Refine
Year of publication
Document Type
Way of publication
- Open Access (3)
Keywords
- Optimization (4)
- Optimisation (3)
- multi-objective optimization (3)
- optimization (3)
- Mehrkriterien-Optimierung (2)
- Simulation (2)
- robust design (2)
- 3D CFD Analysis (1)
- 3D-CFD (1)
- Aero Engines (1)
Institute
BTU
Deterministic optimization is used in all fields of engineering, especially in early design processes based on digital prototypes and simulation. A major disadvantage of deterministic optimization is the unknown robustness of the found solution against uncertainties of system parameters. Therefore, the Robust Design Optimization (RDO) concept as combination of Robustness Analysis (RA) and deterministic optimization was developed. In this paper, such an approach is applied to the optimization of a suspension for passenger cars w.r.t. typical driving maneuvers. The suspension behavior has to be robust against uncertainties without defining strict limits or safety margins. The coupled multi-objective RDO procedure will find a Pareto-front w.r.t. mean value and variance of chosen objectives. As a result, a specific compromise regarding system robustness and mean performance may be chosen from this Pareto-set. To overcome the vast amount of CPU-time, required for expensive direct function evaluations, an adaptive response surface method (aRSM) is integrated. The overall process then consists of an inner loop involving a multi-objective evolutionary algorithm based on response surfaces and an outer loop, where metamodeling is performed on a set of support points. This set is initialized in the first iteration step and updated afterwards by picking promising designs from the Pareto-fronts of the surrogate model, which are then evaluated exactly.
Transfer Matrix Method for the Determination of the Free Vobration of Two Eastically Coupled Beams
(2013)
Contradictory Integer Objective Approaches for Mixed-variables Evolutionary Optimization Problems
(2014)
System Integration as Key for Improving and Speeding up the Preliminary Design Phase of Aero Engines
(2013)
A design change from bladed disk to blisk leads to stronger endurance limits. The paper deals with aerofoil adaption of a blisk within a multi-criterion optimisation loop. Blisks are more sensitive to external excitation since mechanical damping is very low. Thus stronger limits have to be applied which are hard to be fulfilled and hence it requires the use of numerical optimisation. Combination of 2D CFD and FEM in a common optimisation loop to re-design a given blisk. Determination of relevant features for improving blisk endurance. Modern process integration and multi-criterion optimisation tools in combination with today's increased computational power allow to perform mulitphysics design even in the case of complex industrial design tasks.
Beschleunigung der Triebwerksauslegung durch Verwendung problemangepasster Isight-Komponenten
(2011)
Planetary gearboxes in highly sophisticated applications such as turbofan engines are required to have a high power-to-weight ratio and excellent reliability. Hence, thin-rimmed gear units need to be designed as compact as possible which, however, is usually limited by the tooth root load capacity. In order to come up with the best design, a tooth root shape optimization process is developed for thin-rimmed planet gears.
Zhe paper introduces ideas from shape optimization to multibody system dynamics. A disk rolling down a given slope is taken as a simple example, for which it is the goal of the optimization to shape the rolling contour of the disk such that it takes a minimum time to cover a certain distance. The shape of the sontour is described by its radius of curvature. The governing equations of motion result from the kinematics of relative motion and the Nowton-Euler formalism. Three different kinds of spirals are defined and optimized.
The identification of parameters in dynamic systems usually first requires modeling the system as nonlinear differential equations based on physical principles, which then can be evaluated to search for a set of optimal parameters that enable the mathematical model to reproduce some desired behavior of the real system. To overcome the necessity of complex modeling and to possibly reduce the number of extensive experimental evaluations of the real system or numerical evaluations of the differential equations, strategies from machine learning may be applied. The proposed method uses recurrent architectures such as Long Short-Term Memory (LSTM) networks as core to predict the system development from initial conditions and parameter values. The trainable weights of the model are optimized based on a set of training data containing parameter values and corresponding solution trajectories generated by evaluation of the system to be investigated. The trained model may then be used to identify unknown system parameter values related to a specific solution trajectory by solving an optimization problem for the inputs of the machine learning model.