An on-line predictive system for steel wire straightening using genetic programming
Introduction
The dimensional stability of semi products represents one of the important problems in modern mass production method. There are several reasons for keeping the geometry of products within a narrow tolerance field. Highly efficient automatic assembly lines could be addressed, since they are the reason for the production run-outs, when the fluctuation of the semi product geometry is too high. Lever-arch mechanism for maps and files (Fig. 1) can be mentioned as such an example of mass production.
One would agree that the product is very simple at first glance, but considering the production of over 50 million per year it is not quite so. At such a production rate many very specific problems arise. One important problem is the geometrical stability of all the assembling parts of the mechanism. Seeking for the reason and solving this problem is quite a tedious task. The present paper discuses the genetic programming approach.
Section snippets
Definition of the problem
The problem of geometrical accuracy has been mentioned in the introduction. In the paper the focus will be put on the production of the arch (Fig. 2), which is the most problematic from the stability point of view.
The first problem arises from an esthetic point of view. Interaction between the arch and the pin should be as smooth as possible, without more serious steps. The other problem represents the assembling machinery, which is very sensitive to fluctuation of the arch width. So the
State of the art—literature overview
Modelling of real manufacturing systems is typically iteration trial–error process. In the first step the system description is set up (differential equation system) and solving occurs. In the second step the results are fitted to the model. Lopez et al. (2002) describe an approach based on a search of a model of the system in a block diagram representation, where the trial–error process is solved with Genetic Programming.
In metal-forming industry, we deal with many forming processes where
Principle of Ga and GP
GP is probably the most general approach of evolutionary computation methods (Michalewicz, 1996). GP was introduced by Koza in the first half of this decade (Koza (1992), Koza (1994)). In GP the structures subject to adaptation are the hierarchically organised computer programs whose size and form dynamically change during evolution. In order to stick with the biological metaphor, the computer programs are called organisms or also chromosomes.
In GP the space of solutions is the hyperspace of
Modelling of wire straightening using genetic programming
The major objective of the experimental work was looking for the analytical function which connects input and output parameters of the process. Since there are many input and output parameters, we have chosen the genetic programming approach. Input process parameters were the positions of the straightening rollers as presented in Fig. 5, output parameters are the geometrical properties of the bent arch.
The reason for using the genetic programming approach to the modelling of rip production
Experimental equipment for on-line control of the arch geometry
An application for on-line control has been developed within the LabVIEW environment (version 5.0) (LabVIEW, 1996). It is possible to include the genetic model and to adjust the rollers in such a way that the arch geometry (u1) is as stable as possible. Function has to be partially derived to find the necessary adjustment of the rollers. The most important part of the application is control panel (Fig. 7) where the major parameters are set:
- •
Wire diameter,
- •
Tensile strength and yield stress,
- •
Working
Conclusion and further work
The problem of wire straightening and one possible solution is presented in the article. Experimental work on the production machine is because of the large number of input parameters evaluated using a genetic programming approach. The usefullnes of the method is in its ability to develop in the evolutionary process and is not forced to the measurements. Orign for the method can be found in a real life where only the best adapted organism survives. The selection is based on the level of
References (13)
- et al.
Metal-forming process optimisation by inverse evolutionary search
Journal of Materials Processing Technology
(2002) - et al.
Genetic programmingprinciples and applications
Engineering Applications of Artificial Intelligence
(2001) - Balic, J., 1999. Contribution to Integrated Manufacturing (DAAAM publishing series, Manufacturing technology). DAAAM...
- Brezoènik, M., 1998. Modeliranje tehnoloških sistemov z uporabo genetskih metod. Doktorska disertacija,...
Genetic Programming
(1992)Genetic Programming II
(1994)