Results for “robust optimization” — Page 4

NASA NTRS2003

The Role of Design-of-Experiments in Managing Flow in Compact Air Vehicle Inlets

It is the purpose of this study to demonstrate the viability and economy of Design-of-Experiments methodologies to arrive at microscale secondary flow control array designs that maintain optimal inlet performance over a wide range of the mission variables and to explore how these statistical methods provide a better understanding of the management of flow in compact air vehicle inlets. These statistical design concepts were used to investigate the robustness properties of low unit strength micro-effector arrays. Low unit strength micro-effectors are micro-vanes set at very low angles-of-incidence with very long chord lengths. They were designed to influence the near wall inlet flow over an extended streamwise distance, and their advantage lies in low total pressure loss and high effectiveness in managing engine face distortion. The term robustness is used in this paper in the same sense as it is used in the industrial problem solving community. It refers to minimizing the effects of the hard-to-control factors that influence the development of a product or process. In Robustness Engineering, the effects of the hard-to-control factors are often called noise , and the hard-to-control factors themselves are referred to as the environmental variables or sometimes as the Taguchi noise variables. Hence Robust Optimization refers to minimizing the effects of the environmental or noise variables on the development (design) of a product or process. In the management of flow in compact inlets, the environmental or noise variables can be identified with the mission variables. Therefore this paper formulates a statistical design methodology that minimizes the impact of variations in the mission variables on inlet performance and demonstrates that these statistical design concepts can lead to simpler inlet flow management systems.

Anderson, Bernhard H.↗

NASA NTRS1983

Guaranteed robustness properties of multivariable, nonlinear, stochastic optimal regulators

The robustness of optimal regulators for nonlinear, deterministic and stochastic, multi-input dynamical systems is studied under the assumption that all state variables can be measured. It is shown that, under mild assumptions, such nonlinear regulators have a guaranteed infinite gain margin; moreover, they have a guaranteed 50 percent gain reduction margin and a 60 degree phase margin, in each feedback channel, provided that the system is linear in the control and the penalty to the control is quadratic, thus extending the well-known properties of LQ regulators to nonlinear optimal designs. These results are also valid for infinite horizon, average cost, stochastic optimal control problems.

Tsitsiklis, J. N.↗

NASA NTRS1984

Guaranteed robustness properties of multivariable nonlinear stochastic optimal regulators

The robustness of optimal regulators for nonlinear, deterministic and stochastic, multi-input dynamical systems is studied under the assumption that all state variables can be measured. It is shown that, under mild assumptions, such nonlinear regulators have a guaranteed infinite gain margin; moreover, they have a guaranteed 50 percent gain reduction margin and a 60 degree phase margin, in each feedback channel, provided that the system is linear in the control and the penalty to the control is quadratic, thus extending the well-known properties of LQ regulators to nonlinear optimal designs. These results are also valid for infinite horizon, average cost, stochastic optimal control problems.

Tsitsiklis, J. N.↗

NASA NTRS2010

Robustness-Based Design Optimization Under Data Uncertainty

This paper proposes formulations and algorithms for design optimization under both aleatory (i.e., natural or physical variability) and epistemic uncertainty (i.e., imprecise probabilistic information), from the perspective of system robustness. The proposed formulations deal with epistemic uncertainty arising from both sparse and interval data without any assumption about the probability distributions of the random variables. A decoupled approach is proposed in this paper to un-nest the robustness-based design from the analysis of non-design epistemic variables to achieve computational efficiency. The proposed methods are illustrated for the upper stage design problem of a two-stage-to-orbit (TSTO) vehicle, where the information on the random design inputs are only available as sparse point and/or interval data. As collecting more data reduces uncertainty but increases cost, the effect of sample size on the optimality and robustness of the solution is also studied. A method is developed to determine the optimal sample size for sparse point data that leads to the solutions of the design problem that are least sensitive to variations in the input random variables.

