1,721,018 research outputs found
Multidisciplinary and multi-fidelity optimization environment for wing integrated design
No full textThe Ph.D. program has been focused on the development of a multidisciplinary integrated environment for the design of wing for which large changes in shape are expected to be allowed during the flight in order to be better adapted for the different flight segments. The first phase of study has been dedicated to the investigation of the proper Multidisciplinary Design Optimization (MDO) architecture for the integrated management of the design process and a multilevel solution has been proposed and implemented. Such framework involves several disciplinary analysis and optimization loops: in particular aerodynamic analysis, structural analysis, material optimization and mission and performance evaluation are the main components considered for the preliminary design development for such a "morphing" wing. This stage addressed basically the multidisciplinarity and interdisciplinarity issues. The second phase has been dedicated to the investigation of possible techniques for the reduction of the computational burden that characterizes typically this kind of integrated design processes. For this purpose multi-fidelity analysis techniques have been considered involving the use of surrogate models. In particular the attention has been focused on the study of a proper methodology to build an approximated model for the estimation of aerodynamic coefficients to be used for performance evaluation in the mission optimization stage. In this case a procedure involving variables screening phase, data-fit surrogate models evaluation and assessment phase and a final crucial global correction phase of the best surrogate model has been propose
Real-time predictions of vehicle capabilities for reconfigurable mission planning
No full textReliability and operational availability of unmanned vehicles can be augmented through a dynamic reshaping of their operational and mission profile in response to the evolution of their health state and contingencies. In hazardous settings, the dynamic reconfiguration of a mission profile requires real-time predictions of residual capabilities which determine the set of feasible manoeuvres to preserve the vehicle and complete the mission successfully. This work discusses two computational frameworks to predict system capabilities from on-board sensor measurements and actualize a form of self-awareness for unmanned air vehicles in support of reconfigurable mission planning. The first framework relies on a traditional approach to diagnostics and prognostics: model reduction and supervised learning are combined to accelerate both the identification of damage parameters and the prediction of system capabilities. The second framework introduces a priority shift that emphasizes the prediction of vehicle capabilities over the characterization of the damage: an original bypass scheme (named MultiStep-ROM) combines projection-based model reduction and unsupervised machine learning into a form of transfer learning that computes adaptive models directly mapping measurements into capabilities. The two approaches are presented through the example cases of unmanned air vehicles that undergo failures of on-board actuation devices and structural damages. The computational experiments indicate that the bypass approach allows to obtain sensitively faster predictions of vehicle capabilities and is better suited to meet real-time responsiveness requirements than the traditional scheme
Structural assessment and sensor placement strategy for self-aware aerospace vehicles
No full textThis paper addresses real-time structural health assessment as a form of aircraft selfawareness. Limited time and resources available on-board and incomplete measured data affected by uncertainty are the key challenges to face. We discuss a data-driven methodology that combines Multi-Step Reduced Order Modeling to support structural self-awareness, and unsupervised learning (Self-Organizing Maps) to identify optimal sets of sensor locations. In particular, two implementations of our sensor placement strategy are presented and compared for a composite wing panel subjected to a number of damage conditions
Data to decisions: Real-time structural assessment from sparse measurements affected by uncertainty
No full textMultifidelity Optimization for Engineering Design: Space Application
No full textThe design and optimization of space engineering systems requires the implementation of costly high-fidelity models capable to accurately represent complex physical phenomena. Examples are computational fluid dynamic models for the numerical solution of partial differential equations that permit to capture the aerothermodynamic interaction during the re-entry of a space vehicle from an interplanetary transfer. However, simulation-based optimization usually requires a large number of model evaluations during the search for the optimal design, which makes the use of high-fidelity models unfeasible due to their high computational cost. To address this challenge, we discuss a multifidelity strategy for the design and optimization of complex systems capable to combine multiple models at different levels of fidelity in order to contain the computational cost and achieve a better design solution. We propose a strategy for multifidelity active learning that leverages low fidelity models to explore design configurations and refines the quality of the design solution through the principled query of the high-fidelity model. The active learning scheme is formulated to merge data-driven and domain-aware sources of information and is implemented for the multidisciplinary design optimization of an Orion re-entry capsule. The optimization goals are the minimization of the propellant mass burned during the re entry, the minimization of the structural mass of the thermal protection system and the minimization of the temperature reached by the heat shield, all referred to the baseline design configuration. The results illustrate that our multifidelity framework leads to a design improvement of the 15% with respect to the baseline solution with a fraction (7%) of the overall computational cost that would be required by a single-fidelity optimization based on high-fidelity models only
Domain-Aware Active Learning for Multifidelity Optimization
Full text linkBayesian optimization is a popular strategy for the optimization of black-box objective functions [1]. In many engineering applications, the objective can be evaluated with multiple representations at different levels of fidelity, to enhance a trade-off between cost and accuracy. Accordingly, multifidelity methods have been proposed in a Bayesian framework to efficiently combine information sources, using low-fidelity models to enable the exploration of design alternatives, and improve the accuracy of the solution through limited high-fidelity evaluations [2]. Most multifidelity methods based on active learning search the optimal design considering only the information extracted from the surrogate model. This can preclude the evaluation of promising design configurations that can be captured only including the knowledge of the particular physical phenomena involved [3]. To address this issue, this presentation discusses original domain-aware multifidelity Bayesian frameworks to accelerate design analysis and optimization performances. In particular, our strategy comes with an active learning scheme to adaptively sample the design space, combining statistical data from the surrogate model with physical information from the specific domain. Our formulation introduces physics-informed utility functions as additional contributions to the acquisition functions. This permits to enhance the active learning with a physicsbased insight and to realize a form of domain awareness which is beneficial to the efficiency and accuracy of the optimization task. The presentation will discuss several applications and implementations of the proposed approach for single discipline and multidisciplinary aerospace design optimization problems.
[1] Snoek, J., Larochelle, H.. Adams, R.P. Practical bayesian optimization of machine learning algorithms. Advances in neural information processing systems. (2012) 25.
[2] Peherstorfer, B., Willcox, K., Gunzburger, M. Survey of multifidelity methods in uncertainty propagation, inference, and optimization. Siam Review (2018) 60(3): 550–591.
[3] Di Fiore, F., Maggiore, P. Mainini L. Multifidelity domain-aware learning for the design of re-entry vehicles. Structural and Multidisciplinary Optimization (2021
Approximated models for aerodynamic coefficients estimation in a multidisciplinary design environment
Full text linkIn this paper variable fidelity analyses are investigated. Moreover different kind of approximations to be used in a wide multidisciplinary design environment for aircraft design are built. In order to obtain the surrogate models used in the main design process, a proper framework is built by different design of experiments techniques for process and variables management. Approximated models for the estimation of aerodynamic coefficients are evaluated on design spaces of different dimensions and considering different set of variables (i.e. geometric parameters and flight conditions). They are mainly based on the hybrid combination of Vortex Lattice Method (VLM) models
representing the basic low fidelity analysis) and 3D finite volume Computational Fluid Dynamics models (representing the basic high fidelity analysis). Different strategies for the evaluation of the surrogate model are considered and an original methodology for the model construction is here presented
- …
