Evaluación del desempeño de tres algoritmos de inferencia bayesiana, implementados como sistema experto para la identificación de modos de falla en ejes

dc.contributor.advisorEspejo Mora, Edgarspa
dc.contributor.authorMappe Rojas, Kevin Adalbertospa
dc.contributor.researchgroupGrupo de Investigación: AFIS (Análisis de Fallas, Integridad y Superficies).spa
dc.date.accessioned2020-02-19T18:59:06Zspa
dc.date.available2020-02-19T18:59:06Zspa
dc.date.issued2019-11-29spa
dc.description.abstractEl objetivo principal de este proyecto fue la evaluación del desempeño de la inferencia bayesiana implementada como un sistema experto para la identificación de los modos de fallo en ejes. El software experto se dividió en dos módulos, uno para modos de falla que involucran fractura y otro para modos de fallo que involucran deformación plástica, desgaste y corrosión. Se implementaron tres motores de inferencia bayesianos, dos para inferencia exacta y uno para inferencia aproximada que permitieron la identificación de modos de fallo a partir de una base de conocimiento y la evidencia ingresada al software mediante dos cuestionarios, uno para el módulo de deformación plástica, desgaste y corrosión y otro para el módulo de fractura. Para la base del conocimiento o probabilidades a prior se recopiló de un total de 280 casos de falla en ejes diagnosticados por expertos. Para cada modo de fallo identificado por el experto se realizó el análisis de las marcas características presentes en la zona de falla y con la cantidad de marcas encontradas por los expertos para cada modo de fallo, se conformó la base de datos y las redes bayesianas para el software experto. La evaluación y posterior comparación de los motores de inferencia bayesianos para el software experto consistió en un análisis cuantitativo de los resultados, que se obtuvo al evaluar un total de 62 casos de falla en ejes. Para este análisis se utilizaron las medidas de grupo, índice de acuerdo y kappa, usualmente utilizadas para la evaluación de software experto. Adicionalmente se utilizó la metodología de evaluación de ratios de acuerdo con sus índices (Sensibilidad, especificidad, ratio de falsos positivos, ratio de falsos negativos y ROC “Receiver operating characteristic”). Como resultado de la evaluación mediante los índices de acuerdo se determinó que para este desarrollo y específicamente para las redes bayesianas programadas para cada modo de fallo, el motor de inferencia que obtuvo mejores resultados fue Metropolis-hastings tanto para el módulo de fractura como para el módulo de deformación plástica, desgaste y corrosión. Finalmente, de este proyecto se obtiene un software experto implementado en el lenguaje de programación Python para la identificación de modos de fallo en ejes, con tres motores de inferencia bayesiano.spa
dc.description.abstractThe main goal of this project is the evaluation of the Bayesian inference performance implemented as an expert software for failures modes identification in shafts. Expert software has two modules, one for failures modes with fracture and one for failure modes with plastic deformation, wear and corrosion. Three Bayesian inference engines were implemented, two for exact inference and one for approximate inference, the inference engine allow the identification of failure modes in shafts from a knowledge base and the evidence entered into the software through two questionnaires, one for the plastic deformation, wear and corrosion module and another for the fracture module. The prior probabilities or knowledge base were collected from a total of 280 case of failures in shafts diagnosed by experts. For each failure mode identified by expert, the analysis of the attributes present in the fault zone was performed. With the number of attributes found by experts for