摘要
提出了基于模糊神经网络的数控机床温度与热变形的数学模型,该模型根据输入输出样本自动设计和调整模糊系统的参数,并使传统神经网络中没有明确物理含义的权值被赋予模糊逻辑中推理参数的物理含义。将模糊逻辑理论和神经网络结合,提高了网络的泛化能力。文中给出了模糊神经网络结构、算法的具体实现过程。并通过一个仿真实例说明模型可以将20μm内的热变形补偿到1.7μm内,补偿效果明显。针对补偿技术研究与应用中需要快速采集大量的温度及热变形信号问题,以MSC1210微控制器为核心进行模块化设计,温度采集模块通过SPI接口与上位机DSP连接组成数据采集系统。可以方便、灵活的完成数据采集任务。
The mathematical model of temperature and thermal deformation in NC machine tool is proposed based on the fuzzy neural network. According to input and output sample the model automatically designs and adjusts the parameters of fuzzy systems, and endows the weight that is no clear physical meaning of traditional neural network with reasoning parameters of fuzzy logic. The fuzzy logic theory and neural network improve the generalization ability of the network. In this paper the fuzzy neural network structure, the algorithm to realize the particular process are given. And a simulation example shows that the model＇s compensation effect is clear, in which 201xm thermal deformation can be compensated to 1.7ktm. The problem for the thermal error compensation technology research and application needs to rapid acquisition of a large number of required temperature and thermal deformation signal, therefore the modular design takes MSC1210 microcontroller as the core. The temperature acquisition module can link with the host computer DSP through the SPI interface in order to compose the data acquisition system, and complete the data acquisition task conveniently and flexibly.
出处
《电子测量与仪器学报》
CSCD
2009年第9期74-78,共5页
Journal of Electronic Measurement and Instrumentation
基金
先进数控技术江苏省高校重点实验室开放基金(编号:KXJ05020资助项目)
关键词
模糊神经网络
数控机床
热误差补偿
建模
检测系统
模块化
fuzzy neural network
NC machine tool
thermal error compensation
modeling
detection system
modular
