Search NASA⌕ Search

SEARCH · Search NASA

Results for “Robust optimization”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18

Accounting for the tongue-and-groove effect using a robust direct aperture optimization approach

Purpose: Traditionally, the tongue-and-groove effect due to the multileaf collimator architecture in intensity-modulated radiation therapy (IMRT) has typically been deferred to the leaf sequencing stage. The authors propose a new direct aperture optimization method for IMRT treatment planning that explicitly incorporates dose calculation inaccuracies due to the tongue-and-groove effect into the treatment plan optimization stage. Methods: The authors avoid having to accurately estimate the dosimetric effects of the tongue-and-groove architecture by using lower and upper bounds on the dose distribution delivered to the patient. They then develop a model that yields a treatment plan that is robust with respect to the corresponding dose calculation inaccuracies. Results: Tests on a set of ten clinical head-and-neck cancer cases demonstrate the effectiveness of the new method in developing robust treatment plans with tight dose distributions in targets and critical structures. This is contrasted with the very loose bounds on the dose distribution that are obtained by solving a traditional treatment plan optimization model that ignores tongue-and-groove effects in the treatment planning stage. Conclusions: A robust direct aperture optimization approach is proposed to account for the dosimetric inaccuracies caused by the tongue-and-groove effect. The experiments validate the ability of the proposed approach in designing robust treatment plans regardless of the exact consequences of the tongue-and-groove architecture.

60 APPLIED LIFE SCIENCES↗

Surrogate Model Based Optimization for Finding Robust Deep Learning Model Architectures

Deep Learning (DL) models are increasingly used throughout the sciences. However, their performance and usefulness depend greatly on their architecture which is defined by hyperparameters such as the number of nodes, layers, the learning rate, etc. Tuning these hyperparameters is time-consuming because evaluating their performance requires a lengthy training step. Stochastic optimizers used in training lead to performance variability and potentially prediction reliability issues. In this talk, we will describe an automated optimization method based on surrogate models and active learning strategies for tuning DL model architectures. We take into account the prediction variability with the goal to identify architectures that make reliable and robust predictions. We demonstrate our developments on an application arising in particle physics.

deep learning↗

Towards Robust Designs Via Multiple-Objective Optimization Methods

Fabricating and operating complex systems involves dealing with uncertainty in the relevant variables. In the case of aircraft, flow conditions are subject to change during operation. Efficiency and engine noise may be different from the expected values because of manufacturing tolerances and normal wear and tear. Engine components may have a shorter life than expected because of manufacturing tolerances. In spite of the important effect of operating- and manufacturing-uncertainty on the performance and expected life of the component or system, traditional aerodynamic shape optimization has focused on obtaining the best design given a set of deterministic flow conditions. Clearly it is important to both maintain near-optimal performance levels at off-design operating conditions, and, ensure that performance does not degrade appreciably when the component shape differs from the optimal shape due to manufacturing tolerances and normal wear and tear. These requirements naturally lead to the idea of robust optimal design wherein the concept of robustness to various perturbations is built into the design optimization procedure. The basic ideas involved in robust optimal design will be included in this lecture. The imposition of the additional requirement of robustness results in a multiple-objective optimization problem requiring appropriate solution procedures. Typically the costs associated with multiple-objective optimization are substantial. Therefore efficient multiple-objective optimization procedures are crucial to the rapid deployment of the principles of robust design in industry. Hence the companion set of lecture notes (Single- and Multiple-Objective Optimization with Differential Evolution and Neural Networks ) deals with methodology for solving multiple-objective Optimization problems efficiently, reliably and with little user intervention. Applications of the methodologies presented in the companion lecture to robust design will be included here. The evolutionary method (DE) is first used to solve a relatively difficult problem in extended surface heat transfer wherein optimal fin geometries are obtained for different safe operating base temperatures. The objective of maximizing the safe operating base temperature range is in direct conflict with the objective of maximizing fin heat transfer. This problem is a good example of achieving robustness in the context of changing operating conditions. The evolutionary method is then used to design a turbine airfoil; the two objectives being reduced sensitivity of the pressure distribution to small changes in the airfoil shape and the maximization of the trailing edge wedge angle with the consequent increase in airfoil thickness and strength. This is a relevant example of achieving robustness to manufacturing tolerances and wear and tear in the presence of other objectives.

