论文信息 - Application of neural networks for compliant tool polishing operations

Application of neural networks for compliant tool polishing operations

Abstract The purpose of this research is to develop a strategy for automating polishing operations using filamentary brushes. In the first stage of the research, the experiments using the Taguchi system of experimental design are performed to identify the effects of various polishing parameters on the surface finish. It is observed that an abrasive impregnated filamentary brush improves the surface finish more substantially than a nylon brush. In addition, an interaction between brush type and rotational speed, and an interaction between brush rotational speed with brush depth, and type of brush with use of lubricant have been identified. In the second stage of the research, the experimental results are utilized to investigate an applicability of using neural networks in predicting and controlling polishing operations. It is shown that neural networks with one hidden layer learn faster than the two hidden layer designs. The neural network trained to predict the surface finish for a given set of polishing parameters performed with an average percent error between 1%–2%. However, the neural network designed to determine the levels of polishing parameters for a desired surface finish did not perform as well.

[1] S. Y. Kung,et al. An algebraic projection analysis for optimal hidden units size and learning rates in back-propagation learning , 1988, IEEE 1988 International Conference on Neural Networks.

[2] A. Sarkar. Wear of metals , 1976 .

[3] R. J. Stango,et al. Measurement and Analysis of Brushing Tool Performance Characteristics, Part 1: Stiffness Response , 1991 .

[4] R. J. Stango,et al. Aspects of Process Model for Automatic Control of Edge-Deburring with Filamentary Brush , 1992 .

[5] Geoffrey E. Hinton,et al. Learning internal representations by error propagation , 1986 .

[6] William E. Jones,et al. Back-propagation, a generalized delta learning rule , 1987 .

[7] M. C. Shaw. Metal Cutting Principles , 1960 .

[8] Madhan Shridhar Phadke,et al. Quality Engineering Using Robust Design , 1989 .

[9] Clifford Goodman,et al. American Society of Mechanical Engineers , 1988 .

[10] B. Moore,et al. Characterizing the error function of a neural network , 1988, Proceedings., 2nd Symposium on the Frontiers of Massively Parallel Computation.

[11] Geoffrey E. Hinton,et al. Learning representations by back-propagating errors , 1986, Nature.