Análisis mediante redes neuronales del perfil energético de una prensa de conformado de chapa en el sector de la automoción
- CAMILO CARRILLO-GONZÁLEZ 1
- ELOY DÍAZ-DORADO 1
- JOSÉ CIDRÁS-PIDRE 1
- AITOR CORREA-RODRÍGUEZ 2
- ENRIQUE CASAREJOS RUIZ 2
- CRISTINA ISABEL MARTÍNEZ CASTAÑEDA 3
- JOSE FLORENCIO SÁNCHEZ RÚA 3
- 1 Grupo de Ingeniería Eficiente y Digital, Universidad de Vigo, España
- 2 Grupo de Diseño y Simulación Numérica en Ing. Mecánica, Universidad de Vigo, España
- 3 Grupo Stellantis, Vigo
ISSN: 1137-2729
Ano de publicación: 2023
Volume: 27
Número: 2
Páxinas: 17-33
Tipo: Artigo
Outras publicacións en: Revista iberoamericana de ingeniería mecánica
Resumo
This paper analyzes the performance of a press for sheet metal forming, type transfer, in the automotive sector based only on electrical measurements. One of the particularities of this press is the change of the type of part to be manufactured, with more than eight changes per day, with an approximate production of 1000 parts per hour. A non-intrusive method is presented to analyze the behavior of the press only with electrical measurements in the main panel. By means of this analysis it is intended to know the real production parameters, such as cadence and consumption per piece produced, which are currently only known in an approximate way. For this purpose, the consumption of the subsystems is recorded by means of a network analyzer with which RMS values of current and electrical power are obtained every 20 ms. The consumption pattern analysis technique is based on Fourier analysis, which allows obtaining patterns in the frequency domain. These are associated with the different production states of the press. The main result is the recognition of the operating states of the press (start-up, stop, normal production...) in addition to the type of part in production. The result of this trans-formation is used as input to a neural network that performs the recognition of the parts in production in real time. Thanks to the stability of the harmonic spectrum, when the production is in permanent regime, it is possible to determine the part being produced at any time and to detect the start and stop phases of the press.