Search NASA⌕ Search

SEARCH · Search NASA

Results for “Nonlinear model identification”

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.

412 records · Page 23

Using Machine Learning to Infer Material Properties of Debris Fragments from X-ray Images in the DebriSat Project

The DebriSat project is a collaboration effort with the NASA Orbital Debris Program Office, the U.S. Space Force Space Systems Command Center, The Aerospace Corporation, and the University of Florida. To date, over 200,000 fragments from this ground-based, hypervelocity impact experiment have been collected, and processing is underway to determine their physical characteristics, such as material, shape, color, characteristic length, and average cross-sectional area. The x-ray process is primarily used to identify the location of the fragments and estimated size for extraction, so that these physical characteristics can be assessed. This paper proposes a machine learning-based approach to characterize materials from x-ray images of debris fragments embedded in soft-catch foam used in the DebriSat project. The novel methodology discussed in this paper will highlight the use of x-ray imagery data to characterize these fragments without extraction or a human-in-the-loop. Both supervised and unsupervised machine learning techniques are utilized with this approach to infer the physical parameters of the fragments embedded in the soft-catch foam panels used in the impact experiment based on x-ray images of the foam panels. Additionally, 3D reconstructions of the extracted fragments are created with images taken from two different angles using the structure from motion (SfM) method. The characteristic lengths and shape from the 3D reconstruction, alongside the physical characteristics of the debris, are used in the inference of the material type. To develop and test the approach, a dataset of x-ray images of debris fragments of varying sizes and materials is collected. Supervised learning methods such as convolutional neural networks (CNNs), support vector machines (SVM), decision trees, and random forest classifiers are used due to the high-dimensional feature spaces of the debris and nonlinear decision boundaries for material categorization. Given the limited pre-labeled data of embedded debris materials smaller than 10 mm, unsupervised machine learning techniques such as clustering algorithms and autoencoders are used, in addition to supervised learning methods. The clustering algorithms group similar fragments together based on their physical properties, and autoencoders reduce the dimensionality of the x ray images and extract relevant features. The performance of the proposed approach's is analyzed using a range of statistical methods, including confusion matrices, receiver operating characteristic curves, and precision-recall curves. The results are compared with those obtained using a baseline approach that relies on manual identification and classification of debris fragments. To evaluate the effectiveness of different machine learning methods, statistical tests such as t-tests, ANOVA, and cross-validation are performed, comparing the performance of CNNs, SVMs, clustering algorithms, and autoencoders. Additional analysis needs to be conducted to identify any sources of bias or variability that may affect the results, such as variations in imaging conditions or fragmentation patterns. Other topics explored are limitations, refinements, and the potential use of semi-supervised learning techniques, such as self-training to label unlabeled datasets and co-training using x-ray images taken from two different angles as two different models.

Saik Anam Siam↗

Stochastic Learning Approach for Binary Optimization: Application to Bayesian Optimal Design of Experiments

Here, we present a novel stochastic approach to binary optimization suited for optimal experimental design (OED) for Bayesian inverse problems governed by mathematical models such as partial differential equations. The OED utility function, namely, the regularized optimality criterion, is cast into a stochastic objective function in the form of an expectation over a multivariate Bernoulli distribution. The probabilistic objective is then solved by using a stochastic optimization routine to find an optimal observational policy. This formulation (a) is generally applicable to binary optimization problems with soft constraints and is ideal for OED and sensor placement problems; (b) does not require differentiability of the original objective function (e.g., a utility function in OED applications) with respect to the design variable, and thus it enables direct employment of sparsity-enforcing penalty functions such as $\ell_0$, without needing to utilize a continuation procedure or apply a rounding technique; (c) exhibits much lower computational cost than traditional gradient-based relaxation approaches; and (d) can be applied to both linear and nonlinear OED problems with proper choice of the utility function. The proposed approach is analyzed from an optimization perspective with detailed convergence analysis of the optimization approach and is also analyzed from a machine learning perspective with correspondence to policy gradient reinforcement learning. The approach is demonstrated numerically by using an idealized two-dimensional Bayesian linear inverse problem and validated by extensive numerical experiments carried out for sensor placement in a parameter identification setup.