Zaman, Kais↗

NASA NTRS1992

Desensitizing structural-control design

A new method for targeted robustness optimization is presented. This method allows the analyst to 'target' the most critical performance parameter (or family of design requirements with almost the same criticality) for enhanced protection from uncertainties in the model. To create this protection, this method uses knowledge about the uncertainties in some of the parameters of the model - this information is usually available to the structural analyst. The power of the targeted robustness method is demonstrated numerically using an active damping example. The results show that the most critical damping ratio is protected - its sensitivity to the uncertainties is far below those of non-critical damping ratios.

Ponslet, E.↗

NASA NTRS1991

National Launch System Space Transportation Main Engine

The present liquid-oxygen/liquid-hydrogen engine is described as meeting the specific requirements of the National Launch System (NLS) Program including cost-effectiveness and robustness. An overview of the NLS and its objectives is given which indicates that the program aims to develop a flexible launch system to meet security, civil, and commercial needs. The Space Transportation Main Engine (STME) provides core and boost propulsion for the 1.5-stage vehicle and core propulsion for the solid booster vehicle. The design incorporates step-throttling, order-of-magnitude reductions in welds, and configuration targets designed to optimize robustness. The STME is designed to provide adaptable and dependable propulsion while minimizing recurring costs and is designed to meet the needs of NLS and other typical space-transportation programs currently being planned.

Hoodless, Ralph M., Jr.↗

NASA NTRSDate not supplied

Towards an Aviation Large Language Model by Fine-tuning and Evaluating Transformers

In the aviation domain, there are many applications for machine learning and artificial intelligence tools that utilize natural language. For example, there is a desire to know the commonalities in written safety reports such as voluntary post incidents reports or aerial wildfire operations reports to better understand the risks present. Another use-case is the possibility of extracting airspace procedures and constraints currently written in documents such as Letters of Agreement. These applications can benefit from the use of state-of-the-art natural language processing techniques when adapted to the language/phraseology specific to the aviation domain. This paper evaluates the viability of adaptation of NLP tools to the aviation domain by fine-tuning transformer based models using aviation data sets. In 2018, a novel language model based on neural units (also called transformers) was created and became known as “Bidirectional Encoder Representations from Transformers” or BERT. This architecture combined with large amounts of English training data and innovative semi-supervised training tasks set the standard for what would later emerge as Large Language Models. The performance of these models was further improved by hyperparameter tuning and refinement of the semi-supervised training task and resulted in “Robustly Optimized BERT Pre-training Approach through hyperparameter tuning” or RoBERTa models. These pre-trained Large Language Models proved to be useful for a wide variety of natural language processing tasks such as text classification and question answering through a process called fine-tuning. The transformer architecture with pre-trained weights served as the basis with the last few layers replaced with layers fine-tuned to perform a new task e.g., a layer that provides a label for the entire input text. This process of fine-tuning can also be used to adapt the models to new domains; e.g., BioBERT started with the pre-trained BERT model and was completed by additional fine-tuning and training on biomedical documents. Transformer-based architectures can also be used to create rich representations of text called embeddings which can serve as the input to other machine learning models. This allows simpler algorithms such as logistic regression to use context-rich representations of the text while still remaining quick to train and evaluate. In the world of aviation, there is a growing demand for natural language processing and understanding but the domain presents unique challenges. Due to the technical content (and specialized language) of most aviation documents, fine-tuning pre-trained Large Language Models to specific tasks has not met the benchmark on natural language processing tasks set by simpler models trained from scratch on the data. To address this deficiency, this paper evaluates the improvements from fine-tuning a Large Language Model on a large set of aviation documents using the original semi-supervised training tasks before performing specific natural language tasks. In fine-tuning, a domain-specific dataset is used on the original training task but with the pre-trained Large Language Model instead of starting from a random initialization. This approach allows the model to be adapted to the specific domain language without discarding the information gained from training on general English data. This paper utilized two major dataset types to train and assess the RoBERTa fine-tuning performance. The first are 7,057 Letters of Agreement which are Federal Aviation Administration (FAA) documents that formalize airspace operations across the national airspace system. They contain many examples of ‘aviation English’ using domain specific terminology and phrasing which serves as a representative basis to perform the semi-supervised fine-tuning. The second type is the 494 document classification labels to be used for evaluation. This down-stream evaluation aims to show the performance of the fine-tuned model, better understand how much data is needed for an effective fine-tuning, and how fine-tuning can be adapted for different applications in-the domain. After semi-supervised training, evaluation begins by encoding the documents for classification using the fine-tuned RoBERTa model. Then a logistic regression classifier is trained to label the document type and compared against our ground truth labels. This currently leads to a 82.8% accuracy on 10-fold cross validation showing improvement over baseline RoBERTa which achieved 81.0%. We plan to measure the improvements on additional tasks and it is expected that these improvements will lead to more robust models that can tackle the natural language processing challenges present in aviation datasets.

