Nos Abonnements
Ressources Électroniques
Détail de l'auteur
Auteur Jin-Song Pei |
Documents disponibles écrits par cet auteur (5)
Ajouter le résultat dans votre panier Faire une suggestion Affiner la rechercheDeterministic excitation forces for simulation and identification of nonlinear hysteretic SDOF systems / Jin-Song Pei in Journal of engineering mechanics, Vol. 134 N°1 (Janvier 2008)
Titre : Deterministic excitation forces for simulation and identification of nonlinear hysteretic SDOF systems Type de document : texte imprimé Auteurs : Jin-Song Pei, Auteur ; Krisda Piyawat, Auteur Année de publication : 2008 Article en page(s) : pp.35–48. Note générale : Mécanique appliquée Langues : Anglais (eng) Mots-clés : Excitation Simulation Hysteresis Stiffness Degradation Résumé : This paper investigates how to design deterministic excitation forces in studying nonlinear single-degree-of-freedom systems, especially those with rate and path dependency and strength and stiffness degradation. One frequency-modulated periodic excitation and its amplitude-modulated counterpart are proposed as a solution, and a series of numerical exercises are carried out to show that these forces can be designed for sufficient forcing functions to study the complex nonlinear hysteresis. To rapidly reveal the underlying characteristics of the system and also to further lead to an effective system identification, four evaluation tools are proposed to be utilized together with the proposed excitation forces. These tools include the response curves, force-state map, intercycle drift, and intercycle pattern change, based on which some distinctive “patterns” are obtained to reveal the existence of nonlinearities, types of nonlinearities, existence of memory, and degradation. By using both Bouc-Wen and Bouc-Wen-Baber-Noori models for the system in all the simulations, the writers compare the commonly used forces with the proposed excitation forces to further demonstrate the advantages of the proposed excitation forces and evaluation tools. The writers also explore challenges in terms of implementing the proposed excitation forces. The results of this study are expected to benefit both physical testing and numerical simulation of complex nonlinear hysteretic systems in a time- and cost-effective manner, as well as leading to efficient schemes for system identification. ISSN : 0733-9399 En ligne : http://ascelibrary.org/doi/abs/10.1061/%28ASCE%290733-9399%282008%29134%3A1%2835 [...]
in Journal of engineering mechanics > Vol. 134 N°1 (Janvier 2008) . - pp.35–48.[article] Deterministic excitation forces for simulation and identification of nonlinear hysteretic SDOF systems [texte imprimé] / Jin-Song Pei, Auteur ; Krisda Piyawat, Auteur . - 2008 . - pp.35–48.
Mécanique appliquée
Langues : Anglais (eng)
in Journal of engineering mechanics > Vol. 134 N°1 (Janvier 2008) . - pp.35–48.
Mots-clés : Excitation Simulation Hysteresis Stiffness Degradation Résumé : This paper investigates how to design deterministic excitation forces in studying nonlinear single-degree-of-freedom systems, especially those with rate and path dependency and strength and stiffness degradation. One frequency-modulated periodic excitation and its amplitude-modulated counterpart are proposed as a solution, and a series of numerical exercises are carried out to show that these forces can be designed for sufficient forcing functions to study the complex nonlinear hysteresis. To rapidly reveal the underlying characteristics of the system and also to further lead to an effective system identification, four evaluation tools are proposed to be utilized together with the proposed excitation forces. These tools include the response curves, force-state map, intercycle drift, and intercycle pattern change, based on which some distinctive “patterns” are obtained to reveal the existence of nonlinearities, types of nonlinearities, existence of memory, and degradation. By using both Bouc-Wen and Bouc-Wen-Baber-Noori models for the system in all the simulations, the