Estudio comparativo de clasificadores empleados en el diagnóstico de fallos de sistemas industriales // A comparative study of clasification methods used in the fault diagnosis of industrial systems
Abstract
En este artículo se presenta un estudio comparativo del desempeño de cuatro de las técnicas de
clasificación más usadas para el diagnóstico de fallos en procesos industriales. Dentro de las
técnicas seleccionadas se encuentran los clasificadores Vecinos más Cercanos (VMC), Mínimos
Cuadrados Parciales (MCP), Redes Neuronales Artificiales (RNA) y Máquinas de Soporte Vectorial
(MSV). El estudio comparativo se realiza con el objetivo de determinar las técnicas con mayor
capacidad para clasificar de forma correcta los patrones que identifican fallos en procesos
industriales a partir de los datos históricos provenientes de los mismos. Para el estudio se utilizaron
los datos obtenidos de la simulación del modelo del proceso industrial Tennessee Eastman. La
comparación permitió comprobar cómo la capacidad de generalización de las técnicas de
clasificación se incrementa con el aumento de la complejidad en los clasificadores sin que esto
implique necesariamente un mayor esfuerzo computacional en el diagnóstico.
Palabras claves: procesos industriales, diagnóstico de fallos, mantenimiento industrial, máquinas de
soporte vectorial, redes neuronales artificiales, mínimos cuadrados parciales, vecinos más cercanos.
_________________________________________________________________________
This paper, presents a comparative study of the performance of four classification techniques very
used in fault diagnosis of industrial processes. The selected techniques were: k-Nearest neighbor (k-
NN), Partial least-squares (PLS), Artificial Neuronal Networks (ANN) and Support Vector Machines
(SVM). The comparison is based in the classification capacity of the historical data and the
generalization using new observations. The four techniques are applied to historical data of the
known benchmark Tennessee Eastman industrial process. The comparison permitted to prove as the
generalization capacity of the classification techniques grow with the complexity of classifiers without
to increase the computational effort in the fault diagnosis.
Key words: industrial process, fault diagnosis, industrial maintenance, support vector machines, artificial
neural networks, partial least-squares, k-nearest neighbor method.
Downloads
How to Cite
Issue
Section
License
Authors who publish with this journal agree to the following terms:a. Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
b. Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
c. Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).