Search NASA⌕ Search

SEARCH · Search NASA

Results for “variational optimization”

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 1,783 records · Page 99

SU-F-T-188: A Robust Treatment Planning Technique for Proton Pencil Beam Scanning Cranial Spinal Irradiation

Purpose: To propose a proton pencil beam scanning (PBS) cranial spinal irradiation (CSI) treatment planning technique robust against patient roll, isocenter offset and proton range uncertainty. Method: Proton PBS plans were created (Eclipse V11) for three previously treated CSI patients to 36 Gy (1.8 Gy/fractions). The target volume was separated into three regions: brain, upper spine and lower spine. One posterior-anterior (PA) beam was used for each spine region, and two posterior-oblique beams (15° apart from PA direction, denoted as 2PO-15) for the brain region. For comparison, another plan using one PA beam for the brain target (denoted as 1PA) was created. Using the same optimization objectives, 98% CTV was optimized to receive the prescription dose. To evaluate plan robustness against patient roll, the gantry angle was increased by 3° and dose was recalculated without changing the proton spot weights. On the re-calculated plan, doses were then calculated using 12 scenarios that are combinations of isocenter shift (±3mm in X, Y, and Z directions) and proton range variation (±3.5%). The worst-case-scenario (WCS) brain CTV dosimetric metrics were compared to the nominal plan. Results: For both beam arrangements, the brain field(s) and upper-spine field overlap in the T2–T5 region depending on patient anatomy. The maximum monitor unit per spot were 48.7%, 47.2%, and 40.0% higher for 1PA plans than 2PO-15 plans for the three patients. The 2PO-15 plans have better dose conformity. At the same level of CTV coverage, the 2PO-15 plans have lower maximum dose and higher minimum dose to the CTV. The 2PO-15 plans also showed lower WCS maximum dose to CTV, while the WCS minimum dose to CTV were comparable between the two techniques. Conclusion: Our method of using two posterior-oblique beams for brain target provides improved dose conformity and homogeneity, and plan robustness including patient roll.

60 APPLIED LIFE SCIENCES↗

Massively parallel and universal approximation of nonlinear functions using diffractive processors

Nonlinear computation is essential for a wide range of information processing tasks, yet implementing nonlinear functions using optical systems remains a challenge due to the weak and power-intensive nature of optical nonlinearities. Overcoming this limitation without relying on nonlinear optical materials could unlock unprecedented opportunities for ultrafast and parallel optical computing systems. Here, we demonstrate that large-scale nonlinear computation can be performed using linear optics through optimized diffractive processors composed of passive phase-only surfaces. In this framework, the input variables of nonlinear functions are encoded into the phase of an optical wavefront—e.g., via a spatial light modulator (SLM)—and transformed by an optimized diffractive structure with spatially varying point-spread functions to yield output intensities that approximate a large set of unique nonlinear functions–all in parallel. We provide proof establishing that this architecture serves as a universal function approximator for an arbitrary set of bandlimited nonlinear functions, also covering wavelength-multiplexed nonlinear functions as well as multi-variate and complex-valued functions that are all-optically cascadable. Our analysis also indicates the successful approximation of typical nonlinear activation functions commonly used in neural networks, including the sigmoid, tanh, ReLU (rectified linear unit), and softplus. We numerically demonstrate the parallel computation of one million distinct nonlinear functions, accurately executed at wavelength-scale spatial density at the output of a diffractive optical processor. Furthermore, we experimentally validated this framework using in situ optical learning and approximated 35 unique nonlinear functions in a single shot using a compact setup consisting of an SLM and an image sensor. These results establish diffractive optical processors as a scalable platform for massively parallel universal nonlinear function approximation, paving the way for new capabilities in analog optical computing based on linear materials.