each failure mode, the Bayesian database and networks for expert software were formed. The comparison and evaluation of the Bayesian inference engine for the expert software consisted in a quantitative analysis for the result obtained by evaluating a total of 62 cases of shaft failure. For this analysis we used the group measures, agreement index and kappa, usually used for expert software evaluation. In addition, the methodology used to evaluate ratios according to their indices (Sensibility, specificity, false positive rate, false negative rate and ROC “Receiver operating characteristic”) was used. As a result of performance evaluation and according to the index of agreement, in this development and specifically for the Bayesian network programmed for each failure mode, showed that the best inference engine was Metropolis-hastings for fracture and for the module of plastic deformation, wear and corrosion. Finally, from this project we obtain an expert software implemented in the Python programming language for the identification of failure modes in shafts with three Bayesian inference engines.spa
dc.description.additionalMagíster en Ingeniería Mecánica. Línea de Investigación: Modos de falla en elementos de máquinas.spa
dc.description.degreelevelMaestríaspa
dc.format.extent149spa
dc.format.mimetypeapplication/pdfspa
dc.identifier.urihttps://repositorio.unal.edu.co/handle/unal/75649
dc.language.isospaspa
dc.publisher.branchUniversidad Nacional de Colombia - Sede Bogotáspa
dc.relationA performance evaluation of three inference engines as expert systems for failure mode identification in shaftsspa
dc.relation.referencesE. E. Mora and H. Hernández Albañil, Análisis de fallas de estructuras y elementos mecánicos, vol. 1. Bogotá: Universidad Nacional De Colombia, 1 ed., 2015.spa
dc.relation.referencesB. G. Xu, “Intelligent fault inference for rotating flexible rotors using Bayesian belief Network” Expert Systems with Applications, vol. 39, no. 1, pp. 816-822, 2012.spa
dc.relation.referencesC. Li and S. Mahadevan, “Efficient approximate inference in Bayesian networks with continuous variables” Reliability Engineering & System Safety, vol. 169, pp. 269-280, jan 2018.spa
dc.relation.referencesS. J. Russell and P. Norvig, Inteligencia artificial: un enfoque moderno. Madrid: Prentice Hall, segunda ed., 2004.spa
dc.relation.referencesG. Casella, S. Fienberg, and I. Olkin, Bayesian Core: A Practical Approach to Computational Bayesian Statistics. Springer Texts in Statistics, New York, NY: Springer New York, 2007.spa
dc.relation.referencesE. K. A. Karl Heinrich Grote, “Springer handbook of mechanical engineering” Choice Reviews Online, vol. 47, pp. 47-0305-47-0305, sep 2009.spa
dc.relation.referencesS. Climent Serrano, “Sistemas expertos probabilísticos” Universidad De Valencia, p. 301, 1998.spa
dc.relation.referencesE.A. Faigenbaum, “Expert systems in the 1980s” Stanford University, vol. 1, 1980.spa
dc.relation.referencesJay Liebowitz, “The Handbook of Applied Expert Systems” in The Handbook of Applied Expert Systems, CRC Press, 1998.spa
dc.relation.referencesP. J. Graham-Jones and B. G. Mellor, “Expert and knowledge-based systems in failure Analysis” Engineering Failure Analysis, vol. 2, pp. 137-149, jun 1995.spa
dc.relation.referencesJ. Liebowitz, “Expert systems: A short introduction” Engineering Fracture Mechanics, vol. 50, pp. 601-607, mar 1995.spa
dc.relation.referencesC. J. Moreno and E. Espejo, “A performance evaluation of three inference engines as expert systems for failure mode identification in shafts” Engineering Failure Analysis, vol. 53, pp. 24-35, jul 2015.spa
dc.relation.referencesH. Mahersia, H. Boulehmi, and K. Hamrouni, “Development of intelligent systems based on Bayesian regularization network and neuro-fuzzy models for mass detection in mammograms: A comparative analysis” 2016.spa
