Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Automatic Differentiation”

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 289 records · Page 16
All sourcesNASA NTRSDOE OSTI
NASA NTRSDate not supplied

A Formal Verification Framework for Runtime Assurance

The simplex architecture is an instance of Runtime Assurance (RTA) where a trusted component takes control of a safety-critical system when an untrusted component violates a safety property. This paper presents a formalization of the simplex RTA framework in the language of hybrid programs. A feature of this formal verification framework is that, for a given system, a specific instantiation can be created and its safety properties are guaranteed by construction. Instantiations may be kept at varying levels of generality, allowing for black box components, such as ML/AI-based controllers, to be modeled. The framework is written in the Prototype Verification System (PVS) using Plaidypvs, an embedding of differential dynamic logic in PVS. As a proof of concept, the framework is illustrated on an automatic vehicle braking system.

Runtime assurance↗
NASA NTRSDate not supplied

A Formal Verification Framework for Runtime Assurance

The simplex architecture is an instance of Runtime Assurance (RTA) where a trusted component takes control of a safety-critical system when an untrusted component violates a safety property. This paper presents a formalization of the simplex RTA framework in the language of hybrid programs. A feature of this formal verification framework is that, for a given system, a specific instantiation can be created and its safety properties are guaranteed by construction. Instantiations may be kept at varying levels of generality, allowing for black box components, such as ML/AI-based controllers, to be modeled. The framework is written in the Prototype Verification System (PVS) using Plaidypvs, an embedding of differential dynamic logic in PVS. As a proof of concept, the framework is illustrated on an automatic vehicle braking system.

Runtime assurance↗
NASA NTRS1996

Further Development of a New, Flux-Conserving Newton Scheme for the Navier-Stokes Equations

This paper is one of a series of papers describing the development of a new numerical approach for solving the steady Navier-Stokes equations. The key features in the current development are (1) the discrete representation of the dependent variables by way of high order polynomial expansions, (2) the retention of all derivatives in the expansions as unknowns to be explicitly solved for, (3) the automatic balancing of fluxes at cell interfaces, and (4) the discrete simulation of both the integral and differential forms of the governing equations. The main purpose of this paper is, first, to provide a systematic and rigorous derivation of the conditions that are used to simulate the differential form of the Navier-Stokes equations, and second, to extend our previously-presented internal flow scheme to external flows and nonuniform grids. Numerical results are presented for high Reynolds number flow (Re = 100,000) around a finite flat plate, and detailed comparisons are made with the Blasius flat plate solution and Goldstein wake solution. It is shown that the error in the streamwise velocity decreases like r(sup alpha)(Delta)y(exp 2), where alpha approx. 0.25 and r = delta(y)/delta(x) is the grid aspect ratio.

Scott, James R.↗
NASA NTRS1978

Numerical solution of flow problems using body-fitted coordinate systems

The technique of boundary-fitted coordinate systems is based on a method of automatic numerical generation of a general curvilinear coordinate system having a coordinate line coincident with each boundary of a general multi-connected region containing any number of arbitrarily shaped bodies. Once the curvilinear coordinate system is generated, any partial differential system of interest can be solved on this coordinate system by transforming the equations and solving the resulting system in finite difference approximation on the rectangular transformed plane. This method of automatic body-fitted curvilinear coordinate generation is used to construct finite-difference solutions of the full, time dependent Navier-Stokes equations for the unsteady viscous flow about arbitrary two-dimensional airfoils, or any other two-dimensional bodies. Finally, initial results for three-dimensional applications are also presented.

Thompson, J. F.↗
NASA NTRS1972

Algorithm for Liapunov stability analysis

Development of algorithm provides automatic computation of quadratic estimate of domain of stability for stable equilibrium states of nonlinear systems of ordinary differential equations.

Geiss, G. R.↗
NASA NTRS1977

TOMCAT - A code for numerical generation of boundary-fitted curvilinear coordinate systems on fields containing any number of arbitrary two-dimensional bodies

A method for automatic generation of boundary-fitted curvilinear coordinate systems, where the transformed coordinates are solutions of an elliptic differential system in the physical plane, and where the coordinate lines are coincident with all boundaries of a general multiply-connected, two-dimensional region containing any number of arbitrarily shaped bodies, and is described along with a suitable computer code for implementing the method. Any partial differential system can be solved on the boundary-fitted coordinate system by appropriate transformations. The transformed equations are approximated by finite differences and solved numerically in the transformed plane. All computations, whether for generating coordinate system or then solving the transformed equations, can be done on a rectangular field with square mesh with no interpolation required on the boundaries. The physical boundaries may even be time-dependent.

Thompson, J. F.↗
NASA NTRS1980

Advanced Gas Turbine (AGT) powertrain system initial development report

The powertrain consists of a single shaft regenerated gas turbine engine utilizing ceramic hot section components, coupled to a slit differential gearbox with an available variable stator torque converter and an available Ford intergral overdrive four-speed automatic transmission. Predicted fuel economy using gasoline fuel over the combined federal driving cycle (CFDC) is 15.3 km/1, which represents a 59% improvement over the spark-ignition-powered baseline vehicle. Using DF2 fuel, CFDC mileage estimates are 17.43 km/1. Zero to 96.6 km/hr acceleration time is 11.9 seconds with a four-second accleration distance of 21.0 m. The ceramic radial turbine rotor is discussed along with the control system for the powertrain.