Rahman, Md Sadman Sakib [University of California,↗

Multichannel Fiber-Optic Silicon Fabry–Pérot Interferometric Bolometer System for Plasma Radiation Measurements

A single-channel fiber-optic bolometer system based on a high-finesse silicon Fabry–Pérot interferometer (FPI) was previously reported, intended to measure plasma radiation from the magnetically confined fusion chamber. Recently, we developed a multichannel fiber-optic bolometer system with five bolometers multiplexed using a coarse wavelength division multiplexer (CWDM) and interrogated with a white-light system involving a superluminescent light-emission diode source and a high-speed spectrometer. One of the bolometers was used as the reference bolometer to compensate for the ambient temperature variations, and the other four bolometers were used for radiation measurement. The bolometers have a simple structure with a silicon pillar at the end of the single-mode fiber and a gold disk on the other side of the silicon pillar. They are also easy to fabricate without stringent requirements on the optical alignment. Analysis of the system optimization was performed to improve the noise performance and to mitigate the vibration effect that may present in the practical application. The system had a significantly enhanced measurement range compared to the previous high-finesse FPI bolometer system for measuring radiation. Test results performed in air using a 405 nm laser as the radiation source showed that the temperature resolution and the noise-equivalent power density of the sensing bolometers connected to each channel of the CWDM were, respectively, ~0.4 mK and ~0.1 W/m2, with a time constant of ~220 ms, which is comparable to the previous more complicated fiber-optic bolometer systems based on high-finesse FPIs that were interrogated using wavelength-scanning lasers.

Uddin, Nezam↗

MODULATION OF LOW ENERGY BEAM TO GENERATE PREDEFINED BUNCH TRAINS FOR THE NSLS-II TOP-OFF INJECTION

The NSLS II linac will produce a bunch train, 80-150 bunches long with 2 ns bunch spacing. Having the ability to tailor the bunch train can lead to the smaller bunch to bunch charge variation in the storage ring. A stripline is planned to integrate into the linac baseline to achieve this tailoring. The stripline must have a fast field rise and fall time to tailor each bunch. The beam dynamics is minimally affected by including the extra space for the stripline. This paper discusses the linac beam dynamics with stripline, and the optimal design of the stripline. A stripline is to be integrated in the linac to match the storage ring uniform bunch charge requirement, which simplifies the gun pulser electronics and looses the edge uniform requirement. It is located at low energy to lower the stripline power supply requirement and limit the dumped electron radiation. By turning off the stripline, the beam dynamics through linac is comparable with the baseline design. More advanced ideas can be explored. If a DC corrector along with the stripline is used, the core bunch trains gets kick from the stripline while the head and the tail of bunch train just gets a DC kick. The stripline power supply waveform is a single flat top waveform with fast rise and drop and the pulse length is {approx}200 ns long or 100 bunches, which may be easier from the power supply view point. We are also considering the bunch by bunch charge manipulation to match the storage ring uniform bunch charge distribution requirement. By modulating the flat top waveform at 250 MHz with adjustable amplitude, each the bunch center is either at 45 degree or 135 degree. Only the head or tail of the bunch is trimmed out. Although each bunch center deviation from idea center is very different at low energy, it is gradually minimized with beam energy increase.