Man Mohan, Rai↗

Stochastic Simplex Approximate Gradient for Robust Life-cycle Production Optimization: Applied to Brugge Field

In oil and gas production industry, production optimization is a viable technique to maximize the oil/gas production or net present value (NPV). Robust optimization is one type of production optimization techniques where the geological uncertainty of reservoir is considered. When well operating conditions, e.g., well flow rates and bottom-hole pressures, are the optimization variables, ensemble-based optimization (EnOpt) is the most popular ensemble-based algorithm for robust life-cycle production optimization. Recently, however, a superior algorithm, stochastic simplex approximate gradient (StoSAG), was proposed. Finally, the purpose of this study is to provide a theoretical discussion on why StoSAG is generally superior to EnOpt and provide reasonable example (Brugge field) where StoSAG generates estimates of optimal well controls that give a life-cycle net present value (NPV) significantly higher than the NPV obtained from EnOpt.

02 PETROLEUM↗

Numerical derivative techniques for trajectory optimization

The adoption of robust numerical optimization techniques in trajectory simulation programs has resulted in powerful design and analysis tools. These trajectory simulation/optimization programs are widely used, and a representative list includes the GTS system, the POST program, and newer collocation methods such as OTIS and FONPAC. All of these programs rely on optimization algorithms which require objective function and constraint gradient data during the iteration process. However, most trajectory optimization problems lack simple analytical expressions for these derivatives. In the general case a function evaluation involves integrating aerodynamic, propulsive, and gravity forces over multiple trajectory phases with complex control models. With the newer collocation methods, the integration is replaced by defect constraints and cubic approximations for the state. While analytic gradient expressions can sometimes be derived for trajectory optimization problems, the derivation is cumbersome, time consuming, and prone to mistakes. Fortunately, an alternate method exists for the gradient evaluation, namely finite difference approximations. In this paper some finite difference gradient techniques developed for use with the GTS system are presented. These techniques include methods for computing first and second partial derivatives of single and multiple sets of functions. A key feature of these methods is an error control mechanism which automatically adjusts the perturbation size to obtain accurate derivative values.

Hallman, Wayne P.↗

Robust time-optimal control of uncertain structural dynamic systems

A time-optimal open-loop control problem of flexible spacecraft in the presence of modeling uncertainty has been investigated. The results indicate that the proposed approach significantly reduces the residual structural vibrations caused by modeling uncertainty. The results also indicate the importance of proper jet placement for practical tradeoffs among the maneuvering time, fuel consumption, and performance robustness.

Wie, Bong↗

Robustness of Asymptotic Optimal Accelerated Test Plans to Small-Sample Settings

Accelerated testing is a form of testing commonly employed in industry, especially at Sandia National Laboratories, in order to extract information about product lifetime behavior that cannot be acquired under normal operating conditions due to the extensive durations of the intended life- times. As of the writing of this report, there is no known guideline for the design of accelerated tests at the laboratories and yet there is an entire field in the statistical literature on the optimal design of accelerated tests covering a wide range of limitations and constraints available for reference. A potential roadblock to incorporating this knowledge is the common difficulty at Sandia of having severely limited data available for such testing. Nearly all of the methodology presented in the statistical literature is based on asymptotic or large-sample theory and so may not be appropriate for the testing conditions present at Sandia. This report investigates optimal accelerated test plans derived based on small-sample or exact methodology for both a basic test setting as well as a more realistic setting and compares the resulting test plans to those derived based on the large- sample methodology from the literature. It is shown that the large-sample test plans are actually quite robust to the presence of limited data and are generally more flexible than the small-sample test plans.