97 MATHEMATICS AND COMPUTING↗

Non-linear saturation mechanism of electron temperature gradient modes

The electron temperature gradient (ETG) mode is a very plausible candidate to explain the large electron particle transport and thermal conduction. Production and identification of slab ETG modes and measurement electron transport have been already reported [X. Wei, V. Sokolov, and A. K. Sen, Phys. Plasmas 17, 042108 (2010); V. Sokolov and A. K. Sen, Phys. Rev. Lett. (2011)]. Now, we develop a theoretical model of non-linear saturation mechanism of ETG mode based on the three wave coupling of an unstable high frequency ETG mode with a damped ETG radial harmonic and a damped ion acoustic (IA) mode. Bicoherence analysis of Columbia linear machine (CLM) data show coupling between ETG modes ({approx}2.4 MHz) and a low frequency mode ({approx}50 kHz). The large damping drive of the ETG radial harmonic accompanied by the smaller but finite damping of the IA mode presents an energy sink for the unstable ETG mode, thus causing saturation. This model predicts a saturation level of {approx}10% and agrees with the observed levels of ETG modes in the CLM.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Overcoming the Technical Challenges of Coordinating Distributed Load Resources at Scale (Final Report)

Significant recent research has investigated the potential for loads to provide balancing services to the grid. However, this research has not addressed key issues that may arise when such schemes are applied at scale including: 1. Distribution Network Issues. Coordination of large numbers of loads could result in power flows that violate distribution network constraints; 2. Stability Issues. Certain strategies to control loads can exhibit nonlinearity in the form of period-adding bifurcations and chaos. Other control strategies can potentially synchronize the behavior of large numbers of loads. In both cases, the outcome can be power oscillations and instability; 3. Communication Network Issues. Bidirectional low-latency communication channels between a central controller (or several distributed controllers) and each resource are expensive and likely not necessary for effective coordination. Our research questions were: What network, stability, and communication issues might arise in practice when we coordinate large aggregations of loads? How can we coordinate loads to achieve performance objectives in a cost effective manner while avoiding these issues? The ultimate technical goal of the project was the development of network-aware, communication-constrained, non-disruptive load control strategies with stability guarantees that achieve the performance requirements of typical balancing services at a sufficiently low cost to enable the load aggregator and customer to profit. The overall goal was to establish credibility for load control at scale and contribute to U.S. energy security and environmental goals. The team succeeded in answering these research questions and developing these control strategies. The overall approach was based on the development of three testing environments: a simulation testbed, an experimental testbed (20 physical model houses with window-box air conditioners) coupled with the simulation testbed, and a field testbed (100 actual homes in Austin, TX) coupled with the simulation testbed, which enabled controller testing, identification of issues, controller development, and controller validation. The resulting controller was used to demonstrate fast timescale grid balancing (frequency regulation) by aggregations of physical and virtual air conditioners, with sufficient quality to participate in the electricity market. Cost benefit analysis showed overall benefits to the participating households, load aggregators, and the grid, especially if the control technology was integrated directly into existing programmable communicating thermostats. The project provides a variety of wider benefits. Our technology transfer and outreach activities lead us to choose an open-source licensing commercialization pathway, enabling the project results to be available to researchers, industry, and the public. Furthermore, new grid balancing technologies will increase grid flexibility and will enable higher penetrations of intermittent renewable energy resources, such as wind and solar, to be connected to the grid, reducing its environmental impact, and mitigating climate change to the benefit of society.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Effective Materials Property Information Management for the 21st Century

