This news article was originally written in Spanish. It has been automatically translated for your convenience. Reasonable efforts have been made to provide an accurate translation, however, no automated translation is perfect nor is it intended to replace a human translator. The original article in Spanish can be viewed at
Supervisión del estado de la Herramienta mediante conocimiento expertoControl and planning
Supervision of the State of the tool by expert knowledge
David Rodríguez Salgado
Department of electronics and electromechanics engineering. Universidad de ExtremaduraUniversity Center of Mérida c / Santa Teresa de Jornet 38. 06800 Mérida
Tel.: 653 46 68 69 e-mail:drs10@navegalia.com
School of industrial engineering, Elvas Ave. s/n, Badajoz 06071
01/12/2002 1. Introduction
The detection of incipient failures in industrial plants, has acquired great importance with the increased automation of processes in recent years. There are two methodologies to address the problem: techniques based on the analysis of the signal, which extracted characteristics of the State of operation and techniques based on the analysis of the dynamics of the system. In any case, the variation of the parameter examined with respect to the normal values indicates the possibility of the existence of a fault.
This article describes a system of supervision of the State of the tool, based on the analysis of the current consumed in the process of machining.
The wear of the tool, is different depending on the time of court, in a way that they can distinguish three areas, see Figure 1. The area to is characterized by a rapid wear, in zone b wear is proportional to the time of machining, and in the zone c wear is exponential, and can produce a sudden failure of the tool. Wear curve implies that the consumption of electricity will present a similar variation depending on wear and tear. Grows approximately linear with the tool working time and later undergoes a sharp variation.
2 Results and discussion
This article presents a new way to tackle the problem, analyzing the signal of electric current at the time. The variable that is measured is area of the curve, by gauss quadrature, and checking their departure from the normal value. This approach allows a simple technique, without need for algorithms more complex such as those who are
they are used [1].
The sample size is five, and the length of the sliding temporal window at the time may vary depending on the precision required.
3 Findings
The presented methodology, is characterized by simplicity and fast learning of the neural network. In this way, permitted a rapid monitoring of the system, allowing thus act before the failure.
4. References
[1] A. Domínguez Reñones, j. d. Bernardez Pérez, r. Arnanz Gómez, "Monitoring and diagnosis of wear on tools multifilo", Congress of automatic 2000.
[2] J. j. Melendez, j. Colomer, J. De la Rosa, j. Aguilar-Martin, j. Vehi, "Embedding objects into matlab/simulink for process supervision", Proceedings of IEEE Intal. Symposium on Computer Aided Control Systems Design. Dearborn Michigan. USES September 1996