Model predictive control of nonlinear processes using transfer learning-based recurrent neural networks
Not Available
SEARCH · Search NASA
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.
Not Available
Not Available
Abstract not provided.
Abstract not provided.
Digital signal processing is the cornerstone of several modern-day technologies, yet in multiple applications it faces critical bottlenecks related to memory and speed constraints. Thanks to recent advances in metasurface design and fabrication, light-based analog computing has emerged as a viable option to partially replace or augment digital approaches. Several light-based analog computing functionalities have been demonstrated using patterned flat optical elements, with great opportunities for integration in compact nanophotonic systems. So far, however, the available operations have been restricted to the linear regime, limiting the impact of this technology to a compactification of Fourier optics systems. In this paper, we introduce nonlinear operations to the field of metasurface-based analog optical processing, demonstrating that nonlinear optical phenomena, combined with nonlocality in flat optics, can be leveraged to synthesize kernels beyond linear Fourier optics, paving the way to a broad range of new opportunities. As a practical demonstration, we report the experimental synthesis of a class of nonlinear operations that can be used to realize broadband, polarization-selective analog-domain edge detection.
Recurrence is one of the most common phenomena in natural and engineering systems. Process monitoring of dynamic transitions in nonlinear and nonstationary systems is more concerned with aperiodic recurrences and recurrence variations. However, little has been done to investigate the heterogeneous recurrence variations and link with the objectives of process monitoring and anomaly detection. Notably, nonlinear recurrence methodologies are based on homogeneous recurrences, which treat all recurrence states in the same way as black dots, and non-recurrence is white in recurrence plots. Heterogeneous recurrences are more concerned about the variations of recurrence states in terms of state properties (e.g., values and relative locations) and the evolving dynamics (e.g., sequential state transitions). This paper presents a novel approach of heterogeneous recurrence analysis that utilizes a new fractal representation to delineate heterogeneous recurrence states in multiple scales, including the recurrences of both single states and multi-state sequences. Further, we developed a new set of heterogeneous recurrence quantifiers that are extracted from fractal representation in the transformed space. To that end, we integrated multivariate statistical control charts with heterogeneous recurrence analysis to simultaneously monitor two or more related quantifiers. Experimental results on nonlinear stochastic processes show that the proposed approach not only captures heterogeneous recurrence patterns in the fractal representation but also effectively monitors the changes in the dynamics of a complex system.
Nonlinear dynamics induced by seismic sources and seismic waves are common in Earth. Observations range from seismic strong ground motion (the most damaging aspect of earthquakes), intense near-source effects, and distant nonlinear effects from the source that have important consequences. The distant effects include dynamic earthquake triggering-one of the most fascinating topics in seismology today-which may be elastically nonlinearly driven. Dynamic earthquake triggering is the phenomenon whereby seismic waves generated from one earthquake trigger slip events on a nearby or distant fault. Dynamic triggering may take place at distances thousands of kilometers from the triggering earthquake, and includes triggering of the entire spectrum of slip behaviors currently identified. These include triggered earthquakes and triggered slow, silent-slip during which little seismic energy is radiated. It appears that the elasticity of the fault gouge-the granular material located between the fault blocks-is key to the triggering phenomenon.
Here, a novel framework is proposed for integrating biomimetic-based advanced control and multi-agent optimization approaches for nonlinear chemical process applications. In particular, a Biologically-Inspired Optimal Control Strategy, denoted as BIO-CS, is combined with multi-agent optimization (MAO) techniques to provide optimal solutions for dynamic systems. In this combined framework, the BIO-CS algorithm employs gradient-based optimal control solvers for the intermediate problems associated with the leader-follower agents’ local interactions. Also, the MAO uses the capabilities of heuristic-based optimization techniques by sharing process information to obtain optimal operating setpoints for the controller considering an overall process objective. The applicability of the proposed method is demonstrated using a nonlinear, multivariable, process model of a fermentation system. Specifically, the optimal operating points are computed by the MAO implementation for setpoint tracking, trajectory tracking and plant-model mismatch scenarios for BIO-CS application. Results of the developed framework are compared to a gradient-based Sequential Quadratic Programming (SQP) technique and a classical proportional-integral (PI) controller in terms of optimization and control studies, respectively. As an additional contribution, BIO-CS is also cast as a model predictive controller (MPC) for the first time and compared to the agent-based BIO-CS approach in terms of computational time and tracking error. Closed-loop control results show up to 46% improvement in tracking performance during transient for the multi-agent BIO-CS when compared to BIO-CS as MPC for additional computational expense. The obtained results illustrate the capabilities of this novel integrated framework including BIO-CS as MPC to achieve desired nonlinear system performance for various scenarios.