43 PARTICLE ACCELERATORS↗

Massively Parallel Latent Semantic Analyzes using a Graphics Processing Unit

Latent Semantic Indexing (LSA) aims to reduce the dimensions of large Term-Document datasets using Singular Value Decomposition. However, with the ever expanding size of data sets, current implementations are not fast enough to quickly and easily compute the results on a standard PC. The Graphics Processing Unit (GPU) can solve some highly parallel problems much faster than the traditional sequential processor (CPU). Thus, a deployable system using a GPU to speedup large-scale LSA processes would be a much more effective choice (in terms of cost/performance ratio) than using a computer cluster. Due to the GPU s application-specific architecture, harnessing the GPU s computational prowess for LSA is a great challenge. We present a parallel LSA implementation on the GPU, using NVIDIA Compute Unified Device Architecture and Compute Unified Basic Linear Algebra Subprograms. The performance of this implementation is compared to traditional LSA implementation on CPU using an optimized Basic Linear Algebra Subprograms library. After implementation, we discovered that the GPU version of the algorithm was twice as fast for large matrices (1000x1000 and above) that had dimensions not divisible by 16. For large matrices that did have dimensions divisible by 16, the GPU algorithm ran five to six times faster than the CPU version. The large variation is due to architectural benefits the GPU has for matrices divisible by 16. It should be noted that the overall speeds for the CPU version did not vary from relative normal when the matrix dimensions were divisible by 16. Further research is needed in order to produce a fully implementable version of LSA. With that in mind, the research we presented shows that the GPU is a viable option for increasing the speed of LSA, in terms of cost/performance ratio.

99 GENERAL AND MISCELLANEOUS↗

Slater-Pauling behavior within quaternary intermetallic borides of the Ti{sub 3}Co{sub 5}B{sub 2} structure-type

First-principles, density-functional studies of several intermetallic borides of the general type M{sub 2}M'Ru{sub 5-n}Rh{sub n}B{sub 2} (n=0-5; M=Sc, Ti, Nb; M'=Fe, Co) show that the variation in saturation magnetic moment with valence-electron count follows a Slater-Pauling curve, with a maximum moment occurring typically at 66 valence electrons. The magnetic moments in these compounds occur primarily from the 3d electrons of the magnetically active M' sites, with some contribution from the Ru/Rh sites via magnetic polarization. Electronic DOS curves reveal that a rigid-band approach is a reasonable approximation for the estimation of saturation moments and the analysis of orbital interactions in this family of complex borides. COHP analyses of the M'-M' orbital interactions indicate optimized interactions in the minority spin states for Co-containing phases, but strong bonding interactions remaining in Fe-containing phases. - Graphical abstract: Theoretically determined (spin-polarized LMTO-GGA) local magnetic moments as a function of the chemical valence Z for various intermetallic borides.

36 MATERIALS SCIENCE↗

Local Setup Errors in Image-Guided Radiotherapy for Head and Neck Cancer Patients Immobilized With a Custom-Made Device

Purpose: To evaluate the local positioning uncertainties during fractionated radiotherapy of head-and-neck cancer patients immobilized using a custom-made fixation device and discuss the effect of possible patient correction strategies for these uncertainties. Methods and Materials: A total of 45 head-and-neck patients underwent regular control computed tomography scanning using an in-room computed tomography scanner. The local and global positioning variations of all patients were evaluated by applying a rigid registration algorithm. One bounding box around the complete target volume and nine local registration boxes containing relevant anatomic structures were introduced. The resulting uncertainties for a stereotactic setup and the deformations referenced to one anatomic local registration box were determined. Local deformations of the patients immobilized using our custom-made device were compared with previously published results. Several patient positioning correction strategies were simulated, and the residual local uncertainties were calculated. Results: The patient anatomy in the stereotactic setup showed local systematic positioning deviations of 1-4 mm. The deformations referenced to a particular anatomic local registration box were similar to the reported deformations assessed from patients immobilized with commercially available Aquaplast masks. A global correction, including the rotational error compensation, decreased the remaining local translational errors. Depending on the chosen patient positioning strategy, the remaining local uncertainties varied considerably. Conclusions: Local deformations in head-and-neck patients occur even if an elaborate, custom-made patient fixation method is used. A rotational error correction decreased the required margins considerably. None of the considered correction strategies achieved perfect alignment. Therefore, weighting of anatomic subregions to obtain the optimal correction vector should be investigated in the future.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