writers compare the commonly used forces with the proposed excitation forces to further demonstrate the advantages of the proposed excitation forces and evaluation tools. The writers also explore challenges in terms of implementing the proposed excitation forces. The results of this study are expected to benefit both physical testing and numerical simulation of complex nonlinear hysteretic systems in a time- and cost-effective manner, as well as leading to efficient schemes for system identification. ISSN : 0733-9399 En ligne : http://ascelibrary.org/doi/abs/10.1061/%28ASCE%290733-9399%282008%29134%3A1%2835 [...] Exemplaires
Code-barres Cote Support Localisation Section Disponibilité aucun exemplaire
Titre : Linking Nonlinear System Identification with Nonlinear dynamic simulation under openSees : some justifications and implementations Type de document : texte imprimé Auteurs : Krisda Piyawat, Auteur ; Jin-Song Pei, Auteur Année de publication : 2009 Article en page(s) : pp. 1213-1226 Note générale : Mécanique appliquée Langues : Anglais (eng) Mots-clés : Nonlinear systems Simulation Implementation Experimentation. Résumé : This study seeks to bridge the gap between nonlinear system identification and nonlinear dynamic finite-element analysis. Motivated by the needs in earthquake simulation, it is first investigated under which conditions and to what degree the prediction of maximum lateral drift and base shear requires accurate nonlinear hysteretic moment-rotation joint models. A series of simulations is carried out using a simple but typical steel frame under two different earthquake ground motion time histories scaled up to various levels. As one of the two major classes of models in nonlinear system identification, nonparametric models are proposed to be implemented into OpenSees. A methodology with details is provided to effectively implement feedforward neural networks with one hidden layer as a new one-dimensional nonlinear smooth material model directly from a MATLAB environment to OpenSees. The same methodology can be applied to benefit the implementation of other parametric and nonparametric models with linear parameterization. Numerical examples are provided. Challenges are discussed and future work is identified.
DEWEY : 620.1 RAMEAU : 0733-9399 En ligne : http://ascelibrary.aip.org/vsearch/servlet/VerityServlet?KEY=JENMDT&ONLINE=YES&s [...]
in Journal of engineering mechanics > Vol. 135 N° 11 (Novembre 2009) . - pp. 1213-1226[article] Linking Nonlinear System Identification with Nonlinear dynamic simulation under openSees : some justifications and implementations [texte imprimé] / Krisda Piyawat, Auteur ; Jin-Song Pei, Auteur . - 2009 . - pp. 1213-1226.
Mécanique appliquée
Langues : Anglais (eng)
in Journal of engineering mechanics > Vol. 135 N° 11 (Novembre 2009) . - pp. 1213-1226
Mots-clés : Nonlinear systems Simulation Implementation Experimentation. Résumé : This study seeks to bridge the gap between nonlinear system identification and nonlinear dynamic finite-element analysis. Motivated by the needs in earthquake simulation, it is first investigated under which conditions and to what degree the prediction of maximum lateral drift and base shear requires accurate nonlinear hysteretic moment-rotation joint models. A series of simulations is carried out using a simple but typical steel frame under two different earthquake ground motion time histories scaled up to various levels. As one of the two major classes of models in nonlinear system identification, nonparametric models are proposed to be implemented into OpenSees. A methodology with details is provided to effectively implement feedforward neural networks with one hidden layer as a new one-dimensional nonlinear smooth material model directly from a MATLAB environment to OpenSees. The same methodology can be applied to benefit the implementation of other parametric and nonparametric models with linear parameterization. Numerical examples are provided. Challenges are discussed and future work is identified.
