Intelligent Control Concepts for Computer Numerical Control Machines

9625814 Bowden本研究的目的是设计和开发一个原型传感器融合自适应批判神经控制系统，它将产生切割路径的计算机数控（CNC）机床只给出所需的表面形状和最大允许的表面误差。 本文的研究将为开发软件和硬件奠定基础，简化产品设计和自动化制造过程之间的接口，提高产品质量。 它是数控加工过程自适应控制研究的一次技术飞跃，也是数控加工过程自适应控制研究的一个重大进展。 该小组将开发，测试和演示一种新的传感器融合自适应批判神经控制系统切割翼型直纹表面。 神经控制系统将为5轴铣床提供智能，以根据所需形状的数学描述切割翼型形状的模具。 神经控制系统将利用切割过程中可用的多个传感器读数作为自适应评判器的输入。 传感器信息将被融合用于降维，并提供更丰富的输入。 自适应评价器是一种强有力的自适应控制方法。 为了最大限度地提高成功实现研究目标的可能性，为该项目组建了一个跨学科的专家团队。 该团队的成员来自密西西比州立大学（MSU）的工业工程系，MSU的Raspet飞行研究实验室（RFRL）和精确自动化公司。密西西比州立大学的RFRL拥有强大的精密模具制造活动，而Accurate Automation Corporation在神经网络系统开发方面拥有悠久的历史。***

9625814 Bowden The objective of this research is to design and develop a prototype sensor fusion adaptive critic neurocontrol system which will generate the cut paths for a computer numerical control (CNC) machine given only the desired surface shape and the maximum allowable surface error. This research will lay the foundation for the development of software and hardware simplify the interface between the product design and automated fabrication processes and to improve product quality. It attempts a technological leap in adaptive controls research for CNC machining processes and a major advance in as well. This group will develop, test, and demonstrate a novel sensor fusion adaptive critic neurocontrol system for cutting airfoil shaped ruled surfaces. The neurocontrol system will provide a 5-axis milling machine with the intelligence to cut molds for airfoil shapes based on a mathematical description of the desired shape. The neurocontrol system will make use of the multiple sensor readings available during the cutting process as inputs to the adaptive critic. The sensor information will be fused for dimensionality reduction and to provide greater richness of inputs. The adaptive critic is a powerful method for adaptive control. To maximize the probability of successfully achieving the research objective, a cross-disciplinary team of experts has been assembled for the project. Members of the team come from the Department of Industrial Engineering at Mississippi State University (MSU), MSU's Raspet Flight Research Laboratory (RFRL), and Accurate Automaton Corporation. Mississippi State University's RFRL has a strong precision mold making activity and Accurate Automation Corporation has an extensive history in neural network system development. ***