56.03 Methoden im Bauingenieurwesen
Refine
Document Type
- Conference Proceeding (1)
- Master's Thesis (1)
Institute
Keywords
- Nichtlineare Optimierung (2) (remove)
A lot of real-life problems lead frequently to the solution of a complicated (large scale, multicriteria, unstable, nonsmooth etc.) nonlinear optimization problem. In order to cope with large scale problems and to develop many optimum plans a hiearchical approach to problem solving may be useful. The idea of hierarchical decision making is to reduce the overall complex problem into smaller and simpler approximate problems (subproblems) which may thereupon treated independently. One way to break a problem into smaller subproblems is the use of decomposition-coordination schemes. For finding proper values for coordination parameters in convex programming some rapidly convergent iterative methods are developed, their convergence properties and computational aspects are examined. Problems of their global implementation and polyalgorithmic approach are discussed as well.
In displacement oriented methods of structural mechanics may static and dynamic equilibrium conditions lead to large coupled nonlinear systems of equations. In many cases they are solved iteratively utilizing derivatives of Newton's method. Alternatively, the equations may be expressed in terms of the Karush-Kuhn-Tucker conditions of an optimization problem and, therefore, may be solved using methods of mathematical programming. To begin with, the work deals with the fundamentals of the formulation as optimization problem. In particular, the requirements of material nonlinearity and contact situations are analyzed. Proximately, an algorithm is implemented which utilizes the usually sparse structure of the Hessian matrix, whereby particularly the convergence behaviour is analyzed and adjusted. The implementation was tested using examples from statics and dynamics of large systems. The results are verified considering the accuracy comparing alternative solutions (e.g. explicit methods). The potential areas of application is shown and the efficiency of the method is evaluated.