DEWEY : 620.1 RAMEAU : 0733-9399 En ligne : http://ascelibrary.aip.org/vsearch/servlet/VerityServlet?KEY=JENMDT&ONLINE=YES&s [...] Exemplaires
Code-barres Cote Support Localisation Section Disponibilité aucun exemplaire
Titre : New Approach to Designing Multilayer Feedforward Neural Network Architecture for Modeling Nonlinear Restoring Forces. I: Formulation Titre original : Nouvelle Approche à Concevoir l'Architecture Alimentée-vers l'Avant Multi-Couche de Réseau Neurologique pour Modeler les Forces de Reconstitution non linéaires. I : Formulation Type de document : texte imprimé Auteurs : Jin-Song Pei, Auteur ; Schueller, Gerhart L., Éditeur scientifique ; Andrew W. Smyth, Auteur Année de publication : 2007 Article en page(s) : 1290-1300 p. Note générale : Génie Mécanique Langues : Anglais (eng) Mots-clés : Neural networks Mathematics Geometry Models Hysteresis Réseaux neurologiques Géométrie Mathématiques Modèles Hystérésis Index. décimale : 620.1 Essais des matériaux. Défauts des matériaux. Protection des matériaux Résumé : This paper addresses the modeling problem of nonlinear and hysteretic dynamic behaviors through a constructive modeling approach which exploits existing mathematical concepts in artificial neural network modeling. In contrast with many neural network applications, which often result in large and complex “black-box” models, here, the writers strive to produce phenomenologically accurate model behavior starting with network architecture of manageable/small sizes. This affords the potential of creating relationships between model parameter values and observed phenomenological behaviors. Here a linear sum of basis functions is used in modeling nonlinear hysteretic restoring forces. In particular, nonlinear sigmoidal activation functions are chosen as the core building block for their robustness in approximating arbitrary functions. The appropriateness and effectiveness of this set of basis function in modeling a wide variety of nonlinear dynamic behaviors observed in structural mechanics are depicted from an algebraic and geometric perspective.
Cet article adresse le problème modelant des comportements dynamiques non linéaires et par hystérésis par une approche modelante constructive qui exploite des concepts mathématiques existants dans modeler artificiel de réseau neurologique. Contrairement à beaucoup d'applications de réseau neurologique, qui ont souvent comme conséquence de grands et complexes modèles de "boîte-noire", ici, les auteurs tâchent de produire le comportement modèle logiquement précis de phenomeno commençant par l'architecture de réseau de petites tailles maniables. Ceci a les moyens le potentiel de créer des rapports entre les valeurs de paramètre modèles et les comportements phénoménologiques observés. Ici une somme linéaire de fonctions de base est employée en modelant les forces de reconstitution par hystérésis non linéaires. En particulier, des fonctions sigmoïdales non linéaires d'activation sont choisies comme module de noyau pour leur robustesse en rapprochant des fonctions arbitraires. La convenance et l'efficacité de cet ensemble de fonction de base en modelant une grande variété de comportements dynamiques non-linéaires observés dans la mécanique structurale sont dépeintes d'une perspective algébrique et géométrique.
DEWEY : 620.1 ISSN : 0733-9399 En ligne : smyth@civil.columbia.edu
in Journal of engineering mechanics > Vol. 132 N°12 (Decembre 2006) . - 1290-1300 p.[article] New Approach to Designing Multilayer Feedforward Neural Network Architecture for Modeling Nonlinear Restoring Forces. I: Formulation = Nouvelle Approche à Concevoir l'Architecture Alimentée-vers l'Avant Multi-Couche de Réseau Neurologique pour Modeler les Forces de Reconstitution non linéaires. I : Formulation [texte imprimé] / Jin-Song Pei, Auteur ; Schueller, Gerhart L., Éditeur scientifique ; Andrew W. Smyth, Auteur . - 2007 . - 1290-1300 p.
Génie Mécanique
Langues : Anglais (eng)
in Journal of engineering mechanics > Vol. 132 N°12 (Decembre 2006) . - 1290-1300 p.
Mots-clés : Neural networks Mathematics Geometry Models Hysteresis Réseaux neurologiques Géométrie Mathématiques Modèles Hystérésis Index. décimale : 620.1 Essais des matériaux. Défauts des matériaux. Protection des matériaux Résumé : This paper addresses the modeling problem of nonlinear and hysteretic dynamic behaviors through a constructive modeling approach which exploits existing mathematical concepts in artificial neural network modeling. In contrast with many neural network applications, which often result in large and complex “black-box” models, here, the writers strive to produce phenomenologically accurate model behavior starting with network architecture of manageable/small sizes. This affords the potential of creating relationships between model parameter values and observed phenomenological behaviors. Here a linear sum of basis functions is used in modeling nonlinear hysteretic restoring forces. In particular, nonlinear sigmoidal activation functions are chosen as the core building block for their robustness in approximating arbitrary functions. The appropriateness and effectiveness of this set of basis function in modeling a wide variety of nonlinear dynamic behaviors observed in structural mechanics are depicted from an algebraic and geometric perspective.
Cet article adresse le problème modelant des comportements dynamiques non linéaires et par hystérésis par une approche modelante constructive qui exploite des concepts mathématiques existants dans modeler artificiel de réseau neurologique. Contrairement à beaucoup d'applications de réseau neurologique, qui ont souvent comme conséquence de grands et complexes modèles de "boîte-noire", ici, les auteurs tâchent de produire le comportement modèle logiquement précis de phenomeno commençant par l'architecture de réseau de petites tailles maniables. Ceci a les moyens le potentiel de créer des rapports entre les valeurs de paramètre modèles et les comportements phénoménologiques observés. Ici une somme linéaire de fonctions de base est employée en modelant les forces de reconstitution par hystérésis non linéaires. En particulier, des fonctions sigmoïdales non linéaires d'activation sont choisies comme module de noyau pour leur robustesse en rapprochant des fonctions arbitraires. La convenance et l'efficacité de cet ensemble de fonction de base en modelant une grande variété de comportements dynamiques non-linéaires observés dans la mécanique structurale sont dépeintes d'une perspective algébrique et géométrique.
DEWEY : 620.1 ISSN : 0733-9399 En ligne : smyth@civil.columbia.edu Exemplaires
Code-barres Cote Support Localisation Section Disponibilité aucun exemplaire
Titre : New Approach to Designing Multilayer Feedforward Neural Network Architecture for Modeling Nonlinear Restoring Forces. II: Applications Titre original : Nouvelle Approche à Concevoir l'Architecture Multicouche de Réseau Neurologique d'Alimentation en Avant pour Modeler les Forces de Reconstitution Non Linéaires. II : Applications Type de document : texte imprimé Auteurs : Jin-Song Pei, Auteur ; Andrew W. Smyth, Auteur ; Schueller, Gerhart L., Éditeur scientifique Année de publication : 2007 Article en page(s) : 1301-1312 p. Note générale : Génie Mécanique Langues : Anglais (eng) Mots-clés : Neural networks Dynamic response Models Parameters Design Réseaux neurologiques Réponse dynamique Modèles Paramètres Conception Index. décimale : 620.1 Essais des matériaux. Défauts des matériaux. Protection des matériaux Résumé : Based on the basic formulation developed in a companion paper, the writers now present the application of an artificial neural network approach to designing streamlined network models to simulate and identify the nonlinear dynamic response of single-degree-of-freedom oscillators using the restoring force-state mapping interpretation. The neural networks which use sigmoidal activation functions are shown to be highly robust in modeling a wide variety of commonly observed nonlinear structural dynamic response behaviors. By streamlining the networks, individual network model parameters take on physically or geometrically interpretable meaning, and hence, the network initialization can be achieved through an engineered approach rather than through less physically meaningful numerical initialization schemes. Although not proven in general, examples show that by starting with a more meaningful initial design, identification convergence is improved, and the final identified model parameters are seen to have a more physical meaning. A set of model architecture prototypes is developed to capture commonly observed nonlinear response behaviors.
Basé sur la formulation de base qui s'est développé dans un papier de compagnon, les auteurs présentent maintenant l'application d'une approche artificielle de réseau neurologique à concevoir sur les modèles profilés de réseau pour simuler et identifier la réponse dynamique non-linéaire des oscillateurs de simple-degré-de-liberté en utilisant l'force-état de reconstitution traçant l'interprétation. Les réseaux neurologiques qui emploient des fonctions sigmoïdales d'activation s'avèrent fortement robustes en modelant une grande variété de comportements structuraux non linéaires généralement observés de réponse dynamique. Par le jet rayant les réseaux, les différents paramètres de modèle de réseau prennent dessus la signification physiquement ou géométriquement interprétable, et par conséquent, l'initialisation de réseau peut être réalisée par une approche machinée plutôt que par des arrangements numériques moins physiquement signicatifs d'initialisation. Bien que non avéré en général, les exemples montrent cela près commençant par une conception initiale plus signicative, la convergence d'identification est améliorée, et les paramètres modèles identifiés finals sont vus pour avoir une signification plus physique. Un ensemble de prototypes modèles d'architecture est développé pour capturer des comportements non linéaires généralement observés de réponse.
DEWEY : 620.1 ISSN : 0733-9399 En ligne : smyth@civil.columbia.edu
in Journal of engineering mechanics > Vol. 132 N°12 (Decembre 2006) . - 1301-1312 p.[article] New Approach to Designing Multilayer Feedforward Neural Network Architecture for Modeling Nonlinear Restoring Forces. II: Applications = Nouvelle Approche à Concevoir l'Architecture Multicouche de Réseau Neurologique d'Alimentation en Avant pour Modeler les Forces de Reconstitution Non Linéaires. II : Applications [texte imprimé] / Jin-Song Pei, Auteur ; Andrew W. Smyth, Auteur ; Schueller, Gerhart L., Éditeur scientifique . - 2007 . - 1301-1312 p.
Génie Mécanique
Langues : Anglais (eng)
in Journal of engineering mechanics > Vol. 132 N°12 (Decembre 2006) . - 1301-1312 p.
Mots-clés : Neural networks Dynamic response Models Parameters Design Réseaux neurologiques Réponse dynamique Modèles Paramètres Conception Index. décimale : 620.1 Essais des matériaux. Défauts des matériaux. Protection des matériaux Résumé : Based on the basic formulation developed in a companion paper, the writers now present the application of an artificial neural network approach to designing streamlined network models to simulate and identify the nonlinear dynamic response of single-degree-of-freedom oscillators using the restoring force-state mapping interpretation. The neural networks which use sigmoidal activation functions are shown to be highly robust in modeling a wide variety of commonly observed nonlinear structural dynamic response behaviors. By streamlining the networks, individual network model parameters take on physically or geometrically interpretable meaning, and hence, the network initialization can be achieved through an engineered approach rather than through less physically meaningful numerical initialization schemes. Although not proven in general, examples show that by starting with a more meaningful initial design, identification convergence is improved, and the final identified model parameters are seen to have a more physical meaning. A set of model architecture prototypes is developed to capture commonly observed nonlinear response behaviors.
Basé sur la formulation de base qui s'est développé dans un papier de compagnon, les auteurs présentent maintenant l'application d'une approche artificielle de réseau neurologique à concevoir sur les modèles profilés de réseau pour simuler et identifier la réponse dynamique non-linéaire des oscillateurs de simple-degré-de-liberté en utilisant l'force-état de reconstitution traçant l'interprétation. Les réseaux neurologiques qui emploient des fonctions sigmoïdales d'activation s'avèrent fortement robustes en modelant une grande variété de comportements structuraux non linéaires généralement observés de réponse dynamique. Par le jet rayant les réseaux, les différents paramètres de modèle de réseau prennent dessus la signification physiquement ou géométriquement interprétable, et par conséquent, l'initialisation de réseau peut être réalisée par une approche machinée plutôt que par des arrangements numériques moins physiquement signicatifs d'initialisation. Bien que non avéré en général, les exemples montrent cela près commençant par une conception initiale plus signicative, la convergence d'identification est améliorée, et les paramètres modèles identifiés finals sont vus pour avoir une signification plus physique. Un ensemble de prototypes modèles d'architecture est développé pour capturer des comportements non linéaires généralement observés de réponse.
DEWEY : 620.1 ISSN : 0733-9399 En ligne : smyth@civil.columbia.edu Exemplaires
Code-barres Cote Support Localisation Section Disponibilité aucun exemplaire
Titre : Solving dynamical systems involving piecewise restoring force using state event location Type de document : texte imprimé Auteurs : Joseph P. Wright, Auteur ; Jin-Song Pei, Auteur Année de publication : 2012 Article en page(s) : pp.997–1020. Note générale : Mécanique appliquée Langues : Anglais (eng) Mots-clés : Nonlinear hysteretic restoring force Differential-algebraic equations Piecewise solutions Discontinuities Explicit methods Adaptive Runge-Kutta State event location Résumé : Many theoretical and experimental studies of complex path-dependent dynamic systems lead to restoring forces expressed as piecewise nonlinear algebraic equations. Examples include, but are not limited to, bilinear hysteretic, Ramberg-Osgood, Masing, generalized Masing, Clough, and Takeda models, which are popular in engineering mechanics applications. These models relate restoring force to displacement and velocity by means of piecewise relations having only continuity, which leads to two sorts of challenges in numerical simulation. First, the equations of motion may not simply be a set of ordinary differential equations, rather they may fall within the framework of differential-algebraic equations (DAEs). Second, there are unknown locations of discontinuities of low-order derivatives of the solution. This study seeks accurate and efficient numerical solutions of the DAEs with continuity, enabling robust simulation of these complex nonlinear dynamic systems. This study focuses on explicit time integration for single degree-of-freedom problems, while presenting a suitable problem formulation, detailed guidelines, case studies, and convincing insights, while exploiting two built-in MATLAB functions (ode45.m and the Events option). User-defined options are carefully examined, and recommendations are made based on a systematic study of approximation accuracy and computational efficiency, particularly as they relate to global error and tolerance proportionality when using an explicit, adaptive Runge-Kutta (RK) solver. Obtaining accurate values of state event locations results in a robust approach to solving the identified class of problems. This work initiates the possibility of treating many similar models by using the proposed programming module and, more importantly, by applying and further advancing the underlying theoretical concepts. ISSN : 0733-9399 En ligne : http://ascelibrary.org/doi/abs/10.1061/%28ASCE%29EM.1943-7889.0000404
in Journal of engineering mechanics > Vol. 138 N° 8 (Août 2012) . - pp.997–1020.[article] Solving dynamical systems involving piecewise restoring force using state event location [texte imprimé] / Joseph P. Wright, Auteur ; Jin-Song Pei, Auteur . - 2012 . - pp.997–1020.
Mécanique appliquée
Langues : Anglais (eng)
in Journal of engineering mechanics > Vol. 138 N° 8 (Août 2012) . - pp.997–1020.
Mots-clés : Nonlinear hysteretic restoring force Differential-algebraic equations Piecewise solutions Discontinuities Explicit methods Adaptive Runge-Kutta State event location Résumé : Many theoretical and experimental studies of complex path-dependent dynamic systems lead to restoring forces expressed as piecewise nonlinear algebraic equations. Examples include, but are not limited to, bilinear hysteretic, Ramberg-Osgood, Masing, generalized Masing, Clough, and Takeda models, which are popular in engineering mechanics applications. These models relate restoring force to displacement and velocity by means of piecewise relations having only continuity, which leads to two sorts of challenges in numerical simulation. First, the equations of motion may not simply be a set of ordinary differential equations, rather they may fall within the framework of differential-algebraic equations (DAEs). Second, there are unknown locations of discontinuities of low-order derivatives of the solution. This study seeks accurate and efficient numerical solutions of the DAEs with continuity, enabling robust simulation of these complex nonlinear dynamic systems. This study focuses on explicit time integration for single degree-of-freedom problems, while presenting a suitable problem formulation, detailed guidelines, case studies, and convincing insights, while exploiting two built-in MATLAB functions (ode45.m and the Events option). User-defined options are carefully examined, and recommendations are made based on a systematic study of approximation accuracy and computational efficiency, particularly as they relate to global error and tolerance proportionality when using an explicit, adaptive Runge-Kutta (RK) solver. Obtaining accurate values of state event locations results in a robust approach to solving the identified class of problems. This work initiates the possibility of treating many similar models by using the proposed programming module and, more importantly, by applying and further advancing the underlying theoretical concepts. ISSN : 0733-9399 En ligne : http://ascelibrary.org/doi/abs/10.1061/%28ASCE%29EM.1943-7889.0000404 Exemplaires
Code-barres Cote Support Localisation Section Disponibilité aucun exemplaire