36 MATERIALS SCIENCE↗

Robustness, efficiency, and optimality in the Fenna-Matthews-Olson photosynthetic pigment-protein complex

Pigment-protein complexes (PPCs) play a central role in facilitating excitation energy transfer (EET) from light-harvesting antenna complexes to reaction centres in photosynthetic systems; understanding molecular organisation in these biological networks is key to developing better artificial light-harvesting systems. In this article, we combine quantum-mechanical simulations and a network-based picture of transport to investigate how chromophore organization and protein environment in PPCs impacts on EET efficiency and robustness. In a prototypical PPC model, the Fenna-Matthews-Olson (FMO) complex, we consider the impact on EET efficiency of both disrupting the chromophore network and changing the influence of (local and global) environmental dephasing. Surprisingly, we find a large degree of resilience to changes in both chromophore network and protein environmental dephasing, the extent of which is greater than previously observed; for example, FMO maintains EET when 50% of the constituent chromophores are removed, or when environmental dephasing fluctuations vary over two orders-of-magnitude relative to the in vivo system. We also highlight the fact that the influence of local dephasing can be strongly dependent on the characteristics of the EET network and the initial excitation; for example, initial excitations resulting in rapid coherent decay are generally insensitive to the environment, whereas the incoherent population decay observed following excitation at weakly coupled chromophores demonstrates a more pronounced dependence on dephasing rate as a result of the greater possibility of local exciton trapping. Finally, we show that the FMO electronic Hamiltonian is not particularly optimised for EET; instead, it is just one of many possible chromophore organisations which demonstrate a good level of EET transport efficiency following excitation at different chromophores. Overall, these robustness and efficiency characteristics are attributed to the highly connected nature of the chromophore network and the presence of multiple EET pathways, features which might easily be built into artificial photosynthetic systems.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Use of Taguchi design of experiments to optimize and increase robustness of preliminary designs

The research performed this summer includes the completion of work begun last summer in support of the Air Launched Personnel Launch System parametric study, providing support on the development of the test matrices for the plume experiments in the Plume Model Investigation Team Project, and aiding in the conceptual design of a lunar habitat. After the conclusion of last years Summer Program, the Systems Definition Branch continued with the Air Launched Personnel Launch System (ALPLS) study by running three experiments defined by L27 Orthogonal Arrays. Although the data was evaluated during the academic year, the analysis of variance and the final project review were completed this summer. The Plume Model Investigation Team (PLUMMIT) was formed by the Engineering Directorate to develop a consensus position on plume impingement loads and to validate plume flowfield models. In order to obtain a large number of individual correlated data sets for model validation, a series of plume experiments was planned. A preliminary 'full factorial' test matrix indicated that 73,024 jet firings would be necessary to obtain all of the information requested. As this was approximately 100 times more firings than the scheduled use of Vacuum Chamber A would permit, considerable effort was needed to reduce the test matrix and optimize it with respect to the specific objectives of the program. Part of the First Lunar Outpost Project deals with Lunar Habitat. Requirements for the habitat include radiation protection, a safe haven for occasional solar flare storms, an airlock module as well as consumables to support 34 extra vehicular activities during a 45 day mission. The objective for the proposed work was to collaborate with the Habitat Team on the development and reusability of the Logistics Modules.

Carrasco, Hector R.↗

Performance Trades Study for Robust Airfoil Shape Optimization

From time to time, existing aircraft need to be redesigned for new missions with modified operating conditions such as required lift or cruise speed. This research is motivated by the needs of conceptual and preliminary design teams for smooth airfoil shapes that are similar to the baseline design but have improved drag performance over a range of flight conditions. The proposed modified profile optimization method (MPOM) modifies a large number of design variables to search for nonintuitive performance improvements, while avoiding off-design performance degradation. Given a good initial design, the MPOM generates fairly smooth airfoils that are better than the baseline without making drastic shape changes. Moreover, the MPOM allows users to gain valuable information by exploring performance trades over various design conditions. Four simulation cases of airfoil optimization in transonic viscous ow are included to demonstrate the usefulness of the MPOM as a performance trades study tool. Simulation results are obtained by solving fully turbulent Navier-Stokes equations and the corresponding discrete adjoint equations using an unstructured grid computational fluid dynamics code FUN2D.