This paper discusses key principles for the development of materials property information management software systems. There are growing needs for automated materials information management in industry, research organizations and government agencies. In part these are fuelled by the demands for higher efficiency in material testing, product design and development and engineering analysis. But equally important, organizations are being driven to employ sophisticated methods and software tools for managing their mission-critical materials information by the needs for consistency, quality and traceability of data, as well as control of access to proprietary or sensitive information. Furthermore the use of increasingly sophisticated nonlinear, anisotropic and multi-scale engineering analysis approaches, particularly for composite materials, requires both processing of much larger volumes of test data for development of constitutive models and much more complex materials data input requirements for Computer-Aided Engineering (CAE) software. And finally, the globalization of engineering processes and outsourcing of design and development activities generates much greater needs for sharing a single gold source of materials information between members of global engineering teams in extended supply-chains. Fortunately material property management systems have kept pace with the growing user demands. They have evolved from hard copy archives, through simple electronic databases, to versatile data management systems that can be customized to specific user needs. The more sophisticated of these provide facilities for: (i) data management functions such as access control, version control, and quality control; (ii) a wide range of data import, export and analysis capabilities; (iii) mechanisms for ensuring that all data is traceable to its pedigree sources: details of testing programs, published sources, etc; (iv) tools for searching, reporting and viewing the data; and (v) access to the information via a wide range of interfaces, including web browsers, rich clients, programmatic access and clients embedded in third-party applications, such as CAE systems. This paper discusses the important requirements for advanced material data management systems as well as the future challenges and opportunities such as automated error checking, automated data quality assessment and characterization, identification of gaps in data, as well as functionalities and business models to keep users returning to the source: to generate user demand to fuel database growth and maintenance.

36 MATERIALS SCIENCE↗

Weighted Composition Operators for Learning Nonlinear Dynamics

Operator theoretic methods in dynamical system have been dominated by the use of Koopman operators and their continuous time counterparts, such as Koopman Generators and Liouville Operators. The advantage gained from their use primarily stems from the ability to extract subspaces and eigenfunctions within a space of observables that are invariant with respect to the Koopman operator over that space. When this occurs, a dynamic mode decomposition of the systems state provides a linear model for the dynamical system. Not all Koopman operators have eigenfunctions that may be exploited in this manner. However, the framework can still be leveraged for approximations using other operators. In this setting, we present a different operator for the study of dynamical systems, the weighted composition operator. These operators are compact for a wide range of dynamics and spaces, and through their interactions with occupation kernels and vector valued kernels, they admit an estimation of the underlying dynamics. Here, this manuscript presents a new algorithm for the data driven study of dynamical systems from data, and also provides two numerical experiments where convergence is achieved as a proof of concept.

97 MATHEMATICS AND COMPUTING↗

Synthesis and simulation of a two-level magnetic control system for tokamak-reactor plasma

Synthesis and simulation of a hierarchical (two-level) magnetic system for controlling a tokamakreactor plasma throughout the entire divertor discharge stage, including the plasma current ramp-up phase, are carried out. The plasma vertical velocity is stabilized about zero by using a proportional controller in a scalar control loop. The gain of the controller-the coefficient that ensures the required stability margins-is found by using a second-order linear model constructed by solving the identification problem on the basis of numerical experiments carried out with the DINA plasmophysical computer code. The internal cascade (the lower level of the system) for tracking the scenario currents in the poloidal magnetic field coils is synthesized by using the complete dynamic channel decoupling method. The external cascade (the upper level of the system) for tracking the plasma current and shape is synthesized by using the method of pseudoseparation of the control channels and the multidimensional diagonal proportional-integral controller, with proportional, integrating, and double integrating units connected in parallel in each channel. In the hierarchical control system, the lower level (the internal cascade) is subordinated to the upper level (the external cascade). The external cascade acts on the internal one by the signals that set the required currents in the coils of the central solenoid and of the poloidal magnetic field in order to ensure the required plasma current and shape in accordance with the output signals from the plant, which are transmitted through the vector feedback channel. The lower level is aimed exclusively at tracking the reference inputs by tracking the currents in the control coils. An operating mode of the system under the conditions of current saturation in the control coils is proposed and implemented. Results are presented from numerical simulations of the two-level (cascade) control system for reference scenario no. 2 of the ITER database (www.iter.org) with the DINA nonlinear code.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Revisiting the blocking force test on ferroelectric ceramics using high energy x-ray diffraction

