SEARCH · Search NASA

Results for “adaptive weights”

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.

SourceResearch connectionFrom yearThrough year

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

At least 73 records · Page 4

All sources NASA NTRS DOE OSTI

NASA NTRS2004

Fabrication and Testing of Active and Adaptive Cyanate Ester Composite Mirrors

The objective of the NASA/Bennett Optical Research Inc. (BOR) NAS8-02008 Phase II Program, which also incorporated ideas developed under the earlier NASA NAS8-01035 Phase 1 Program, was to develop a large mirror fabrication and test facility with emphasis on producing large, light weight active and adaptive optics. A principle objective was to develop mandrels on which to make large composite graphite-filled cyanate ester mirrors, Deliverables were two of these superpolished lightweight active/adaptive optic composite mirrors, one 12" (approx.1/3 meter) in diameter and one 22" (approx.1/2 meter) in diameter. In addition optical superpolishers for mandrels up to 1.2 meters in diameter, test instruments for determining optical figure and scattered light, novel design actuators for making the composite mirrors both active and adaptive, and passive and active means for measuring actuator performance were developed at BOR. We are now installing a superpolisher capable of producing 3 meter diameter mirror/mandrels. All polishers utilize the principle of centrifugal elutriation and produce superpolished mandrels with surface microroughnesses under 1 nm rms.

Bennett, H. E.↗

NASA NTRS1988

Adaptive Control For Space-Station Joints

Input gain weighting added to existing adaptive-control algorithm. Report discusses concept for adaptive control of rotations of various appended structures about joints on space station, with suppression of vibrations in both station and appendages. Concept relevant to such terrestrial applications as control of aircraft, suppression of vibrations in large buildings, and robotics.

Ih, Che-Hang Charles↗

DOE OSTI2022

Scalable and Memory-Efficient Algorithms for Controlling Networked Epidemic Processes Using Multiplicative Weights Update Method

We study the problem of designing scalable algorithms to find effective intervention strategies for controlling stochastic epidemic processes on networks. This is a common problem arising in agent based models for epidemic spread. Previous approaches to this problem focus on either heuristics with no guarantees or approximation algorithms that scale only to networks corresponding to county-sized populations, typically, with less than a million nodes. In particular, the mathematical-programming based approaches need to solve the Linear Program (LP) relaxation of the problem using an LP solver, which restricts the scalability of this approach. In this work, we overcome this restriction by designing an algorithm that adapts the multiplicative weights update (MWU) framework, along with the sample average approximation (SAA) technique, to approximately solve the linear program (LP) relaxation for the problem. To scale this approach further, we provide a memory-efficient algorithm that enables scaling to large networks, corresponding to country-size populations, with over 300 million nodes and 30 billion edges. Furthermore, we show that this approach provides near-optimal solutions to the LP in practice.

Sambaturu, Prathyush↗

DOE OSTI2025

Conformal Hierarchical Simulation-Based Inference with Local Validity

Trustworthy and interpretable uncertainty quantification is a long-standing challenge in artificial intelligence. Simulation-based inference (SBI) comprises a broad swath of approaches for estimating latent parameters with uncertainties. Although flexible neural density estimators in SBI can be remark- ably expressive capturing highly structured, high-dimensional posteriors their credible regions can be badly mis-calibrated and are often only accompanied by heuristic coverage checks. We present the first SBI framework that delivers finite-sample local valid coverage guarantees that hold in the neighborhood of each observation. Our framework can couple any off-the-shelf hierarchical SBI engine with a confor- mal Bayesian post-processing step that operates on the posterior predictive density. A kernel-weighted conformity score adapts the conformal quantile to the local geometry of the data, yielding prediction sets that are simultaneously (i) marginally calibrated, (ii) locally valid, and (iii) hierarchical, handling global and observation-specific parameters in a single pass. Through experiments on synthetic data and benchmarks from neuroscience and physics, we show that our approach attains 1 − α coverage, where prior SBI methods under- or over-cover. Our approach also maintains a competitive, credible set size with minimal computational overhead. Finally, our approach can be used to make predictions on real data and give valid credible regions modulo weight-initialization-based model mis-specification.

Trivedi, Shubhendu [Fermilab]↗

NASA NTRS2007

Thin, nearly wireless adaptive optical device

A thin, nearly wireless adaptive optical device capable of dynamically modulating the shape of a mirror in real time to compensate for atmospheric distortions and/or variations along an optical material is provided. The device includes an optical layer, a substrate, at least one electronic circuit layer with nearly wireless architecture, an array of actuators, power electronic switches, a reactive force element, and a digital controller. Actuators are aligned so that each axis of expansion and contraction intersects both substrate and reactive force element. Electronics layer with nearly wireless architecture, power electronic switches, and digital controller are provided within a thin-film substrate. The size and weight of the adaptive optical device is solely dominated by the size of the actuator elements rather than by the power distribution system.

