Diseño y simulación de un controlador neuro-difuso desacoplado sobre un modelo NARX del sistema cocina-prensa de una planta pesquera para mejorar su eficiencia
Cargando...
Fecha
Authors
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad de Ingeniería y Tecnología
Resumen
Una producción eficiente del sistema cocina-prensa de una planta pesquera permite producir una harina con bajo contenido de humedad y grasa, denominada “prime”. Austral Group S.A.A., una empresa pesquera, ha priorizado su producción porque genera mayores ingresos económicos. Para ello, se ha automatizado el sistema priorizando la aplicación del control PID clásico en los procesos: temperatura de cocina, nivel de tolvin prensa y corriente del motor de prensa. Sin embargo, dichos controladores son insuficientes para estabilizar las variables y generar una operación eficiente. Bajo esto en consideración, el presente trabajo propone un control neuro-difuso desacoplado para mejorar la eficiencia del sistema cocina-prensa. Debido a que no se cuenta con acceso a la planta, pero sí a mediciones de las variables del proceso, se obtiene un modelo computacional del sistema a partir de datos experimentales. Para ello, se propone una red neuronal NARX. Para sintonizar las funciones de membresía de los controladores, se propone una red ANFIS. Tras ello, se evalúan diferentes estructuras de controlares difusos: tipo PD, tipo PI y tipo PI-PD. Para fines comparativos, se simulan los controladores PID Compact actualmente implementados. Para acercar el comportamiento de simulación a la realidad, se someten los sistemas de control a disturbios de ruido y perturbaciones típicas. Por último, se implementan los bloques de control en TIA Portal y se analizan las consideraciones de implementación basado en el uso de recursos. Se concluye que, mediante la aplicación de controladores neuro-difusos en el sistema cocina prensa, se cumplen las especificaciones de diseño para la elaboración de una harina de alta calidad y se obtiene un mejor rechazo a las perturbaciones típicas del sistema.
Efficient production of the heating-pressing system of a fish meal plant allows the production of a meal with low moisture and fat content, called “prime”. Austral Group S.A.A., a fishing company, has prioritized its production because it generates higher economic income. To achieve this, the system has been automated, prioritizing applying classic PID control in kitchen temperature, press chute level, and press current. However, these controllers are insufficient to stabilize the variables and generate an efficient operation. Considering this, the present work proposes a decoupled neuro-fuzzy control to improve the efficiency of the heating-pressing system. Since there is no access to the plant but to measurements of the process variables, a computational model of the system is obtained from experimental data. For this, a NARX neural network is proposed. To tune the membership functions of controllers, an ANFIS network is proposed. After that, different structures of fuzzy controls are evaluated: PD type, PI type, and PI-PD type. For comparison purposes, currently implemented PID Compact controllers are simulated. The control systems are subjected to noise and typical disturbances to bring the simulation behavior closer to reality. Finally, the control blocks are implemented in TIA Portal software, and implementation considerations based on resource usage are discussed. It is concluded that, through the application of neuro-fuzzy controllers in the heating-pressing system, the design specifications for producing high-quality flour are met, and better rejection of typical system disturbances is obtained.
Efficient production of the heating-pressing system of a fish meal plant allows the production of a meal with low moisture and fat content, called “prime”. Austral Group S.A.A., a fishing company, has prioritized its production because it generates higher economic income. To achieve this, the system has been automated, prioritizing applying classic PID control in kitchen temperature, press chute level, and press current. However, these controllers are insufficient to stabilize the variables and generate an efficient operation. Considering this, the present work proposes a decoupled neuro-fuzzy control to improve the efficiency of the heating-pressing system. Since there is no access to the plant but to measurements of the process variables, a computational model of the system is obtained from experimental data. For this, a NARX neural network is proposed. To tune the membership functions of controllers, an ANFIS network is proposed. After that, different structures of fuzzy controls are evaluated: PD type, PI type, and PI-PD type. For comparison purposes, currently implemented PID Compact controllers are simulated. The control systems are subjected to noise and typical disturbances to bring the simulation behavior closer to reality. Finally, the control blocks are implemented in TIA Portal software, and implementation considerations based on resource usage are discussed. It is concluded that, through the application of neuro-fuzzy controllers in the heating-pressing system, the design specifications for producing high-quality flour are met, and better rejection of typical system disturbances is obtained.
Descripción
Citación
Gutarra Sanchez, A. G. (2025). Diseño y simulación de un controlador neuro-difuso desacoplado sobre un modelo NARX del sistema cocina-prensa de una planta pesquera para mejorar su eficiencia [Tesis de Título Profesional, Universidad de Ingeniería y Tecnología]. Repositorio Institucional UTEC. https://hdl.handle.net/20.500.12815/516
Colecciones
Aprobación
Revisión
Complementado por
Referenciado por
Licencia Creative Commons
Excepto donde se indique lo contrario, la licencia de este ítem se describe como info:eu-repo/semantics/openAccess

