摘要
不确定性信息的处理是目前控制器设计的关键和难点 ,变切削深度加工过程有大量的不确定性信息 ,本文以其为实验对象 ,研究神经网络控制效果 ,虽然网络学习率已根据实际需要作了修改 ,但系统响应速度较慢。据此 ,提出基于信息熵的神经网络优化控制算法 ,对熵函数中的概率分别采用均匀分布规则和根据最大熵原理求取两种策略。比较发现 ,前者比最初神经网络控制更能显著提高系统响应速度 ,后者能在此基础上将震荡次数减少 2 / 3。实验证明 ,基于信息熵测度的控制对处理不确定性系统具有很好的效果。
How to deal with uncertain information is the key and difficult proble m in the modern manufacturing, and varied cutting-depth process is a typical un c ertain system. Firstly, a neural network controller is designed for the machinin g process, and its learning speed has been adjusted with different conditions, b ut the system responses slowly. Then a new algorithm of neural network based on information entropy is put forward. Its output probability is uniform distributi ng, and the result shows that the adjusting time is shorter than that of neural network control, finally the controller is revised to get the probability accord ing to maximum entropy theory with the system-shock time reducing 2/3.
出处
《工具技术》
北大核心
2005年第2期11-14,共4页
Tool Engineering
基金
国家自然科学基金资助项目 (项目编号 :5 0 175 0 2 9)
教育部留学回国人员科研启动基金资助项目
关键词
信息熵
神经网络
智能控制
切削加工
information entropy, neural network, cutting process, intelligent control