MODELING THE SUNSPOT NUMBER DISTRIBUTION WITH A FOKKER-PLANCK EQUATION

Sunspot numbers exhibit large short-timescale (daily-monthly) variation in addition to longer-timescale variation due to solar cycles. A formal statistical framework is presented for estimating and forecasting randomness in sunspot numbers on top of deterministic (including chaotic) models for solar cycles. The Fokker-Planck approach formulated assumes a specified long-term or secular variation in sunspot number over an underlying solar cycle via a driver function. The model then describes the observed randomness in sunspot number on top of this driver function. We consider a simple harmonic choice for the driver function, but the approach is general and can easily be extended to include other drivers which account for underlying physical processes and/or empirical features of the sunspot numbers. The framework is consistent during both solar maximum and minimum, and requires no parameter restrictions to ensure non-negative sunspot numbers. Model parameters are estimated using statistically optimal techniques. The model agrees both qualitatively and quantitatively with monthly sunspot data even with the simplistic representation of the periodic solar cycle. This framework should be particularly useful for solar cycle forecasters and is complementary to existing modeling techniques. An analytic approximation for the Fokker-Planck equation is presented, which is analogous to the Euler approximation, which allows for efficient maximum likelihood estimation of large data sets and/or when using difficult to evaluate driver functions.

79 ASTRONOMY AND ASTROPHYSICS↗

MEASURING THE REDSHIFT DEPENDENCE OF THE COSMIC MICROWAVE BACKGROUND MONOPOLE TEMPERATURE WITH PLANCK DATA

We study the capability of Planck data to constrain deviations of the cosmic microwave background (CMB) blackbody temperature from adiabatic evolution using the thermal Sunyaev-Zeldovich anisotropy induced by clusters of galaxies. We consider two types of data sets depending on how the cosmological signal is removed: using a CMB template or using the 217 GHz map. We apply two different statistical estimators, based on the ratio of temperature anisotropies at two different frequencies and on a fit to the spectral variation of the cluster signal with frequency. The ratio method is biased if CMB residuals with amplitude {approx}1 {mu}K or larger are present in the data, while residuals are not so critical for the fit method. To test for systematics, we construct a template from clusters drawn from a hydro-simulation included in the pre-launch Planck Sky Model. We demonstrate that, using a proprietary catalog of X-ray-selected clusters with measured redshifts, electron densities, and X-ray temperatures, we can constrain deviations of adiabatic evolution, measured by the parameter {alpha} in the redshift scaling T(z) = T{sub 0}(1 + z){sup 1-{alpha}}, with an accuracy of {sigma}{sub {alpha}} = 0.011 in the most optimal case and with {sigma}{sub {alpha}} = 0.018 for a less optimal case. These results represent a factor of 2-3 improvement over similar measurements carried out using quasar spectral lines and a factor 6-20 with respect to earlier results using smaller cluster samples.

79 ASTRONOMY AND ASTROPHYSICS↗

Data-driven wind turbine wake modeling via probabilistic machine learning

Wind farm design primarily depends on the variability of the wind turbine wake flows to the atmospheric wind conditions and the interaction between wakes. Physics-based models that capture the wake flow field with high-fidelity are computationally very expensive to perform layout optimization of wind farms, and, thus, data-driven reduced-order models can represent an efficient alternative for simulating wind farms. In this work, we use real-world light detection and ranging (LiDAR) measurements of wind-turbine wakes to construct predictive surrogate models using machine learning. Specifically, we first demonstrate the use of deep autoencoders to find a low-dimensional latent space that gives a computationally tractable approximation of the wake LiDAR measurements. Then, we learn the mapping between the parameter space and the (latent space) wake flow fields using a deep neural network. Additionally, we also demonstrate the use of a probabilistic machine learning technique, namely, Gaussian process modeling, to learn the parameter-space-latent-space mapping in addition to the epistemic and aleatoric uncertainty in the data. Finally, to cope with training large datasets, we demonstrate the use of variational Gaussian process models that provide a tractable alternative to the conventional Gaussian process models for large datasets. Furthermore, we introduce the use of active learning to adaptively build and improve a conventional Gaussian process model predictive capability. Overall, we find that our approach provides accurate approximations of the wind-turbine wake flow field that can be queried at an orders-of-magnitude cheaper cost than those generated with high-fidelity physics-based simulations.