Li, Wu↗

TH-C-BRD-12: Robust Intensity Modulated Proton Therapy Plan Can Eliminate Junction Shifts for Craniospinal Irradiation

Purpose: The passive scattering proton therapy (PSPT) technique is the commonly used radiotherapy technique for craniospinal irradiation (CSI). However, PSPT involves many numbers of junction shifts applied over the course of treatment to reduce the cold and hot regions caused by field mismatching. In this work, we introduced a robust planning approach to develop an optimal and clinical efficient techniques for CSI using intensity modulated proton therapy (IMPT) so that junction shifts can essentially be eliminated. Methods: The intra-fractional uncertainty, in which two overlapping fields shift in the opposite directions along the craniospinal axis, are incorporated into the robust optimization algorithm. Treatment plans with junction sizes 3,5,10,15,20,25 cm were designed and compared with the plan designed using the non-robust optimization. Robustness of the plans were evaluated based on dose profiles along the craniospinal axis for the plans applying 3 mm intra-fractional shift. The dose intra-fraction variations (DIV) at the junction are used to evaluate the robustness of the plans. Results: The DIVs are 7.9%, 6.3%, 5.0%, 3.8%, 2.8% and 2.2%, for the robustly optimized plans with junction sizes 3,5,10,15,20,25 cm. The DIV are 10% for the non-robustly optimized plans with junction size 25 cm. The dose profiles along the craniospinal axis exhibit gradual and tapered dose distribution. Using DIVs less than 5% as maximum acceptable intrafractional variation, the overlapping region can be reduced to 10 cm, leading to potential reduced number of the fields. The DIVs are less than 5% for 5 mm intra-fractional shifts with junction size 25 cm, leading to potential no-junction-shift for CSI using IMPT. Conclusion: This work is the first report of the robust optimization on CSI based on IMPT. We demonstrate that robust optimization can lead to much efficient carniospinal irradiation by eliminating the junction shifts.

60 APPLIED LIFE SCIENCES↗

A Measure Approximation for Distributionally Robust PDE-Constrained Optimization Problems

In numerous applications, scientists and engineers acquire varied forms of data that partially characterize the inputs to an underlying physical system. This data is then used to inform decisions such as controls and designs. Consequently, it is critical that the resulting control or design is robust to the inherent uncertainties associated with the unknown probabilistic characterization of the model inputs. Here in this work, we consider optimal control and design problems constrained by partial differential equations with uncertain inputs. We do not assume a known probabilistic model for the inputs, but rather we formulate the problem as a distributionally robust optimization problem where the outer minimization problem determines the control or design, while the inner maximization problem determines the worst-case probability measure that matches desired characteristics of the data. We analyze the inner maximization problem in the space of measures and introduce a novel measure approximation technique, based on the approximation of continuous functions, to discretize the unknown probability measure. Finally, we prove consistency of our approximated min-max problem and conclude with numerical results.

54 ENVIRONMENTAL SCIENCES↗

Optimal Control Modification for Robust Adaptation of Singularly Perturbed Systems with Slow Actuators

Recently a new optimal control modification has been introduced that can achieve robust adaptation with a large adaptive gain without incurring high-frequency oscillations as with the standard model-reference adaptive control. This modification is based on an optimal control formulation to minimize the L2 norm of the tracking error. The optimal control modification adaptive law results in a stable adaptation in the presence of a large adaptive gain. This study examines the optimal control modification adaptive law in the context of a system with a time scale separation resulting from a fast plant with a slow actuator. A singular perturbation analysis is performed to derive a modification to the adaptive law by transforming the original system into a reduced-order system in slow time. The model matching conditions in the transformed time coordinate results in increase in the feedback gain and modification of the adaptive law.

Nguyen, Nhan T.↗