Modelado y control Neuroborroso de sistemas complejos. Aplicación a procesos de mecanizado de alto rendimiento
[EN] This work presents a methodology for the design and implementation of an intelligent control system and an intelligent monitoring system. This methodology is successfully applied to highly complex processes. To that end, it proposes a procedure based on the neurofuzzy modeling of the process in...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Fecha de publicación: | 2010 |
| País: | España |
| Institución: | Universidad de Salamanca (USAL) |
| Repositorio: | GREDOS. Repositorio Institucional de la Universidad de Salamanca |
| OAI Identifier: | oai:gredos.usal.es:10366/76447 |
| Acceso en línea: | http://hdl.handle.net/10366/76447 |
| Access Level: | acceso abierto |
| Palabra clave: | Tesis y disertaciones académicas Universidad de Salamanca (España) Academic dissertations Inteligencia artificial Artificial Intelligence |
| Sumario: | [EN] This work presents a methodology for the design and implementation of an intelligent control system and an intelligent monitoring system. This methodology is successfully applied to highly complex processes. To that end, it proposes a procedure based on the neurofuzzy modeling of the process in question. The models are obtained through an identification process which uses representative input-output data of the system to be studied. Once models which describe the dynamic process have been obtained, these are used as the basis of the monitoring and control systems. In the case of design and implementation of neurofuzzy control systems, it proposes a procedure for obtaining a neurofuzzy model of the process dynamics (direct dynamic) and a neurofuzzy model of its inverse dynamics. These models are used according to the internal model control paradigm to control the complex process. Thus, it designs and implements a neurofuzzy control system based on the internal model control paradigm to control the cutting force in a drilling process (complex electrochemical process) by modifying the feed rate of the tool. Moreover, in the case of neurofuzzy monitoring of complex systems, it proposes a procedure to obtain a neurofuzzy model which relates directly measured process parameters through sensors with a range of phenomena difficult to measure on-line. Thus, the monitoring system is implemented on the basis of the obtained neurofuzzy model. The proposed procedure for developing a neurofuzzy monitoring system, it has been applied to monitoring cutting tool wear in a turning process (complex physical-mechanical process). The information obtained from cutting forces sensors, acceleration (vibration) sensors, acoustic emission sensors and using the operating time, it has developed a neurofuzzy model to estimate the flank wear of the cutting tool. Both in the monitoring system design and in the design and implementation of the control system, it has used different neurofuzzy modeling strategies: an inductive strategy and another transductive strategy. Through the inductive strategy, it has obtained global models representing the entire dynamic process. Instead, through transductive strategies, it has obtained local models to study the particular situation. The strategies used have been ANFIS, TNFIS and TWNFI-i. The use of neurofuzzy models (transductive and inductive) to control and monitoring machining processes stems from the nature of these processes, i.e., its complexity. The nonlinear behavior and the presence of uncertainties (difficult to modeling) both in drilling as in turning processes open the door to the use of these techniques. The advantage of the proponed method is that it eliminates the need for an accurate mathematical model of the complex process to design/adjust the control/monitoring system. The results obtained with neurofuzzy modeling, control and monitoring systems have been very significant results and they are based on real experiments carried out in industrial environments. Moreover, from the technical point of view, significant benefits were obtained such as increasing material removal rate, effective utilization of the cutting tool life, increasing safety for the process (operator, machine and workpiece) and better control of downtimes due to breakage of the cutting tool. |
|---|