ATM↗

NASA NTRS1987

Constrained optimization techniques for active control of aeroelastic response

Active control of aeroelastic response is a complex problem in which the designer usually tries to satisfy many design criteria which are often conflicting in nature. To further complicate the design problem, the state space equations describing this type of control problem are usually of high order, involving a large number of states to represent the flexible structure and unsteady aerodynamics. Control laws based on the standard Linear - Quadratic - Gaussian method are of the same high order as the aeroelastic plant and may be difficult to implement in the flight computer. To overcome this disadvantage a new approach was developed for designing low-order optimized robust control laws. In this approach, a nonlinear programming algorithm is used to search for the values of control law design variables that minimize a performance index while satisfying several inequality constraints that describe the design criteria on the stability robustness and responses. The method is applied to a gust load alleviation problem and a stability robustness improvement problem of a drone aircraft.

Mukhopadhyay, Vivekananda↗

NASA NTRS1996

A New Technique for Compensating Joint Limits in a Robot Manipulator

A new robust, optimal, adaptive technique for compensating rate and position limits in the joints of a six degree-of-freedom elbow manipulator is presented. In this new algorithm, the unmet demand as a result of actuator saturation is redistributed among the remaining unsaturated joints. The scheme is used to compensate for inadequate path planning, problems such as joint limiting, joint freezing, or even obstacle avoidance, where a desired position and orientation are not attainable due to an unrealizable joint command. Once a joint encounters a limit, supplemental commands are sent to other joints to best track, according to a selected criterion, the desired trajectory.

Litt, Jonathan↗

NASA NTRS2003

Structured Uncertainty Bound Determination From Data for Control and Performance Validation

This report attempts to document the broad scope of issues that must be satisfactorily resolved before one can expect to methodically obtain, with a reasonable confidence, a near-optimal robust closed loop performance in physical applications. These include elements of signal processing, noise identification, system identification, model validation, and uncertainty modeling. Based on a recently developed methodology involving a parameterization of all model validating uncertainty sets for a given linear fractional transformation (LFT) structure and noise allowance, a new software, Uncertainty Bound Identification (UBID) toolbox, which conveniently executes model validation tests and determine uncertainty bounds from data, has been designed and is currently available. This toolbox also serves to benchmark the current state-of-the-art in uncertainty bound determination and in turn facilitate benchmarking of robust control technology. To help clarify the methodology and use of the new software, two tutorial examples are provided. The first involves the uncertainty characterization of a flexible structure dynamics, and the second example involves a closed loop performance validation of a ducted fan based on an uncertainty bound from data. These examples, along with other simulation and experimental results, also help describe the many factors and assumptions that determine the degree of success in applying robust control theory to practical problems.

