SEARCH · Engineering Papers
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.
This document describes the Airport Configuration Prediction Model (ACPM) Machine Learning (ML) service developed under the ATD-2 subproject as a reference implementation and released by NASA as Open Source software on github.
Here, this paper presents implementation of a model predictive controller (MPC) for an experimental solid oxide fuel cell (SOFC) system. The MPC controller is based on a gain-scheduled predictor with block-oriented structure that can capture important non-linear effects while still keeping the computational complexity low enough to meet real time control requirements. Experimental results show the MPC is able to regulate the SOFC cathode outlet temperature in the face of startup transients and input perturbations.
A data-driven model predictive control (MPC) was developed to enable the self-regulating capability of heat pipe (HP) nuclear microreactors. The MPC can proactively respond to potential disturbances of HP microreactors using three approaches for system identifications: linear state-space model, feedforward neural network, and recurrent neural networks with long short-term memory units. We present numerical results of data-driven MPCs to control the temperatures of selected HPs in a 37-HP test article. Our results show qualitatively that all data-driven MPCs produced similar control actions, while quantitatively, with artificial neural nets (especially feedforward neural nets), MPC can better follow drastic changes in setpoints with small errors.
Abstract In‐silico screening of novel biofuel molecules based on chemical and fuel properties is a critical first step in the biofuel evaluation process due to the significant volumes of samples required for experimental testing, the destructive nature of engine tests, and the costs associated with bench‐scale synthesis of novel fuels. Predictive models are limited by training sets of few existing measurements, often containing similar classes of molecules that represent just a subset of the potential molecular fuel space. Software tools can be used to generate every possible molecular descriptor for use as input features, but most of these features are largely irrelevant and training models on datasets with higher dimensionality than size tends to yield poor predictive performance. Feature selection has been shown to improve machine learning models, but correlation‐based feature selection fails to provide scientific insight into the underlying mechanisms that determine structure–property relationships. The implementation of causal discovery in feature selection could potentially inform the biofuel design process while also improving model prediction accuracy and robustness to new data. In this study, we investigate the benefits causal‐based feature selection might have on both model performance and identification of key molecular substructures. We found that causal‐based feature selection performed on par with alternative filtration methods, and that a structural causal model provides valuable scientific insights into the relationships between molecular substructures and fuel properties.
New workflows have been developed for predictive modeling of magnetohydrodynamic (MHD) equilibrium in tokamak plasmas. The goal of this work is to predict the MHD equilibrium in tokamak discharges without having measurements of the kinetic profiles. The workflows include a cold start tool, which constructs all the profiles and power flows needed by transport codes; a Grad–Shafranov equilibrium solver; and various codes for the sources and sinks. For validation purposes, a database of DIII-D tokamak discharges has been constructed that is comprised of scans in the plasma current, toroidal magnetic field, and triangularity. Initial efforts focused on developing a workflow utilizing an empirically derived pressure model tuned to DIII-D discharges with monotonic safety factor profiles. This workflow shows good agreement with experimental kinetic equilibrium calculations, but is limited in that it is a single fluid (equal ion and electron temperatures) model and lacks H-mode pedestal predictions. The best agreement with the H-mode database is obtained using a theory-based workflow utilizing pressure profile predictions from a coupled TGLF turbulent transport and EPED pedestal models together with external magnetics and Motional Stark Effect (MSE) data to construct the equilibrium. Here, we obtain an average root mean square error of 5.1% in the safety factor profile when comparing the predicted and experimental kinetic equilibrium. We also find good agreement with the plasma stored energy, internal inductance, and pressure profiles. Including MSE data in the theory-based workflow results in noticeably improved agreement with the q-profiles in high triangularity discharges in comparison with the results obtained with magnetic data only. The predictive equilibrium workflow is expected to have wide applications in experimental planning, between-shot analysis, and reactor studies.
Stationary batteries in residential and commercial buildings are often used to smooth customer load profiles and to lower customer electricity bills. Controllers for these battery systems should account for customer energy consumption, rate structures, and high internal battery temperatures, which can lead to reduced performance over the battery lifetime. It is important to consider the uncertainty in forecasting energy consumption and temperature, especially for customers with highly variable and uncertain loads. We propose a novel battery controller using stochastic model predictive control that accounts for these uncertainties and can handle complex rate structures, including demand charges. We show that the controller performs better than standard model predictive control when there is significant uncertainty in the forecast. We also show improvements in the performance with more accurate forecasts and with a more aggressive control strategy.
Stationary batteries in residential and commercial buildings are often used to smooth customer load profiles and to lower customer electricity bills. Controllers for these battery systems should account for customer energy consumption, rate structures, and high internal battery temperatures, which can lead to reduced performance over the battery lifetime. It is important to consider the uncertainty in forecasting energy consumption and temperature, especially for customers with highly variable and uncertain loads. We propose a novel battery controller using stochastic model predictive control that accounts for these uncertainties and can handle complex rate structures, including demand charges. We show that the controller performs better than standard model predictive control when there is significant uncertainty in the forecast. We also show improvements in the performance with more accurate forecasts and with a more aggressive control strategy.