The blocking force test is a standard test to characterise the properties of piezoelectric actuators. The aim of this study is to understand the various contributions to the macroscopic behaviour observed during this experiment that involves the intrinsic piezoelectric effect, ferroelectric domain switching, and internal stress development. For this purpose, a high energy diffraction experiment is performed in-situ during a blocking force test on a tetragonal lead zirconate titanate (PZT) ceramic (Pb{sub 0.98}Ba{sub 0.01}(Zr{sub 0.51}Ti{sub 0.49}){sub 0.98}Nb{sub 0.02}O{sub 3}). It is shown that the usual macroscopic linear interpretation of the test can also be performed at the single crystal scale, allowing the identification of local apparent piezoelectric and elastic properties. It is also shown that despite this apparent linearity, the blocking force test involves significant non-linear behaviour mostly due to domain switching under electric field and stress. Although affecting a limited volume fraction of the material, domain switching is responsible for a large part of the macroscopic strain and explains the high level of inter- and intra-granular stresses observed during the course of the experiment. The study shows that if apparent piezoelectric and elastic properties can be identified for PZT single crystals from blocking stress curves, they may be very different from the actual properties of polycrystalline materials due to the multiplicity of the physical mechanisms involved. These apparent properties can be used for macroscopic modelling purposes but should be considered with caution if a local analysis is aimed at.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Nonlinear dynamics of drops and bubbles and chaotic phenomena

Nonlinear phenomena associated with the dynamics of free drops and bubbles are investigated analytically, numerically and experimentally. Although newly developed levitation and measurement techniques have been implemented, the full experimental validation of theoretical predictions has been hindered by interfering artifacts associated with levitation in the Earth gravitational field. The low gravity environment of orbital space flight has been shown to provide a more quiescent environment which can be utilized to better match the idealized theoretical conditions. The research effort described in this paper is a closely coupled collaboration between predictive and guiding theoretical activities and a unique experimental program involving the ultrasonic and electrostatic levitation of single droplets and bubbles. The goal is to develop and to validate methods based on nonlinear dynamics for the understanding of the large amplitude oscillatory response of single drops and bubbles to both isotropic and asymmetric pressure stimuli. The first specific area on interest has been the resonant coupling between volume and shape oscillatory modes isolated gas or vapor bubbles in a liquid host. The result of multiple time-scale asymptotic treatment, combined with domain perturbation and bifurcation methods, has been the prediction of resonant and near-resonant coupling between volume and shape modes leading to stable as well as chaotic oscillations. Experimental investigations of the large amplitude shape oscillation modes of centimeter-size single bubbles trapped in water at 1 G and under reduced hydrostatic pressure, have suggested the possibility of a low gravity experiment to study the direct coupling between these low frequency shape modes and the volume pulsation, sound-radiating mode. The second subject of interest has involved numerical modeling, using the boundary integral method, of the large amplitude shape oscillations of charged and uncharged drops in the presence of a static or time-varying electric field. Theoretically predicted non linearity in the resonant frequency of the fundamental quadrupole mode has been verified by the accompanying experimental studies. Additional phenomena such as hysteresis in the frequency response of ultrasoncially levitated droplets in the presence of a time varying electric field, and mode coupling in the oscillations of ultrasonically modulated droplets, have also been uncovered. One of the results of this ground-based research has been the identification and characterization of phenomena strictly associated with the influence of the gravitational field. This has also allowed us to identify the specific requirements for potential microgravity investigations yielding new information not obtainable on Earth.

Trinh, Eugene H.↗

Quantitative Phenomena Identification and Ranking Table (QPIRT) for Bayesian uncertainty quantification

Propagating parameter uncertainty for a nuclear reactor system code is a challenging problem due to often non-linear system response to the numerous parameters involved and lengthy computational times; issues that compound when a statistical sampling procedure is adopted, since the code must be run many times. The number of parameters sampled must therefore be limited to as few as possible that still accurately characterize the uncertainty in the system response. A Quantitative Phenomena Identification and Ranking Table (QPIRT) was developed to accomplish this goal. The QPIRT consists of two steps: a 'Top-Down' step focusing on identifying the dominant physical phenomena controlling the system response, and a 'Bottom-Up' step which focuses on determining the correlations from those key physical phenomena that significantly contribute to the response uncertainty. The Top-Down step evaluates phenomena using the governing equations of the system code at nominal parameter values, providing a 'fast' screening step. The Bottom-Up step then analyzes the correlations and models for the phenomena identified from the Top-Down step to find which parameters to sample. The QPIRT, through the Top-Down and Bottom-Up steps thus provides a systematic approach to determining the limited set of physically relevant parameters that influence the uncertainty of the system response. This strategy was demonstrated through an application to the RELAP5-based analysis of a PWR Total Loss of main Feedwater Flow (TLOFW) accident, also known as feed and bleed' scenario, . Ultimately, this work is the first component in a larger task of building a calibrated uncertainty propagation framework. The QPIRT is an essential piece because the uncertainty of those selected parameters will be calibrated to data from both Separate and Integral Effect Tests (SETs and IETs). Therefore the system response uncertainty will incorporate the knowledge gained from the database of past large IETs. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Application of Nonlinear Seismic Soil-Structure Interaction Analysis for Identification of Seismic Margins at Nuclear Power Plants