dc.relation.referencesT. P. Burghardt and K. Ajtai, “Neural/Bayes network predictor for inheritable cardiac disease pathogenicity and phenotype” Journal of Molecular and Cellular Cardiology, vol. 119, pp. 19-27, jun 2018.spa
dc.relation.referencesJ. del Sagrado, J. Sánchez, F. Rodríguez, and M. Berenguel, “Bayesian networks for greenhouse temperature control” Journal of Applied Logic, vol. 17, pp. 25-35, sep 2016.spa
dc.relation.referencesM. Galagedarage Don and F. Khan, “Dynamic process fault detection and diagnosis based on a combined approach of hidden Markov and Bayesian network model” Chemical Engineering Science, vol. 201, pp. 82-96, jun 2019.spa
dc.relation.referencesR. G. Cowell, “Finex: A Probabilistic Expert System for forensic identification” Forensic Science International, vol. 134, no. 2-3, pp. 196-206, 2003.spa
dc.relation.referencesE. Chojnacki, W. Plumecocq, and L. Audouin, “An expert system based on a bayesian network for fire safety analysis in nuclear area” Fire Safety Journal, vol. 105, pp. 28-40, 2019.spa
dc.relation.referencesS. Nasiri, M. R. Khosravani, and K.Weinberg, “Fracture mechanics and mechanical fault detection by artificial intelligence methods: A review” Engineering Failure Analysis, vol. 81, pp. 270-293, nov 2017.spa
dc.relation.referencesB. Kareem, “Evaluation of failures in mechanical crankshafts of automobile based on expert opinion” Case Studies in Engineering Failure Analysis, vol. 3, pp. 25-33, apr 2015spa
dc.relation.referencesY. Wang, H. Yang, X. Yuan, and Y. Cao, “An improved Bayesian network method for fault diagnosis” IFAC-PapersOnLine, vol. 51, no. 21, pp. 341-346, 2018.spa
dc.relation.referencesJ. Wang, Z. Yang, J. Su, Y. Zhao, S. Gao, X. Pang, and D. Zhou, “Root-cause analysis of occurring alarms in thermal power plants based on Bayesian networks” International Journal of Electrical Power & Energy Systems, vol. 103, pp. 67-74, dec 2018.spa
dc.relation.referencesV. H. Jacobo, A. Ortiz, Y. Cerrud, and R. Schouwenaars, “Hybrid expert system for the failure analysis of mechanical elements” Engineering Failure Analysis, vol. 14, no. 8 SPEC. ISS., pp. 1435-1443, 2007.spa
dc.relation.referencesK. Medjaher, J.-Y. Moya, and N. Zerhouni, “Failure prognostic by using Dynamic Bayesian Networks” IFAC Proceedings Volumes, vol. 42, no. 5, pp. 257-262, 2009.spa
dc.relation.referencesM. S. Kan, A. C. Tan, and J. Mathew, “A review on prognostic techniques for nonstationary and non-linear rotating systems” Mechanical Systems and Signal Processing, vol. 62, pp. 1-20, 2015.spa
dc.relation.referencesM. A. Zaidan, R. Relan, A. R. Mills, and R. F. Harrison, “Prognostics of gas turbine engine: An integrated approach” Expert Systems with Applications, vol. 42, pp. 8472-8483, dec 2015.spa
dc.relation.referencesH. P. Bloch and F. K. Geitner, “Metallurgical Failure Analysis” in Machinery Failure Analysis and Troubleshooting (H. P. B. Geitner and F. K., eds.), ch. Chapter 2, pp. 10-78, Gulf Professional Publishing, 1999.spa
dc.relation.referencesN. Waterman, “Failure analysis” Materials & Design, vol. 3, pp. 425-426, apr 1982.spa
dc.relation.referencesASM, “Asm Handbook - Vol 11 - Failure Analysis and Prevention.”spa
dc.relation.referencesS. S. Pelaseyed, F. Mashayekhi, and A. Movahedi-Rad, “Investigation of the Shaft Failure Connected to Extruder” Journal of Failure Analysis and Prevention, vol. 15, no. 6, pp. 775-781, 2015.spa
dc.relation.referencesP. S. Sell, Expert Systems - A Practical Introduction. London: Macmillan Education UK, 1985.spa
dc.relation.referencesM. Correa, C. Bielza, and J. Pamies-Teixeira, “Comparison of Bayesian networks and artificial neural networks for quality detection in a machining process” Expert Systems with Applications, vol. 36, pp. 7270-7279, apr 2009.spa