We investigate magnetohydrodynamic turbulence driven by the magnetorotational instability (MRI) in Keplerian disks with a nonzero net azimuthal magnetic field using shearing box simulations. As distinct from previous studies, we analyze turbulence dynamics in Fourier (k-) space to understand its sustenance. The linear growth of the MRI with the azimuthal field has a transient character and is anisotropic in Fourier space, leading to anisotropy of nonlinear processes in Fourier space. As a result, the main nonlinear process appears to be a new type of angular redistribution of modes in Fourier space—the nonlinear transverse cascade—rather than the usual direct/inverse cascade. We demonstrate that the turbulence is sustained by the interplay of the linear transient growth of the MRI (which is the only energy supply for the turbulence) and the transverse cascade. These two processes operate at large length scales, comparable to the box size (disk scale height); the corresponding small wavenumber area, called the vital area in Fourier space, is crucial for the sustenance, while outside the vital area, direct cascade dominates. The interplay of the linear and nonlinear processes in Fourier space is generally too intertwined for a vivid schematization. Nevertheless, we reveal the basic subcycle of the sustenance that clearly shows the synergy of these processes in the self-organization of the magnetized flow system. This synergy is quite robust and persists for the considered different aspect ratios of the simulation boxes. The spectral characteristics of the dynamical processes in these boxes are qualitatively similar, indicating the universality of the sustenance mechanism of the MRI turbulence.
The focus of this initial work is to apply recent breakthroughs in sparse linear modeling and nonlinear signal processing to the problem of removing microphonic noise from nuclear spectrum measurements. The resolution of nuclear spectrum measurements is commonly degraded by microphonic noise caused by the vibration of supporting equipment (i.e. cryocoolers) as well as motion of the detector itself in mobile applications. Current signal processing techniques cannot remove microphonic noise without modifying the desired characteristics of the spikes induced by nuclear detection events. The LANL Engineering Institute has been working with NEN‐1 to remedy this problem. Initial work focused on developing active control systems to cancel the vibrations experienced by nuclear detectors. As work on progressed along these lines the team realized that a more robust, lower‐energy, lower‐cost solution to the microphonic noise problem could be realized using recent breakthroughs in nonlinear signal processing. Over the course of the last decade significant advances have been made in the field of sparse linear modeling using L1 norm minimization to find linear representations of sparse signals. The problem of removing microphonic noise from nuclear detection measurements appears to be well suited to the L1 minimization approach because these measurements tend to be inherently sparse in two different domains. The nuclear detections are sparse in a spike domain, and the microphonic noise is sparse in the Fourier basis. L1 minimization allows us to decompose the measurements into these two domains. The result is that we can potentially separate the spikes from the microphonic noise. Initial investigations into this technique look very promising. In this work microphonic noise is induced in an HE‐3 detector by a wrench impact. A pulser is used to generate uniform, synthetic, nuclear events and is fed into the detector front end electronics. Nearly all the microphonic noise is separated from the spike measurements. The histograms in Figure 2 illustrate that spectrum widening caused by noise is significantly reduced. The goal of future work to further develop this technique to account for variations in spike characteristics, as well as to use adaptive signal processing and modern Graphical Processing Unit (GPU) technology to speed up the signal decompositions.
We show theoretically that accelerating light beams can be used to manipulate nonlinear optical processes through spatiotemporal quasi-phase-matching, allowing for unprecedented temporal and spectral shaping of the generated light. As a proof of principle, we demonstrate exquisite control over the high-order harmonic frequency conversion process, showing efficient enhancement of an extremely broad range of harmonics emitted during a selected quarter-cycle of the driving laser pulse.
A comprehensive vibronic coupling model based on the time-dependent wavepacket approach is derived to simulate linear optical processes, such as one-photon absorbance and resonance Raman scattering, and nonlinear optical processes, such as two-photon absorbance and resonance hyper-Raman scattering. This approach is particularly well suited for combination with first-principles calculations. Expressions for the Franck-Condon terms, and non-Condon effects via the Herzberg-Teller coupling approach in the independent-mode displaced harmonic oscillator model are presented. The significance of each contribution to the different spectral types is discussed briefly.
Her,e we studied dynamical balances in magnetorotational instability (MRI) turbulence with a net vertical field in the shearing box model of disks. Analyzing the turbulence dynamics in Fourier ($\boldsymbol{k}$-)space, we identified three types of active modes that define the turbulence characteristics. These modes have lengths similar to the box size, i.e., lie in the small wavenumber region in Fourier space labeled "the vital area" and are (i) the channel mode, uniform in the disk plane with the smallest vertical wavenumber; (ii) the zonal flow mode, azimuthally and vertically uniform with the smallest radial wavenumber; and (iii) the rest (parasitic) modes. The rest modes comprise those harmonics in the vital area whose energies reach more than 50% of the maximum spectral energy. The rest modes individually are not so significant compared to the channel and zonal flow modes; however, the combined action of their multitude is dominant over these two modes. These three mode types are governed by the interplay of the linear and nonlinear processes, leading to their interdependent dynamics. The linear processes consist of disk flow nonmodality modified classical MRI with a net vertical field. The main nonlinear process is the transfer of modes over wavevector angles in Fourier space—the transverse cascade. The channel mode exhibits episodic bursts supplied by linear MRI growth, while the nonlinear processes mostly oppose this, draining the channel energy and redistributing it to the rest modes. As for the zonal flow, it does not have a linear source and is fed by nonlinear interactions of the rest modes.