Seismic probabilistic risk assessment (SPRA) methods and approaches at nuclear power plants (NPP) were first developed in the 1970s and aspects of them have matured over time as they were applied and incrementally improved. SPRA provides information on risk and risk insights and allows for some accounting for uncertainty and variability. As a result, SPRA is now used as an important basis for risk-informed decision making for both new and operating NPPs in the US and in an increasing number of countries globally. SPRAs are intended to provide best estimates of the various combinations of structural and equipment failures that can lead to a seismic induced core damage event. However, in some instances the current SPRA approach contains large uncertainties, and potentially masks other important events (for instance, it was not the seismic motions that caused the Fukushima core melt events, but the tsunami ingress into the facility). INL has an advanced SPRA research and development (R&D) activity that will identify areas in the calculation process that contain significant uncertainties. One current area of focus is the use of nonlinear soil-structure interaction (NLSSI) analysis methods to accurately capture: 1) nonlinear soil behavior and 2) gapping and sliding between the NPP and soil. The goal of this study is to compare numerical NLSSI analysis results with recorded earthquake ground motions at Fukushima Daichii (Great Tohuku Earthquake) and evaluate the sources of nonlinearity contributing to the observed reduction in peak acceleration. Comparisons are made using recorded data in the free-field (soil column with no structural influence) and recorded data on the NPP basemat (in-structure response). Results presented in this study should identify areas of focus for future R&D activities with the goal of minimizing uncertainty in SPRA calculations. This is not a validation activity since there are too many sources of uncertainty that a numerical analysis would need to consider (variability in soil material properties, structural material properties, etc.). Rather the report will determine if the NLSSI calculations are following similar trends observed in the recorded data (i.e. reductions in maximum acceleration between the free-field and basemat) Numerical NLSSI results presented show maximum accelerations between the free field and basemat were reduced the EW and NS directions. The maximum acceleration in the UD direction increased slightly. The largest reduction in maximum accelerations between the modeled free-field and the NPP basemat resulted in nearly 50% reduction. The observation in reduction of numerical maximum accelerations in the EW and NS directions follows the observed trend in the recorded data. The maximum reductions observed in these NLSSI studies were due to soil nonlinearities, not gapping and sliding (although additional R&D is needed to develop an appropriate approach to model gapping and sliding). This exploratory study highlights the need for additional R&D on developing: (i) improved modeling of soil nonlinearities (soil constitutive models that appropriately capture cyclic soil behavior), (ii) improved modeling of gapping and sliding at the soil-structure interface (to appropriately capture the dissipation of energy at this interface), and (iii) experimental laboratory test data to calibrate the items (i) and (ii).

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Identification of Rate, Extent, and Mechanisms of Hot Metal Resulfurization with CaO-SiO{sub 2}-Na{sub 2}O Slag Systems

The resulfurization of hot metal has not been comprehensively studied in literature. This study presents an experimental and mathematical modeling study of resulfurization in thermodynamic and kinetic point of view. The rate, extent, and mechanisms of resulfurization were evaluated by analyzing concurrently the physical properties and sulfur-extracting ability of the slag. Experiments were conducted in a chamber furnace in an argon atmosphere, and the hot metal was sampled with pre-defined basis. The experiments were continued until the metal–slag system reached an apparent thermodynamic equilibrium. To obtain a quantitative measure on the effect of system properties on the rate and extent of resulfurization, the results of this study were combined with previous studies handling the sulfide capacities of Na{sub 2}O-SiO{sub 2} and CaO-SiO{sub 2}-Na{sub 2}O slag systems. The sulfide capacities of the slag and corresponding metal–slag sulfur partition ratios were mathematically modeled with data-driven techniques such as multiple linear and non-linear regression and artificial neural networks. Finally, with the help of these, to study the kinetics of resulfurization, a simple mechanistic reaction model was derived. The results suggest that resulfurization of hot metal follows 1st-order kinetics and that the rate and extent can be regulated through the control of the associated thermodynamic driving force and by modifying the physical properties of the slag. The rate-limiting factor was found to be determined by the morphology of the slag phase.