Stationary batteries in residential and commercial buildings are often used to smooth customer load profiles and to lower customer electricity bills. Controllers for these battery systems should account for customer energy consumption, rate structures, and high internal battery temperatures, which can lead to reduced performance over the battery lifetime. It is important to consider the uncertainty in forecasting energy consumption and temperature, especially for customers with highly variable and uncertain loads. We propose a novel battery controller using stochastic model predictive control that accounts for these uncertainties and can handle complex rate structures, including demand charges. We show that the controller performs better than standard model predictive control when there is significant uncertainty in the forecast. We also show improvements in the performance with more accurate forecasts and with a more aggressive control strategy.
We describe a genetic programming system which learns nonlinear predictive models for lossless image compression.
A prediction model of signal degradation in a Land Mobile Satellite Service (LMSS) for urban areas is proposed. This model treats shadowing effects caused by buildings statistically and can predict a Cumulative Distribution Function (CDF) of signal diffraction losses in urban areas as a function of system parameters such as frequency and elevation angle and environmental parameters such as number of building stories and so on. In order to examine the validity of the model, we compared the percentage of locations where diffraction losses were smaller than 6 dB obtained by the CDF with satellite visibility measured by a radiometer. As a result, it was found that this proposed model is useful for estimating the feasibility of providing LMSS in urban areas.
An enhanced Finite Control Set Model Predictive Control (FCS-MPC) strategy with a two-step prediction horizon for a Doubly-Fed Induction Generator (DFIG) is the focus of this paper. The DFIG and RL filter's discrete-time model is used in the proposed control scheme for a two-step prediction horizon of rotor and filter currents for the converter's eight possible switching states. Afterward, the control algorithm selects the ideal switching state, which minimizes currents' objective function. The proposed control scheme does not require a modulation stage for internal controllers. We use Lagrange extrapolations to predict the dynamic changes in currents. The switching frequency reduction is achieved by integrating an optimization constraint in the algorithm's cost function. Furthermore, the Total Harmonic Distortion level of currents is kept below 5%, according to IEEE Std 519-14. Obtained results reveal that the switching frequency of the DFIG converters is considerably reduced without losing control; it is reduced by 13.89KHz for the Grid Side Converter (GSC) and by 1.78KHz for the Rotor Side Converter (RSC).
Experimental verification of mathematical model performance predictions for gamma ray atmospheric density sensors
We developed a new prediction model for forecasting relativistic (greater than 2MeV) electrons, which provides a VERY HIGH correlation between predicted and actually measured electron fluxes at geostationary orbit. This model implies the multi-step particle acceleration and is based on numerical integrating two linked continuity equations for primarily accelerated particles and relativistic electrons. The model includes a source and losses, and used solar wind data as only input parameters. We used the coupling function which is a best-fit combination of solar wind/interplanetary magnetic field parameters, responsible for the generation of geomagnetic activity, as a source. The loss function was derived from experimental data. We tested the model for four year period 2004-2007. The correlation coefficient between predicted and actual values of the electron fluxes for whole four year period as well as for each of these years is stable and incredibly high (about 0.9). The high and stable correlation between the computed and actual electron fluxes shows that the reliable forecasting these electrons at geostationary orbit is possible.
Abstract Genetic redundancy refers to a situation where an individual with a loss-of-function mutation in one gene (single mutant) does not show an apparent phenotype until one or more paralogs are also knocked out (double/higher-order mutant). Previous studies have identified some characteristics common among redundant gene pairs, but a predictive model of genetic redundancy incorporating a wide variety of features derived from accumulating omics and mutant phenotype data is yet to be established. In addition, the relative importance of these features for genetic redundancy remains largely unclear. Here, we establish machine learning models for predicting whether a gene pair is likely redundant or not in the model plant Arabidopsis thaliana based on six feature categories: functional annotations, evolutionary conservation including duplication patterns and mechanisms, epigenetic marks, protein properties including posttranslational modifications, gene expression, and gene network properties. The definition of redundancy, data transformations, feature subsets, and machine learning algorithms used significantly affected model performance based on holdout, testing phenotype data. Among the most important features in predicting gene pairs as redundant were having a paralog(s) from recent duplication events, annotation as a transcription factor, downregulation during stress conditions, and having similar expression patterns under stress conditions. We also explored the potential reasons underlying mispredictions and limitations of our studies. This genetic redundancy model sheds light on characteristics that may contribute to long-term maintenance of paralogs, and will ultimately allow for more targeted generation of functionally informative double mutants, advancing functional genomic studies.