Knowles, Gareth↗

NASA NTRS2008

Thin, nearly wireless adaptive optical device

Knowles, Gareth↗

NASA NTRS2009

Thin nearly wireless adaptive optical device

A thin nearly wireless adaptive optical device capable of dynamically modulating the shape of a mirror in real time to compensate for atmospheric distortions and/or variations along an optical material is provided. The device includes an optical layer, a substrate, at least one electronic circuit layer with nearly wireless architecture, an array of actuators, power electronic switches, a reactive force element, and a digital controller. Actuators are aligned so that each axis of expansion and contraction intersects both substrate and reactive force element. Electronics layer with nearly wireless architecture, power electronic switches, and digital controller are provided within a thin-film substrate. The size and weight of the adaptive optical device is solely dominated by the size of the actuator elements rather than by the power distribution system.

Knowles, Gareth J.↗

DOE OSTI2022

Improved Architectures and Training Algorithms for Deep Operator Networks

Operator learning techniques have recently emerged as a powerful tool for learning maps between infinite-dimensional Banach spaces. Trained under appropriate constraints, they can also be effective in learning the solution operator of partial differential equations (PDEs) in an entirely self-supervised manner. In this work we analyze the training dynamics of deep operator networks (DeepONets) through the lens of Neural Tangent Kernel theory, and reveal a bias that favors the approximation of functions with larger magnitudes. To correct this bias we propose to adaptively re-weight the importance of each training example, and demonstrate how this procedure can effectively balance the magnitude of back-propagated gradients during training via gradient descent. We also propose a novel network architecture that is more resilient to vanishing gradient pathologies. Taken together, our developments provide new insights into the training of DeepONets and consistently improve their predictive accuracy by a factor of 10-50x, demonstrated in the challenging setting of learning PDE solution operators in the absence of paired input-output observations.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

On the influence of over-parameterization in manifold based surrogates and deep neural operators

Constructing accurate and generalizable approximators (surrogate models) for complex physico-chemical processes exhibiting highly non-smooth dynamics is challenging. The main question is what type of surrogate models we should construct and should these models be under-parameterized or over-parameterized. In this work, we propose new developments and perform comparisons for two promising approaches: manifold-based polynomial chaos expansion (m-PCE) and the deep neural operator (DeepONet), and we examine the effect of over-parameterization on generalization. While m-PCE enables the construction of a mapping by first identifying low-dimensional embeddings of the input functions, parameters, and quantities of interest (QoIs), a neural operator learns the nonlinear mapping via the use of deep neural networks. Here, we demonstrate the performance of these methods in terms of generalization accuracy by solving the 2D time-dependent Brusselator reaction-diffusion system with uncertainty sources, modeling an autocatalytic chemical reaction between two species. We first propose an extension of the m-PCE by constructing a mapping between latent spaces formed by two separate embeddings of the input functions and the output QoIs. To further enhance the accuracy of the DeepONet, we introduce weight self-adaptivity in the loss function. We demonstrate that the performance of m-PCE and DeepONet is comparable for cases of relatively smooth input-output mappings. However, when highly non-smooth dynamics is considered, DeepONet shows higher approximation accuracy. We also find that for m-PCE, modest over-parameterization leads to better generalization, both within and outside of distribution, whereas aggressive over-parameterization leads to over-fitting. In contrast, an even highly over-parameterized DeepONet leads to better generalization for both smooth and non-smooth dynamics. Furthermore, we compare the performance of the above models with another recently proposed operator learning model, the Fourier Neural Operator, and show that its over-parameterization also leads to better generalization. Taken together, our studies show that m-PCE can provide very good accuracy at very low training cost, whereas a highly over-parameterized DeepONet can provide better accuracy and robustness to noise but at higher training cost. In both methods, the inference cost is negligible.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

A Multi-Objective Bayesian Optimized Human Assessed Multi-Target Generated Spectral Recommender System for Rapid Pareto Discoveries of Material Properties