Lim, Kyong B.↗

NASA NTRS2010

NASA Tech Briefs, October 2010

Topics covered include: Hybrid Architecture Active Wavefront Sensing and Control; Carbon-Nanotube-Based Chemical Gas Sensor; Aerogel-Positronium Technology for the Detection of Small Quantities of Organic and/or Toxic Materials; Graphene-Based Reversible Nano-Switch/Sensor Schottky Diode; Inductive Non-Contact Position Sensor; High-Temperature Surface-Acoustic-Wave Transducer; Grid-Sphere Electrodes for Contact with Ionospheric Plasma; Enabling IP Header Compression in COTS Routers via Frame Relay on a Simplex Link; Ka-Band SiGe Receiver Front-End MMIC for Transponder Applications; Robust Optimization Design Algorithm for High-Frequency TWTs; Optimal and Local Connectivity Between Neuron and Synapse Array in the Quantum Dot/Silicon Brain; Method and Circuit for In-Situ Health Monitoring of Solar Cells in Space; BGen: A UML Behavior Network Generator Tool; Platform for Post-Processing Waveform-Based NDE; Electrochemical Hydrogen Peroxide Generator; Fabrication of Single, Vertically Aligned Carbon Nanotubes in 3D Nanoscale Architectures; Process to Create High-Fidelity Lunar Dust Simulants; Lithium-Ion Electrolytes Containing Phosphorous-Based, Flame-Retardant Additives; InGaP Heterojunction Barrier Solar Cells; Straight-Pore Microfilter with Efficient Regeneration; Determining Shear Stress Distribution in a Laminate; Self-Adjusting Liquid Injectors for Combustors; Handling Qualities Prediction of an F-16XL-Based Reduced Sonic Boom Aircraft; Tele-Robotic ATHLETE Controller for Kinematics - TRACK; Three-Wheel Brush-Wheel Sampler; Heterodyne Interferometer Angle Metrology; Aligning Astronomical Telescopes via Identification of Stars; Generation of Optical Combs in a WGM Resonator from a Bichromatic Pump; Large-Format AlGaN PIN Photodiode Arrays for UV Images; Fiber-Coupled Planar Light-Wave Circuit for Seed Laser Control in High Spectral Resolution Lidar Systems; On Calculating the Zero-Gravity Surface Figure of a Mirror; Optical Modification of Casimir Forces for Improved Function of Micro- and Nano-Scale Devices; Analysis, Simulation, and Verification of Knowledge-Based, Rule-Based, and Expert Systems; Core and Off-Core Processes in Systems Engineering; Digital Reconstruction Supporting Investigation of Mishaps; and Template Matching Approach to Signal Prediction.

Source record↗

NASA NTRS2014

NASA Tech Briefs, January 2014

Topics include: Multi-Source Autonomous Response for Targeting and Monitoring of Volcanic Activity; Software Suite to Support In-Flight Characterization of Remote Sensing Systems; Visual Image Sensor Organ Replacement; Ultra-Wideband, Dual-Polarized, Beam-Steering P-Band Array Antenna; Centering a DDR Strobe in the Middle of a Data Packet; Using a Commercial Ethernet PHY Device in a Radiation Environment; Submerged AUV Charging Station; Habitat Demonstration Unit (HDU) Vertical Cylinder Habitat; Origami-Inspired Folding of Thick, Rigid Panels; A Novel Protocol for Decoating and Permeabilizing Bacterial Spores for Epifluorescent Microscopy; Method and Apparatus for Automated Isolation of Nucleic Acids from Small Cell Samples; Enabling Microliquid Chromatography by Microbead Packing of Microchannels; On-Command Force and Torque Impeding Devices (OC-FTID) Using ERF; Deployable Fresnel Rings; Transition-Edge Hot-Electron Microbolometers for Millimeter and Submillimeter Astrophysics; Spacecraft Trajectory Analysis and Mission Planning Simulation (STAMPS) Software; Cross Support Transfer Service (CSTS) Framework Library; Arbitrary Shape Deformation in CFD Design; Range Safety Flight Elevation Limit Calculation Software; Frequency-Modulated, Continuous-Wave Laser Ranging Using Photon-Counting Detectors; Calculation of Operations Efficiency Factors for Mars Surface Missions; GPU Lossless Hyperspectral Data Compression System; Robust, Optimal Subsonic Airfoil Shapes; Protograph-Based Raptor-Like Codes; Fuzzy Neuron: Method and Hardware Realization; Kalman Filter Input Processor for Boresight Calibration; Organizing Compression of Hyperspectral Imagery to Allow Efficient Parallel Decompression; and Temperature Dependences of Mechanisms Responsible for the Water-Vapor Continuum Absorption.

