摘要
在机械加工过程中,工件尺寸误差影响着产品的质量。建立输入和输出参数之间的预测模型尤为必要。构建基于BP神经网络的机械切削误差预测模型,并采用L-M算法优化训练权值参数,克服了BP神经网络计算复杂、训练时间过长、容易限于局部极小等缺陷。给出实验方案,并以一个机械加工实例验证该BP神经网络误差预测模型的准确性和有效性。
The product quality is affected by error of workpiece in course of mechanical processing.It is very important to set up parameter model of input-output system.A mechanical cutting error predictive model based on BP neural networks was built with L-M algorithm to optimize training parameters,so some disadvantages of BP neural networks such as complex calculation,too long training time and easy to falling into local minimum,were solved.An example of mechanical cutting was given to test the accuracy and effectiveness of BP neural networks predictive model.
出处
《机床与液压》
北大核心
2013年第11期67-71,共5页
Machine Tool & Hydraulics
基金
重庆市教委科学技术研究项目(KJ110818)
重庆理工大学科研启动项目(2010ZD24)
重庆理工大学青年基金项目(2011ZQ25)
关键词
机械加工
BP神经网络
误差预测模型
Mechanical processing
BP neural networks
Error predictive model