The response of the cross-beam energy transfer instability (CBET) to laser bandwidth is investigated through a combination of theory and simulation. Existing linear theory is generalized to treat broadband lasers, demonstrating that CBET is most effectively suppressed when the bandwidth exceeds the ion-acoustic wave (IAW) frequency. It is shown that for such bandwidths, reverse (seed to pump) transfer becomes possible, which reduces the net energy transfer rapidly as bandwidth is increased. The CBET gain exponent in this regime scales with bandwidth (Δω) as Δω−3 for Gaussian or Lorentzian laser spectra with different scalings possible for other spectra. Comparison of our theory with linearized fluid and particle-in-cell simulations, performed with the laser-plasma simulation environment (LPSE) and vector particle in cell (VPIC) codes, respectively, finds that the model is accurate in the absence of nonlinear processes. However, linear analysis also finds that the IAW energy density scales as Δω−1, implying that nonlinear effects may be more difficult to control than the CBET scaling would suggest. Indeed, nonlinear effects are found to be present in VPIC simulations with high-intensity lasers, despite minimal apparent CBET. Nonlinear processes in the VPIC cases include particle trapping, the two-ion wave decay, and ion wave self-focusing. In some high intensity VPIC cases, these effects lead to net energy transfer from seed to pump and increases to backscatter stimulated Brillouin scattering reflectivities. Finally, for a given bandwidth, we show that improved control of nonlinear processes can be achieved via smoothing by spectral dispersion.
We propose a stochastic method for solving Schwinger-Dyson equations in large-N quantum field theories. Expectation values of single-trace operators are sampled by stationary probability distributions of the so-called nonlinear random processes. The set of all the histories of such processes corresponds to the set of all planar diagrams in the perturbative expansions of the expectation values of singlet operators. We illustrate the method on examples of the matrix-valued scalar field theory and the Weingarten model of random planar surfaces on the lattice. For theories with compact field variables, such as sigma models or non-Abelian lattice gauge theories, the method does not converge in the physically most interesting weak-coupling limit. In this case one can absorb the divergences into a self-consistent redefinition of expansion parameters. A stochastic solution of the self-consistency conditions can be implemented as a 'memory' of the random process, so that some parameters of the process are estimated from its previous history. We illustrate this idea on the two-dimensional O(N) sigma model. The extension to non-Abelian lattice gauge theories is discussed.
Abstract Similarity laws (SL) are essential for correlating the characteristics of plasmas at different dimensional scales, which have been validated for radio frequency (rf) discharges at low pressure but under limited conditions. In this work, we evaluate the effects of nonlinear collisions (e.g. stepwise ionization) on the similarity properties in capacitive rf discharges in argon across a wide range of pressure regimes via fully kinetic particle-in-cell simulations. The SL scalings of fundamental discharge parameters, e.g. the electron power absorption and electron energy probability function, are examined with and without nonlinear collisions, respectively. Without the nonlinear collisions, the similarity scalings are found to be rigorously valid. When the nonlinear collisions are considered, the similarity properties in rf discharges still exist approximately, which indicates that the violations caused by the nonlinear collisions are still minor in the studied cases. The reasons for the effectiveness of SL scalings with nonlinear collision processes are also discussed. The results from this study confirmed the validity of similarity transformations with more complex reaction kinetics, which may promote the applicability of the SL scalings for the design and fabrications of plasma devices.
Using one-dimensional numerical simulations, we study the elementary process of Alfvén wave reflection in a uniform medium, including nonlinear effects. In the linear regime, Alfvén wave reflection is triggered only by the inhomogeneity of the medium, whereas in the nonlinear regime, it can occur via nonlinear wave–wave interactions. Such nonlinear reflection (backscattering) is typified by decay instability. In most studies of decay instabilities, the initial condition has been a circularly polarized Alfvén wave. In this study we consider a linearly polarized Alfvén wave, which drives density fluctuations by its magnetic pressure force. For generality, we also assume a broadband wave with a red-noise spectrum. In the data analysis, we decompose the fluctuations into characteristic variables using local eigenvectors, thus revealing the behaviors of the individual modes. Different from the circular-polarization case, we find that the wave steepening produces a new energy channel from the parent Alfvén wave to the backscattered one. Such nonlinear reflection explains the observed increasing energy ratio of the sunward to the anti-sunward Alfvénic fluctuations in the solar wind with distance against the dynamical alignment effect.
The nonlinear optical response of a system of molecules often contains contributions whereby the products of lower-order processes in two separate molecules give signals that appear on top of a genuine direct higher-order process with a single molecule. These many-body contributions are known as cascading and complicate the interpretation of multidimensional stimulated Raman and other nonlinear signals. In a quantum electrodynamic treatment, these cascading processes arise from second-order expansion in the molecular coupling to vacuum modes of the radiation field, i.e., single-photon exchange between molecules, which also gives rise to other collective effects. Furthermore, we predict the relative phase of the direct and cascading nonlinear signals and its dependence on the microscopic dynamics as well as the sample geometry. This phase may be used to identify experimental conditions for distinguishing the direct and cascading signals by their phase. Higher-order cascading processes involving the exchange of several photons between more than two molecules are discussed.