Source record↗

NASA NTRS2017

The Future of Ground Magnetometer Arrays in Support of Space Weather Monitoring and Research

A community workshop was held in Greenbelt, Maryland, on 5-6 May 2016 to discussrecommendations for the future of ground magnetometer array research in space physics. The community reviewed findings contained in the 2016 Geospace Portfolio Review of the Geospace Section of the Division of Atmospheric and Geospace Science of the National Science Foundation and discussed the present state of ground magnetometer arrays and possible pathways for a more optimal, robust, and effective organization and scientific use of these ground arrays. This paper summarizes the report of that workshop to the National Science Foundation (Engebretson & Zesta, 2017) as well as conclusions from two follow-up meetings. It describes the current state of U.S.-funded ground magnetometer arrays and summarizes community recommendations for changes in both organizational and funding structures. It also outlines a variety of new and/or augmented regional and global data products and visualizations that can be facilitated by increased collaboration among arrays. Such products will enhance the value of ground-based magnetometer data to the community's effort for understanding of Earth's space environment and space weather effects.

earths space environment↗

NASA NTRSDate not supplied

Examination of Unified Control Incorporating Generalized Control Allocation

Hybrid vehicles which combine vertical take off and landing with cruise configurations pose a unique challenge for control design and implementation. For this class of vehicle, successful control designs have historically broken the flight envelope into phases of flight and modified the control approach for each phase. This research approaches control in a unified way across the entire envelope utilizing a robust optimal design which provides effect or weighting then implemented in a generalized Affine Generalized Inverse control allocation algorithm. System performance for a Lift plus Cruise class vehicle is presented.

Unified Control,Control Allocation↗

NASA NTRSDate not supplied

Towards an Aviation Large Language Model by Fine-tuning and Evaluating Transformers

In the aviation domain, there are many applications for machine learning and artificial intelligence tools that utilize natural language. For example, there is a desire to know the commonalities in written safety reports such as voluntary post incidents reports or create more accurate transcripts of air traffic management conversations. Another use-case is the possibility of extracting airspace procedures and constraints currently written in documents such as Letters of Agreement (LOA) which is used as the evaluation case in this paper. These applications can benefit from the use of state-of-the-art Natural Language Processing (NLP) techniques when adapted to the language/phraseology specific to the aviation domain. This paper evaluates the viability of transferring pre-trained large language models to the aviation domain by adapting transformer based models using aviation datasets. This paper utilized two datasets to adapt a ‘Robustly Optimized Bidirectional Encoder Representations from Transformers Approach’ (RoBERTa) model and two down-stream classification tasks to assess its performance. These datasets are all built upon Letters of Agreement which are Federal Aviation Administration (FAA) documents that formalize airspace operations across the national airspace system. The first two datasets are used for the adaptation of RoBERTa to the aviation domain and were of different sizes to assess the number of documents needed to adapt to the aviation domain. They contain many examples of ‘aviation English’ using domain specific terminology and phrasing which serves as a representative basis to perform the unsupervised adaptation. The second dataset is a separate set of LOA documents with two sets of classification labels to be used for evaluation; one at the document level and one at the line level. These down-stream evaluations allowed the measurement of improvement by adapting RoBERTa. The accuracy increased by 4-6% on both tasks and the F1 score on the class of interest increased by 4-8% from the adaptation.