36 MATERIALS SCIENCE↗

Real-time Identification and Control of Satellite Signal Impairments Solution and Application of the Stratonovich Equation Part 1. Theoretical Development

As satellite communications systems become both more complex and reliant with respect to their operating environment, it has become imperative to be able to identify, during real-time operation, the onset of one or more impairments to the quality of overall communications system integrity. One of the most important aspects to monitor of a satellite link operating within the Earth's atmosphere is the signal fading due to the occurrence of rain and/or phase scintillations. This, of course, must be done in the presence of the associated measurement uncertainty or potentially faulty measurement equipment such as in the Advanced Communication Technology Satellite (ACTS) experiment. In the present work, an approach originally suggested in 1991, and apparently still considered iconoclastic, will be significantly developed and applied to the satellite communications link on which the deleterious composite signal fade is the result of one or many component fade mechanisms. Through the measurement (with the attendant uncertainty or 'error' in the measurement) of such a composite fading satellite signal, it is desired to extract the level of each of the individual fading mechanisms so they can be appropriately mitigated before they impact the overall performance of the communications network. Rather than employing simple-minded deterministic filtering to the real-time fading, the present approach is built around all the models and/or descriptions used to describe the individual fade components, including their dynamic evolution. The latter is usually given by a first-order Langevin equation. This circumstance allows the description of the associated temporal transition probability densities of each of the component processes. By using this description, along with the real-time measurements of the composite fade (along with the measurement errors), one can obtain statistical estimates of the levels of each of the component fading mechanisms as well as their predicted values into the future. This is all accomplished by the use of the well-known Stratonovich integro-differential equation that results from the model of the measured signal fade that is also tailored to adaptively adjust the values of the parameters used in the statistical models of the individual fade mechanisms. Three examples of increasing complexity are addressed and solved for the iterative determination of fade component levels from the measured composite signal fade in the presence of measurement error and, in the last case, with uncertainty in the model parameters.

Countermeasure↗

Identification and Quantification of Beta-Emitters in Waste Packages by Segmented Gamma-Scanning - 20114

A commonly applied method for the non-destructive characterization of the radioactive inventory of waste packages (e.g. 200-l drums) is segmented gamma-scanning. The gamma-emitting inventory is identified via their characteristic lines in the measured spectra. The corresponding net peak areas allow for quantification applying appropriate models. The identification and quantification of beta-emitters present in the waste package require further evaluation. While some of them emit gamma-rays, too, which can be used for identification and quantification, pure beta-emitters contribute to the gamma-spectra only by their Bremsstrahlung-spectra. {sup 32}P, {sup 90}Sr and {sup 90}Y are typical representatives of these isotopes. A method is developed for identification and quantification of these pure beta-emitters in radioactive waste packages. Origin of the procedure is the evaluation of the measured gamma-spectrum. After the identification of all lines present in the spectrum (characteristic gamma-lines, single- and double-escape-lines, X-ray-lines, annihilation peak etc.) this information is used together with calibration data of the detection system and further a-priori information like matrix composition, cross-section data from databases etc. to simulate the spectra without the Bremsstrahlung-components by an appropriate method. This may be Monte Carlo simulations or, like in our case, a computer code taking into account mathematical descriptions of the physical effects like photo effect, (multiple) Compton scattering, backscattering etc. and the attenuation effects of the waste matrix. In any case, the resulting spectrum must simulate the real spectrum without its Bremsstrahlung-component as good as possible. Subtracting the simulated from the measured spectrum then results in the Bremsstrahlung-components caused by the beta-emitters. Prior to evaluation, a set of Bremsstrahlung-spectra is calculated and stored for different beta-emitters in different matrices using the Bethe-Heitler equation. This data set can be extended for new conditions, if necessary, and is allocated in a data base for future application. Applying a non-linear optimization using the generalized reduced gradient method for comparing the calculated with the derived Bremsstrahlung-spectra results in the identification of the beta-emitter and a factor, which is proportional to its activity. The proportionality is determined by previous calibration measurements. The application of the method is demonstrated by measurements on a 200-l mock-up drum and a real waste package. (authors)