Source record↗
NASA NTRS1991

The symbolic computation and automatic analysis of trajectories

Research was generally done on computation of trajectories of dynamical systems, especially control systems. Algorithms were further developed for rewriting expressions involving differential operators. The differential operators involved arise in the local analysis of nonlinear control systems. An initial design was completed of the system architecture for software to analyze nonlinear control systems using data base computing.

Grossman, Robert↗
DOE OSTI2023

Differentiable Neural Architecture, Mixed Precision and Accelerator Co-Search

Quantization, effective Neural Network architecture, and efficient accelerator hardware are three important design paradigms to maximize accuracy and efficiency. Mixed Precision Quantization is a process of assigning different precision to different Neural Network layers for optimized inference. Neural Architecture Search (NAS) is a process of automatically designing the neural network for a task and can also be extended to search for the precision of each weight and activation matrix. In this paper, we develop the following three methods: (i) Fast Differentiable Hardware-aware Mixed Precision Quantization Search method to find optimal precision, (ii) Joint Differentiable hardware-aware Architecture and Mixed Precision Quantization Co-search, (iii) Joint Accelerator, Architecture, and Precision triple co-search to find best possibilities in all the three worlds. We demonstrate the effectiveness of our proposed methods targeting Bitfusion accelerator by searching mixed precision models on MobilenetV2. We achieve better accuracy-latency trade-off models than the manually designed and previously proposed search methods.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Regularized Differentiation for Bioburden Density Estimation in Planetary Protection

In this paper, we propose and investigate the performance of two novel shrinkage estimators for bioburden density estimation in planetary protection. The estimators are based on the regularized differentiation of a cumulative count of colony forming units collected throughout the data collecting session or the life cycle of the entire mission. The regularized differentiation recasts the problem of bioburden density estimation as a linear least squares problem. The least squares problem is then solved through regularization techniques, such as truncated singular value decomposition and penalized least squares. The regularization is necessary to avoid noise amplification during the differentiation of noisy data. The two regularization estimators are compared with four other commonly used estimators to simultaneously evaluate the means of multivariable independent Poisson distributions: the maximum likelihood, noninformative Bayes estimator with Jeffreys prior, Empirical Bayes using conjugate gamma-Poisson model with gamma parameters selected by method of moments, and the Clevenson-Zidek estimator. It is shown through computer-simulated data that the regularized differentiation based on ridge regression has the smallest mean-squared error among all estimators. The analysis of shrinkage mechanism implemented by regularized differentiation is performed, and it is shown that the regularized differentiation amounts to performing a weighted averaging of all the samples. The weights are determined by the regularization parameter automatically selected by the L-curve technique. Since the method of least squares makes no distributional assumptions about the data, it presents an attractive technique for bioburden density estimation when there are concerns about the misspecification of the distributional model. The paper concludes with the analysis of the bioburden data collected during InSight mission and directions for future work.

97 - MATHEMATICS AND COMPUTING↗
NASA NTRS1984

Advanced AFCS developments on the XV-15 tilt rotor research aircraft

The design criteria and control and handling qualities of the Automatic Flight Control System (AFCS), developed in the framework of the XV-15 tilt-rotor research aircraft, are evaluated, differentiating between the stability and control criteria. A technically aggressive SCAS control law was implemented, demonstrating that significant benefits accrue when stability criteria are separated from design criteria; the design analyses for application of the control law are presented, and the limit bandwidth for stabilization in hovering flight is shown to be defined by rotor or control lag functions. Flight tests of the aircraft resulted in a rating of 3 on the Cooper-Harper scale; a possibility of achieving a rating of 2 is expected if the system is applied to the yaw and heave control modes.

Churchill, G. B.↗
DOE OSTI2023

Applying Physics-Informed Neural Networks to Solve Navier–Stokes Equations for Laminar Flow around a Particle

In recent years, Physics-Informed Neural Networks (PINNs) have drawn great interest among researchers as a tool to solve computational physics problems. Unlike conventional neural networks, which are black-box models that “blindly” establish a correlation between input and output variables using a large quantity of labeled data, PINNs directly embed physical laws (primarily partial differential equations) within the loss function of neural networks. By minimizing the loss function, this approach allows the output variables to automatically satisfy physical equations without the need for labeled data. The Navier–Stokes equation is one of the most classic governing equations in thermal fluid engineering. This study constructs a PINN to solve the Navier–Stokes equations for a 2D incompressible laminar flow problem. Flows passing around a 2D circular particle are chosen as the benchmark case, and an elliptical particle is also examined to enrich the research. The velocity and pressure fields are predicted by the PINNs, and the results are compared with those derived from Computational Fluid Dynamics (CFD). Additionally, the particle drag force coefficient is calculated to quantify the discrepancy in the results of the PINNs as compared to CFD outcomes. The drag coefficient maintained an error within 10% across all test scenarios.

