精度,并为切削参数的合理选择提供了基础理论。
[Abstract]:Aiming at the influence of cutting parameters on tool wear condition and service life, a tool wear detection and control method based on neural network and genetic algorithm is studied. The milling experiment of martensitic stainless steel plane was carried out by means of multi-factor orthogonal experimental design. The training sample was obtained by measuring the wear amount of the tool surface by universal tool microscope. With the help of the nonlinear mapping ability of BP neural network, a tool wear prediction model for cutting speed, feed per tooth, back feed and cutting time is established through a limited training sample. Experiments show that the prediction error of the neural network prediction model is less than 5.4. Finally, a cutting parameter optimization model with minimum tool wear is constructed. According to the individual fitness evaluation function defined for each generation of tool wear, a genetic algorithm is proposed to solve the cutting parameter optimization model. Compared with Taguchi method, the optimal cutting parameters obtained by genetic algorithm (GA) are reduced by 6.734%. Experimental results show that the proposed tool wear detection and control technology improves the efficiency and accuracy of tool wear calculation and provides a basic theory for the reasonable selection of cutting parameters.
【作者单位】：南昌航空大学无损检测教育部重点实验室;
【基金】：国家自然科学基金资助项目(No.51165039,No.51465045) 江西省自然科学基金资助项目(No.20142BAB206018) 无损检测技术教育部重点实验室开放基金资助项目(No.ZD201029004)
【分类号】：TG71;TP274

【参考文献】