07 ISOTOPE AND RADIATION SOURCES↗

Constrained or unconstrained? Neural-network-based equation discovery from data

Throughout many fields, practitioners often rely on differential equations to model systems. Yet, for many applications, the theoretical derivation of such equations and/or the accurate resolution of their solutions may be intractable. Instead, recently developed methods, including those based on parameter estimation, operator subset selection, and neural networks, allow for the data-driven discovery of both ordinary and partial differential equations (PDEs), on a spectrum of interpretability. The success of these strategies is often contingent upon the correct identification of representative equations from noisy observations of state variables and, as importantly and intertwined with that, the mathematical strategies utilized to enforce those equations. Specifically, the latter has been commonly addressed via unconstrained optimization strategies. Representing the PDE as a neural network, we propose to discover the PDE (or the associated operator) by solving a constrained optimization problem and using an intermediate state representation similar to a physics-informed neural network (PINN). The objective function of this constrained optimization problem promotes matching the data, while the constraints require that the discovered PDE is satisfied at a number of spatial collocation points. We present a penalty method and a widely used trust-region barrier method to solve this constrained optimization problem, and we compare these methods on numerical examples. Our results on several example problems demonstrate that the latter constrained method outperforms the penalty method, particularly for higher noise levels or fewer collocation points. This work motivates further exploration into using sophisticated constrained optimization methods in scientific machine learning, as opposed to their commonly used, penalty-method or unconstrained counterparts. For both of these methods, we solve these discovered neural network PDEs with classical methods, such as finite difference methods, as opposed to PINNs-type methods relying on automatic differentiation. Here, we briefly highlight how simultaneously fitting the data while discovering the PDE improves the robustness to noise and other small, yet crucial, implementation details.

Data-driven discovery↗

Reorganization of Water at Aqueous Aluminum Chloride (AlCl 3 ) Interfaces: Vibrational Sum Frequency Generation and Molecular Dynamics Simulations

AlCl 3 hydration states and complexation are not well understood both in solutions and at the air–aqueous interface despite their potential significance in natural waters and their industrial and energy-related applications. Here, we investigated Al 3+ and Cl – ion behaviors in an AlCl 3 aqueous bulk solution and at the air–aqueous interface using interface-selective vibrational sum frequency generation (SFG), Raman and infrared spectroscopies, molecular dynamics (MD) simulation, as well as molecular-informed reduced modeling. Our reduced modeling reveals relatively long-range effects for Al 3+ as compared to monovalent ions such as Na + indicating that the interfacial depth of trivalent ions can be significantly larger than that of monovalent ions at the air–water interface. MD simulations reveal interfacial stratification and multiple layering of the ions. Compression of the Al 3+ and Cl – distributions with increasing concentrations from 0.5 to 2.5 m is also observed in the subsurface regions. Significant SSP- and PPP-polarized SFG OH spectral intensity increases are observed from 0.5 to 1.5 m and 0.5 to 2.5 m, respectively, indicative of interfacial depth increases and a change in average orientation above 1.5 m. Extensive evaluation of SFG spectra, Fresnel-corrected using several approaches, shows the same trends. The nonmonotonic trend points to a changing structure in surface and subsurface water orientation and hydrogen bonding environment generally consistent with the MD simulation of stratification and water orientation changes. Furthermore, solvent-shared ion pairing is implicated with MD simulation radial distribution analysis and consistent with infrared spectral identification of the hexaaqua aluminum ion in the solution phase. Spectral evidence of a strong Al 3+ hydration shell and the acidic behavior of the Al 3+ ions is obvious in the Raman and infrared spectra of the bulk solution. In conclusion, we show that the MD dipole potential is directly related to the MD second-order susceptibility of the interface, χ SFG–MD (2) , both of which correlate up to ∼35 Å with the spectral observations of increasing and then saturating intensities, suggesting that both ion stratification and interfacial depth determine the water orientations at an air–water interface of 1-3 electrolyte solutions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