dc.relation.referencesH. E. Kyburg and J. Pearl, “Probabilistic Reasoning in Intelligent Systems: Networks of Plausible Inference.” The Journal of Philosophy, vol. 88, no. 8, p. 434, 1991.spa
dc.relation.referencesC. Miranda, Apunts d'inteligencia artificial. Barcelona: Universitat Politécnica de Catalunya, 1 ed., 2013.spa
dc.relation.referencesJ. T. P. Méndez and R. M. Morales, Inteligencia artificial Métodos, técnicas y aplicaciones. Madrid: McGRAW-HILL, 1 ed., 2008.spa
dc.relation.referencesE. Castillo, J. M. Gutiérrez, and A. S. Hadi, Expert Systems and Probabilistic Network Models. Monographs in Computer Science, New York, NY: Springer New York, 1997.spa
dc.relation.referencesG. F. Cooper, “The computational complexity of probabilistic inference using Bayesian belief networks” Artificial Intelligence, vol. 42, pp. 393-405, mar 1990.spa
dc.relation.referencesP. Dagum and M. Luby, “Approximating probabilistic inference in Bayesian belief networks is NP-hard” Artificial Intelligence, vol. 60, pp. 141-153, mar 1993.spa
dc.relation.referencesC. J. Butz, J. S. Oliveira, and A. L. Madsen, “Bayesian network inference using marginal Trees” International Journal of Approximate Reasoning, vol. 68, pp. 127-152, jan 2016.spa
dc.relation.referencesH. Broersma, E. Dahlhaus, and T. Kloks, “A linear time algorithm for minimum fill-in and treewidth for distance hereditary graphs” Discrete Applied Mathematics, vol. 99, pp. 367-400, feb 2000.spa
dc.relation.referencesN. Koller, Daphne and Friedman, Probabilistic graphical models: principles and techniques. London: MIT press, 2009.spa
dc.relation.referencesK.-L. Du and M. N. S. Swamy, Neural Networks and Statistical Learning. London: Springer London, 2014.spa
dc.relation.referencesCentro de Investigación sobre Sistemas Inteligentes de la UNED, “OpenMarkov” 2015.spa
dc.relation.referencesC. Ross, “TKInter” 2019.spa
dc.relation.referencesMunirAD, “GitHub - MunirAD/BayesNet: Implemented Variable Elimination algorithm for inference in Bayesian Network” 2016.spa
dc.relation.referencesThe PyMC Development Team, “PyMC3” 2018.spa
dc.rightsDerechos reservados - Universidad Nacional de Colombiaspa
dc.rights.accessrightsinfo:eu-repo/semantics/openAccessspa
dc.rights.licenseAtribución-NoComercial 4.0 Internacionalspa
dc.rights.spaAcceso abiertospa
dc.rights.urihttp://creativecommons.org/licenses/by-nc/4.0/spa
dc.subject.ddcIngeniería y operaciones afinesspa
dc.subject.ddcTecnología (Ciencias aplicadas)spa
dc.subject.proposalFailure analysiseng
dc.subject.proposalexpert systemeng
dc.subject.proposalinferencia bayesianaspa
dc.subject.proposalredes bayesianasspa
dc.subject.proposaleliminación de variablesspa
dc.subject.proposalMetropolis-hastingeng
dc.subject.proposalenumerationeng
dc.titleEvaluación del desempeño de tres algoritmos de inferencia bayesiana, implementados como sistema experto para la identificación de modos de falla en ejesspa
dc.title.alternativePerformance evaluation of three Bayesian inference algorithms, implemented as an expert system for the identification of shaft failure modes.spa
dc.typeDocumento de trabajospa
dc.type.coarhttp://purl.org/coar/resource_type/c_8042spa
dc.type.coarversionhttp://purl.org/coar/version/c_ab4af688f83e57aaspa
dc.type.contentTextspa
dc.type.driverinfo:eu-repo/semantics/workingPaperspa
dc.type.redcolhttp://purl.org/redcol/resource_type/WPspa
dc.type.versioninfo:eu-repo/semantics/acceptedVersionspa
oaire.accessrightshttp://purl.org/coar/access_right/c_abf2spa

Archivos

Bloque original

Mostrando 1 - 1 de 1
Cargando...
Miniatura
Nombre:
1032450351-2019.pdf
Tamaño:
8.17 MB
Formato:
Adobe Portable Document Format

Bloque de licencias

Mostrando 1 - 1 de 1
Cargando...
Miniatura
Nombre:
license.txt
Tamaño:
3.9 KB
Formato:
Item-specific license agreed upon to submission
Descripción: