This is a report on a study of parameters identification of mechanical constants of anisotropic rock mass and initial ground pressure from set of displacements on assumption that rock. ass is anisotropic continuous elastic medium. A local iterated extended Kalman filter combined by global iterated extended Kalman filter is applied to the parameters identification method in order to consider probabilistic feature of measured data. It is concluded that this method is effective for parameters identification of mechanical constants of rock mass with enough accuracy.
Ceci est rapport sur une etude d"identification parametrique des constantes mecaniques de la masse rocheuse anisotrope et de la pression initiale sur le sol à partir d"un jeu de deplacements dans la supposition que la masse rocheuse soit un medium elastique constinu anisotrope. Un filtre de Kalman localement iteratif non lineaire est utilise pour la methode d"identification parametrique afin de considerer les caracteristiques probabilistes des donnees mesurees. Il se conclut que cette methode est efficace pour d"identification parametrique des constantes mecaniques de la masse rocheuse avec une precision suffisante.
Dieser Bericht behandelt Errungenschaften einer Forschung, die durchgefuehrt wurde, um Parameter, die sich auf mechanische Konstanten einer anisotropen Felsenmasse und eines Ausgangsgrunddruckes beziehen, aus einer Reihe von Verschiebungen festzustellen unter der Annahme, dass die Felsen.asse ein anisotrop fortlaufendes elastisches Medium darstelle. Bei der Feststellung der Parameter durch diese Methode wurde ein sowohl lokal wie global iterativ ausgedehntes Kalman Filter angewandt, um wahrscheinliche Eigenschaften der ge.essen Daten zu erwaegen. Diese Forschung hat zur Ueberzeugung gefuehrt, dass diese Methode so wirksa. ist, dass die Parameter, die sich auf mechanische Konstanten einer Felsenmasse beziehen, mit zureichend hoher Genauigkeit festgestellt werden koennen.
Recently, back analysis methods have been proposed for identification of mechanical constants and initial ground pressure of rock mass on basis of measured data. This method enables review" of the conditions given at designing stage and the assessment of stability during excavation works "of rock mass structures. Many mechanical models used for the back analysis methods assume that rock mass is isotropic and linear elastic medium. However, recent research has revealed that rock mass becomes mechanically anisotropic depending on distributions and mechanical properties of various discontinuous planes present in the rock mass (Kyoya et al. 1985, Yamabe et al. 1987).
This paper describes a study of parameters identification of mechanical constants of anisotropic rock mass and initial ground pressure from set of displacements on condition that rock mass is anisotropic continuous elastic medium. Note -that identification problem in this paper is non-linear. As the identification technique for, this non-linear system, insted of the optimization technique for mathematical program.ing which handles observed data deterministically (Gioda 1984, Yeh 1985), a method using the iterated extended Kalman filter (Jazwinsky 1970) combined by conventional extended Kalman filter is proposed to handle measured data probabilistically. The proposed method is constructed by modifying the extended Kalman filter and adopting an algorithm to reduce the influence of non- lineality of the system and the difference of initial values on the back analysis solutions.