A dynamic rain attenuation prediction model is developed for use in obtaining the temporal characteristics, on time scales of minutes or hours, of satellite communication link availability. Analagous to the associated static rain attenuation model, which yields yearly attenuation predictions, this dynamic model is applicable at any location in the world that is characterized by the static rain attenuation statistics peculiar to the geometry of the satellite link and the rain statistics of the location. Such statistics are calculated by employing the formalism of Part I of this report. In fact, the dynamic model presented here is an extension of the static model and reduces to the static model in the appropriate limit. By assuming that rain attenuation is dynamically described by a first-order stochastic differential equation in time and that this random attenuation process is a Markov process, an expression for the associated transition probability is obtained by solving the related forward Kolmogorov equation. This transition probability is then used to obtain such temporal rain attenuation statistics as attenuation durations and allowable attenuation margins versus control system delay.
Efforts to build a Predictive Material Modeling (PMM) framework from the micro-scale to the macro-scale are presented in this abstract. To reduce the need for extensive testing, accelerate the design cycle process, and reduce uncertainty margins applied to final designs, NASA is developing simulation and modeling tools that enable characterization of material properties and response to high-enthalpy environments. The Porous Microstructure Analysis (PuMA) code has been developed for computing macroscale (volume averaged) properties of porous materials using microscale images from micro-computed tomography (micro-CT). Microscale modeling requires a realistic representation of a material microstructure; these are obtained either synthetically during the design of the material or through X-ray micro-CT. Volume averaged properties are then used to inform macroscale material response models, such as those implemented in the Porous-material Analysis Toolbox based on OpenFOAM (PATO) software, also actively developed by NASA. The computational model in PATO is a generic heat and mass transfer model for porous reactive materials containing several solid phases and a single gas phase. The detailed chemical interactions occurring between the solid phases and the gas phase are modeled at the pore scale assuming local thermal equilibrium. These tools were developed to efficiently interface with other pre-existing codes such as SPARTA (direct simulation Monte Carlo), DPLR (hypersonic CFD), NEQAIR (radiative transport) and DAKOTA (uncertainty quantification and optimization). Detailed flight data (Mars Science Laboratory [MSL] Entry Descent and Landing Instrument [MEDLI]) is critical for validating these computational tools for NASA applications. Examples of modeling ablative material response using these codes will be presented including 3D simulations of the full-scale heatshield of the MSL capsule. The simulations demonstrate the ability of the modern material response code, PATO, to handle the material response of geometrically complex and large domains, through the use of massively parallel computations.
A robustly stabilizing MPC (model predictive control) algorithm for uncertain nonlinear systems is developed that guarantees resolvability. With resolvability, initial feasibility of the finite-horizon optimal control problem implies future feasibility in a receding-horizon framework. The control consists of two components; (i) feed-forward, and (ii) feedback part. Feed-forward control is obtained by online solution of a finite-horizon optimal control problem for the nominal system dynamics. The feedback control policy is designed off-line based on a bound on the uncertainty in the system model. The entire controller is shown to be robustly stabilizing with a region of attraction composed of initial states for which the finite-horizon optimal control problem is feasible. The controller design for this algorithm is demonstrated on a class of systems with uncertain nonlinear terms that have norm-bounded derivatives and derivatives in polytopes. An illustrative numerical example is also provided.
This paper presents a summary of the life prediction methods developed under the NASA Lewis Research Center's Hot Section Technology (HOST) program. A major objective of the fatigue and fracture efforts under the HOST program was to significantly improve the analytic life prediction tools used by the aeronautical gas turbine engine industry. This has been achieved in the areas of high-temperature thermal and mechanical fatigue of bare and coated high-temperature superalloys. Such technical improvements will eventually reduce life cycle costs. The cyclic crack initiation and propagation resistance of nominally isotropic polycrystalline alloys and highly anisotropic single crystal alloys has been addressed. A sizeable data base has been generated for three alloys [cast PWA 1455 (B-1900 + Hf), wrought Inconel 718, and cast single crystal PWA 1480] in bare and coated conditions. Two coatings systems, diffusion aluminide (PWA 273) and plasma sprayed MCrAIY overlay (PWA 286) were employed. Life prediction modeling efforts were devoted to creep-fatigue interaction, oxidation, coatings interactions, multiaxiality of stress-strain states, mean stress effects, cumulative damage, and thermomechanical fatigue. The fatigue crack initiation life models developed to date include the Cyclic Damage Accumulation (CDA) Model of Pratt & Whitney and the Total Strain Version of Strainrange Partitioning (TS-SRP) of NASA Lewis for nominally isotropic materials, and the Tensile Hysteretic Energy Model of Pratt & Whitney for anisotropic superalloys. The fatigue model being developed by the General Electric Company is based upon the concepts of Path-lndependent Integrals (PII) for describing cyclic crack growth under complex nonlinear response at the crack tip due to thermomechanical loading conditions. A micromechanistic oxidation crack extension model has been derived by researchers at Syracuse University. The models are described and discussed in the paper. Only limited verification has been achieved to-date as several of the technical programs are still in progress and the verification tasks are scheduled, quite naturally, near the conclusion of the program. To-date, efforts have concentrated on development of independent models for cyclic constitutive behavior, cyclic crack initiation, and cyclic crack propagation. The transition between crack initiation and crack propagation has not been thoroughly researched as yet, and the integration of these models into a unified life prediction method has not been addressed.