Optimization for different tasks like material characterization, synthesis, and functional properties for desired applications over multi-dimensional control parameter and function spaces need a rapid strategic search through active learning. However, in all cases prior to optimization, the target material properties are assumed known and fixed, which mostly deviates from real-world scenarios in material synthesis. This can be critical for running expensive experiments on new materials, when the experimental results are fuzzy for any scientific outcomes due to improper target setting, ultimately wasting time and cost. The failure rate and cost are even higher over exploring on multi-target space, where we want to learn the pareto among multiple properties, to jointly optimize during material synthesis for desired applications. To address the challenge, here we introduce the human-operator attempt flexibility in the active learning based automated experiment framework, with generating multiple human assessed targets through a voting-based recommender system during real-time microscope measurements over the large material image space, sequentially learn/update multiple desired targets through a weighting system, and adaptively search in multiple material properties functional space for non-dominated pareto discoveries to maximize the custom structural similarity based acquisition function. We term this a multi-objective Bayesian optimized human assessed multi-target generated spectral recommender systems (MOBO-HAM-SRS). The approach has been demonstrated to peizoresponse force spectroscopy of a ferroelectric thin film, exploring with different kernels and acquisition functions. This work shows an advancement towards human-AI collaborated automated experiments, steering optimization trajectories through human overpowering AI at the early stage when uncertainty is high and AI overpowering human at the later stage with rapid exploration towards optimal goal, following human-assessed multiple targets properties.

Biswas, Arpan↗

NASA NTRS1991

Experimental study of adaptive pointing and tracking for large flexible space structures

This paper describes an experimental study of adaptive pointing and tracking control for flexible spacecraft conducted on a complex ground experiment facility. The algorithm used in this study is based on a multivariable direct model reference adaptive control law. Several experimental validation studies were performed earlier using this algorithm for vibration damping and robust regulation, with excellent results. The current work extends previous studies by addressing the pointing and tracking problem. As is consistent with an adaptive control framework, the plant is assumed to be poorly known to the extent that only system level knowledge of its dynamics is available. Explicit bounds on the steady-state pointing error are derived as functions of the adaptive controller design parameters. It is shown that good tracking performance can be achieved in an experimental setting by adjusting adaptive controller design weightings according to the guidelines indicated by the analytical expressions for the error.

Boussalis, D.↗

NASA NTRS1990

A 4800 bps CELP vocoder with an improved excitation

The Stochastic or Code Excited Linear Predictive Coder (CELP) is among the promising candidates for producing good quality speech at low bit rates. However, the speech quality produced suffers from perceived roughness. Many researchers have used pole-zero postfilters to mask the roughness at the output of the synthesis filter. Although the postfilters are effective in masking the noise at low bit rates, they produce spectral distortions. It is proposed that speech can be improved by introducing two modifications to the fixed stochastic codebook. In the first modification, the stochastic codebook is used only when the long term correlations are low. Otherwise, a pulse like codebook is selected. In the second modification, the selected codebook output is weighted using an adaptive spectral shaping procedure. These two modifications were incorporated in a 4800 bps CELP coder and have resulted in a perceptually improved vocoded speech.

Hassanein, Hisham↗

NASA NTRS1992

Neural network based adaptive control of nonlinear plants using random search optimization algorithms

This paper presents a method for utilizing artificial neural networks for direct adaptive control of dynamic systems with poorly known dynamics. The neural network weights (controller gains) are adapted in real time using state measurements and a random search optimization algorithm. The results are demonstrated via simulation using two highly nonlinear systems.

Boussalis, Dhemetrios↗

NASA NTRS1992

A neural network prototyping package within IRAF

We outline our plans for incorporating a Neural Network Prototyping Package into the IRAF environment. The package we are developing will allow the user to choose between different types of networks and to specify the details of the particular architecture chosen. Neural networks consist of a highly interconnected set of simple processing units. The strengths of the connections between units are determined by weights which are adaptively set as the network 'learns'. In some cases, learning can be a separate phase of the user cycle of the network while in other cases the network learns continuously. Neural networks have been found to be very useful in pattern recognition and image processing applications. They can form very general 'decision boundaries' to differentiate between objects in pattern space and they can be used for associative recall of patterns based on partial cures and for adaptive filtering. We discuss the different architectures we plan to use and give examples of what they can do.

Bazell, D.↗

NASA NTRS1998

Application of Least Mean Square Algorithms to Spacecraft Vibration Compensation

This paper describes the application of the Least Mean Square (LMS) algorithm in tandem with the Filtered-X Least Mean Square algorithm for controlling a science instrument's line-of-sight pointing. Pointing error is caused by a periodic disturbance and spacecraft vibration. A least mean square algorithm is used on-orbit to produce the transfer function between the instrument's servo-mechanism and error sensor. The result is a set of adaptive transversal filter weights tuned to the transfer function. The Filtered-X LMS algorithm, which is an extension of the LMS, tunes a set of transversal filter weights to the transfer function between the disturbance source and the servo-mechanism's actuation signal. The servo-mechanism's resulting actuation counters the disturbance response and thus maintains accurate science instrumental pointing. A simulation model of the Upper Atmosphere Research Satellite is used to demonstrate the algorithms.