Air Traffic Management↗

NASA NTRSDate not supplied

Towards an Aviation Large Language Model by Fine-tuning and Evaluating Transformers

In the aviation domain, there are many applications for machine learning and artificial intelligence tools that utilize natural language. For example, there is a desire to know the commonalities in written safety reports such as voluntary post incidents reports or create more accurate transcripts of air traffic management conversations. Another use-case is the possibility of extracting airspace procedures and constraints currently written in documents such as Letters of Agreement (LOA) which is used as the evaluation case in this paper. These applications can benefit from the use of state-of-the-art Natural Language Processing (NLP) techniques when adapted to the language/phraseology specific to the aviation domain. This paper evaluates the viability of transferring pre-trained large language models to the aviation domain by adapting transformer based models using aviation datasets. This paper utilized two datasets to adapt a ‘Robustly Optimized Bidirectional Encoder Representations from Transformers Approach’ (RoBERTa) model and two down-stream classification tasks to assess its performance. These datasets are all built upon Letters of Agreement which are Federal Aviation Administration (FAA) documents that formalize airspace operations across the national airspace system. The first two datasets are used for the adaptation of RoBERTa to the aviation domain and were of different sizes to assess the number of documents needed to adapt to the aviation domain. They contain many examples of ‘aviation English’ using domain specific terminology and phrasing which serves as a representative basis to perform the unsupervised adaptation. The second dataset is a separate set of LOA documents with two sets of classification labels to be used for evaluation; one at the document level and one at the line level. These down-stream evaluations allowed the measurement of improvement by adapting RoBERTa. The accuracy increased by 4-6% on both tasks and the F1 score on the class of interest increased by 4-8% from the adaptation.

Air Traffic Management↗

NASA NTRSDate not supplied

Sensitivity of Optimal Midcourse Correction Scheduling for Robust Cislunar Trajectory Design

A new approach to optimal trajectory design is the determination of optimal trajectories that are robust to initial trajectory dispersions, navigation errors, maneuver execution errors, and environment modeling errors. This paper investigates the sensitivity of cislunar robust optimal trajectory design to launch date, duration of navigation measurement passes, and navigation measurement frequency. For a given cislunar trajectory from translunar injection (TLI) to lunar orbit insertion (LOI), the optimal locations of midcourse corrections, also known as trajectory correction maneuvers (TCM) are determined by minimizing the final 3-σ ∆v subject to a final 3-σ position dispersion constraint for a given launch date, specified measurement pass duration prior to each maneuver, and measurement frequency. Optical navigation (OpNav) is assumed, and OpNav field-of-view (FOV) and lighting constraints are employed. These constraints turn out to be important elements of the problem. The sensitivity of the optimal TCM locations are then investigated by varying the launch date, duration of OpNav measurement passes, and the OpNav measurement frequency, and then re-optimizing the locations of the TCMs. Given the problem parameters provided herein, results show that while the optimal TCM locations with respect to TLI vary greatly from one launch date to another, their locations with respect to LOI are nearly invariant over a 2-month launch window. Results also show that in all cases the optimal location of the last TCM is found to be at the point where the OpNav lunar FOV constraint is first violated. For all other TCMs, OpNav measurement pass duration and measurement frequency can have a moderate to large affect on the optimal TCM locations.

Linear Covariance Analysis↗

NASA NTRS1992

On actuator placement for robust time-optimal control of uncertain flexible spacecraft

The problem of computing open-loop, on-off jet firing logic for flexible spacecraft in the face of plant modeling uncertainty is investigated. The primary control objective is to achieve a fast maneuvering time with a minimum of structural vibrations during and/or after a maneuver. This paper is also concerned with the problem of selecting a proper pair of jets for practical trade-offs among the maneuvering time, fuel consumption, structural mode excitation, and performance robustness. A time-optimal control problem subject to parameter robustness constraints is formulated. A three-mass-spring model of flexible spacecraft with a rigid-body mode and two flexible modes is used to illustrate the concept.

Wie, Bong↗