An inverse problem via cross-entropy method for calibration of a drill string torsional dynamic model

Eber Dantas (1) , Americo Cunha Jr (1) , Francisco Soeiro (1) , Bruno Cesar Cayres (2) , Hans Ingo Weber (3)

1 UERJ - Universidade do Estado do Rio de Janeiro [Brasil] = Rio de Janeiro State University [Brazil] = Université d'État de Rio de Janeiro [Brésil]
2 CEFET/RJ - Centro Federal de Educação Tecnológica Celso Suckow da Fonseca [Rio de Janeiro]
3 PUC-Rio - Pontifícia Universidade Católica do Rio de Janeiro [Brasil] = Pontifical Catholic University of Rio de Janeiro [Brazil] = Université catholique pontificale de Rio de Janeiro [Brésil]

Model selection and parameter identification are challenging tasks in drill string dynamics due to the high degree of nonlinearity that abound the diverse and complex mechanisms involved. This work explores the application of a stochastic metaheuristic procedure for parameter identification over the torsional mode of drill string vibration. A proposed model is calibrated with data from a validated experimental setup , adjusting stiffness, damping and friction parameters. The resulting simulations display a reasonable fit with almost ideal correlation coefficients even in the presence of stick-slip. The optimization strategy is compared with an Genetic Algorithm, revealing significantly greater efficiency and showcasing how the cross-entropy method may be a viable tool in the demanding context of drill string modeling.

Mots clés: drillstring dynamics; inverse problem; nonlinear dynamics; parameter identification; cross-entropy method

Domaines: Vibrations [physics.class-ph]; Probabilités [math.PR]; Statistiques [math.ST]; Mécanique des structures [physics.class-ph]; Mécanique des solides [physics.class-ph]; Génie mécanique [physics.class-ph]

Eber Dantas, Americo Cunha Jr, Francisco Soeiro, Bruno Cesar Cayres, Hans Ingo Weber. An inverse problem via cross-entropy method for calibration of a drill string torsional dynamic model. 25th ABCM International Congress on Mechanical Engineering (COBEM 2019), Oct 2019, Uberlândia, Brazil. ⟨hal-02388463⟩