Hu, Beichao (ORCID:0009000163215151)↗
NASA NTRS1976

Numerical simulation of low gravity draining

A boundary value problem was solved numerically for a liquid that is assumed to be inviscid and incompressible, having a motion that is irrotational and axisymmetric, and having a constant (5 degrees) solid-liquid contact angle. The avoidance of excessive mesh distortion, encountered with strictly Lagrangian or Eulerian kinematics, was achieved by introducing an auxiliary kinematic velocity field along the free surface in order to vary the trajectories used in integrating the ordinary differential equations simulating the moving boundary. The computation of the velocity potential was based upon a nonuniform triangular mesh which was automatically revised to varying depths to accommodate the motion of the free surface. These methods permitted calculation of draining induced axisymmetric slosh through the many (or fractional) finite amplitude oscillations that can occur depending upon the balance of draining, gravitational, and surface tension forces. Velocity fields, evolution of the free surface with time, and liquid residual volumes were computed for three and one half decades of Weber number and for two Bond numbers, tank fill levels, and drain radii. Comparisons with experimental data are very satisfactory.

Bizzell, G. D.↗
NASA NTRS1989

The soft x ray telescope for Solar-A

The Solar-A satellite being prepared by the Institute for Sapce and Astronautical Sciences (ISAS) in Japan is dedicated to high energy observations of solar flares. The Soft X Ray Telescope (SXT) is being prepared to provide filtered images in the 2 to 60 A interval. The flight model is now undergoing tests in the 1000 foot tunnel at MSFC. Launch will be in September 1991. Earlier resolution and efficiency tests on the grazing incidence mirror have established its performance in soft x rays. The one-piece, two mirror grazing incidence telescope is supported in a strain free mount separated from the focal plane assembly by a carbon-epoxy metering tube whose windings and filler are chosen to minimize thermal and hygroscopic effects. The CCD detector images both the x ray and the concentric visible light aspect telescope. Optical filters provide images at 4308 and 4700 A. The SXT will be capable of producing over 8000 of the smallest partial frame images per day, or fewer but larger images, up to 1024 x 1024 pixel images. Image sequence with two or more of the five x ray analysis filters, with automatic exposure compensation to optimize the charge collection by the CCD detector, will be used to provide plasma diagnostics. Calculations using a differential emission measure code were used to optimize filter selection over the range of emission measure variations and to avoid redundancy, but the filters were chosen primarily to give ratios that are monotonic in plasma temperature.

Brown, W. A.↗
NASA NTRS1995

Design of an anti-Rician-fading modem for mobile satellite communication systems

To design a demodulator applicable to mobile satellite communication systems using differential phase shift keying modulation, we have developed key technologies including an anti-Rician-fading demodulation scheme, an initial acquisition scheme, automatic gain control (AGC), automatic frequency control (AFC), and bit timing recovery (BTR). Using these technologies, we have developed one-chip digital signal processor (DSP) modem for mobile terminal, which is compact, of light weight, and of low power consumption. Results of performance test show that the developed DSP modem achieves good performance in terms of bit error ratio in mobile satellite communication environment, i.e., Rician fading channel. It is also shown that the initial acquisition scheme acquires received signal rapidly even if the carrier-to-noise power ratio (CNR) of the received signal is considerably low.

Kojima, Toshiharu↗
NASA NTRS1995

A technique for optimizing grid blocks

A new technique for automatically combining grid blocks of a given block-structured grid into logically-rectangular clusters which are 'optimal' is presented. This technique uses the simulated annealing optimization method to reorganize the blocks into an optimum configuration, that is, one which minimizes a user-defined objective function such as the number of clusters or the differential in the sizes of all the clusters. The clusters which result from applying the technique to two different two-dimensional configurations are presented for a variety of objective function definitions. In all cases, the automatically-generated clusters are significantly better than the original clusters. While this new technique can be applied to block-structured grids generated from any source, it is particularly useful for operating on block-structured grids containing many blocks, such as those produced by the emerging automatic block-structured grid generators.

Dannenhoffer, John F., III↗
NASA NTRS1972

Numerical method for the solution of large systems of differential equations of the boundary layer type

A numerical method for the solution of large systems of nonlinear differential equations of the boundary-layer type is described. The method is a modification of the technique for satisfying asymptotic boundary conditions. The present method employs inverse interpolation instead of the Newton method to adjust the initial conditions of the related initial-value problem. This eliminates the so-called perturbation equations. The elimination of the perturbation equations not only reduces the user's preliminary work in the application of the method, but also reduces the number of time-consuming initial-value problems to be numerically solved at each iteration. For further ease of application, the solution of the overdetermined system for the unknown initial conditions is obtained automatically by applying Golub's linear least-squares algorithm. The relative ease of application of the proposed numerical method increases directly as the order of the differential-equation system increases. Hence, the method is especially attractive for the solution of large-order systems. After the method is described, it is applied to a fifth-order problem from boundary-layer theory.

Green, M. J.↗