Woodard , Stanley E.↗

NASA NTRS2001

Perception of tilt (somatogravic illusion) in response to sustained linear acceleration during space flight

During the 1998 Neurolab mission (STS-90), four astronauts were exposed to interaural and head vertical (dorsoventral) linear accelerations of 0.5 g and 1 g during constant velocity rotation on a centrifuge, both on Earth and during orbital space flight. Subjects were oriented either left-ear-out or right-ear-out (Gy centrifugation), or lay supine along the centrifuge arm with their head off-axis (Gz centrifugation). Pre-flight centrifugation, producing linear accelerations of 0.5 g and 1 g along the Gy (interaural) axis, induced illusions of roll-tilt of 20 degrees and 34 degrees for gravito-inertial acceleration (GIA) vector tilts of 27 degrees and 45 degrees , respectively. Pre-flight 0.5 g and 1 g Gz (head dorsoventral) centrifugation generated perceptions of backward pitch of 5 degrees and 15 degrees , respectively. In the absence of gravity during space flight, the same centrifugation generated a GIA that was equivalent to the centripetal acceleration and aligned with the Gy or Gz axes. Perception of tilt was underestimated relative to this new GIA orientation during early in-flight Gy centrifugation, but was close to the GIA after 16 days in orbit, when subjects reported that they felt as if they were 'lying on side'. During the course of the mission, inflight roll-tilt perception during Gy centrifugation increased from 45 degrees to 83 degrees at 1 g and from 42 degrees to 48 degrees at 0.5 g. Subjects felt 'upside-down' during in-flight Gz centrifugation from the first in-flight test session, which reflected the new GIA orientation along the head dorsoventral axis. The different levels of in-flight tilt perception during 0.5 g and 1 g Gy centrifugation suggests that other non-vestibular inputs, including an internal estimate of the body vertical and somatic sensation, were utilized in generating tilt perception. Interpretation of data by a weighted sum of body vertical and somatic vectors, with an estimate of the GIA from the otoliths, suggests that perception weights the sense of the body vertical more heavily early in-flight, that this weighting falls during adaptation to microgravity, and that the decreased reliance on the body vertical persists early post-flight, generating an exaggerated sense of tilt. Since graviceptors respond to linear acceleration and not to head tilt in orbit, it has been proposed that adaptation to weightlessness entails reinterpretation of otolith activity, causing tilt to be perceived as translation. Since linear acceleration during in-flight centrifugation was always perceived as tilt, not translation, the findings do not support this hypothesis.

STS-90 Shuttle Project↗

NASA NTRS2007

Quaternion Averaging

Many applications require an algorithm that averages quaternions in an optimal manner. For example, when combining the quaternion outputs of multiple star trackers having this output capability, it is desirable to properly average the quaternions without recomputing the attitude from the the raw star tracker data. Other applications requiring some sort of optimal quaternion averaging include particle filtering and multiple-model adaptive estimation, where weighted quaternions are used to determine the quaternion estimate. For spacecraft attitude estimation applications, derives an optimal averaging scheme to compute the average of a set of weighted attitude matrices using the singular value decomposition method. Focusing on a 4-dimensional quaternion Gaussian distribution on the unit hypersphere, provides an approach to computing the average quaternion by minimizing a quaternion cost function that is equivalent to the attitude matrix cost function Motivated by and extending its results, this Note derives an algorithm that deterniines an optimal average quaternion from a set of scalar- or matrix-weighted quaternions. Rirthermore, a sufficient condition for the uniqueness of the average quaternion, and the equivalence of the mininiization problem, stated herein, to maximum likelihood estimation, are shown.

Markley, F. Landis↗

NASA NTRS2011

Enchanced interference cancellation and telemetry reception in multipath environments with a single paraboic dish antenna using a focal plane array

An Advanced Focal Plane Array ("AFPA") for parabolic dish antennas that exploits spatial diversity to achieve better channel equalization performance in the presence of multipath (better than temporal equalization alone), and which is capable of receiving from two or more sources within a field-of-view in the presence of multipath. The AFPA uses a focal plane array of receiving elements plus a spatio-temporal filter that keeps information on the adaptive FIR filter weights, relative amplitudes and phases of the incoming signals, and which employs an Interference Cancelling Constant Modulus Algorithm (IC-CMA) that resolves multiple telemetry streams simultaneously from the respective aero-nautical platforms. This data is sent to an angle estimator to calculate the target's angular position, and then on to Kalman filters FOR smoothing and time series prediction. The resulting velocity and acceleration estimates from the time series data are sent to an antenna control unit (ACU) to be used for pointing control.

Mukai, Ryan↗