Deep neural networks↗

Climate Resilience and the Design of Smart Buildings

How will buildings of the future achieve increased climate resilience while operating at low energy utilization indices? Could smart buildings be networked and designed to foresee energy consumption variations and change their performance in real-time? This chapter will attempt to answer these questions by a) reviewing existing literature on expected climate change impacts to cities and buildings, b) reviewing of existing climate change action plans and initiatives for cities, c) discussing current efforts in designing smart, climate resilient buildings and d) discussing building design strategies for optimizing energy use and addressing future climate change in a smart city network. The objective of this chapter is to provide readers with state of the art knowledge and actionable approaches to addressing climate change, in the context of future city developments, and its impact on building energy consumption and resilience.

smart buildings, climate change, greenhouse gas em↗

Antimony stable isotope fractionation during adsorption onto birnessite: A molecular perspective from X-ray absorption spectroscopy and density functional theory

Sorption of antimony (Sb) onto birnessite significantly influences the fate of Sb in oceanic and terrestrial environments and fractionates Sb isotopes. Nevertheless, little is known about Sb isotopic fractionation during its adsorption on birnessite. Here, in this study, we show the value of Δ 123 Sb adsorbed-aqueous increases from −0.398 to −0.332 ‰ in 1 h and then decreases and stabilizes at −0.384 ‰ in 72 h. The enrichment of the light Sb isotope is predominantly due to the distortion of the octahedral symmetry. X-ray absorption spectroscopy results indicate Sb first forms a double-corner-sharing complex on birnessite and then transforms to a double-edge-sharing complex during adsorption. The optimized bond distances for double-corner-sharing (3.37 Å) and double-edge-sharing (2.90 Å) complexes calculated using density functional theory (DFT) fits well with the structure (3.41 and 3.00 Å) revealed by X-ray absorption spectroscopy, respectively. The fractionation derived from reduced partition function ratios calculated using DFT aligns well with the experimental results. Therefore, the variation in Sb isotopic fractionation during adsorption is attributed to the evolving structure of Sb complexes on birnessite. Our results demonstrate the isotopic fractionation of Sb during adsorption on birnessite and provide a molecular-scale understanding of Sb behavior, contributing to the correct reconstruction of the Sb isotope composition of ancient seawater using ferromanganese crusts and nodules, and efforts to trace Sb migration in epigenetic mining environments.

Adsorption↗

Electrochemical dynamics of imidazolium ionic liquids at graphene electrodes for energy storage applications

Electric double layer capacitors (EDLCs) are prominent energy storage systems that constitute the foundation of more reliable and sustainable energy infrastructures. Modern EDLCs often incorporate ionic liquids (ILs) as a key component in their electrolytes, leveraging the high electrochemical stability of ILs to enhance device performance. The performance and functionality of these capacitors also hinge on the interfacial behavior of ILs at electrode surfaces, which remain insufficiently understood. Here, we performed synchrotron infrared nanospectroscopy (SINS) in combination with density functional theory (DFT) calculations to investigate the electric double layers (EDLs) of three imidazolium-based ILs in a custom-designed graphene liquid cell. This approach revealed new insights into the dynamics of IL EDLs and the underpinning factors originating from the IL structures. Variations in anion size and structure were found to tune the ILs’ ability to form EDLs with compact and closely-correlated ion arrangements, which are critical for enhancing their capacitive performance. These findings highlight the intricate interactions between ions governed by their structures and charge behaviors, which underscores the opportunities for targeted design of IL-based electrolytes to optimize EDLC functionality.

Electric double layers↗

Elucidating the differences in oxidation of high-performance α- and β- diisobutylene biofuels via Synchrotron photoionization mass spectrometry

Abstract Biofuels are a promising ecologically viable and renewable alternative to petroleum fuels, with the potential to reduce net greenhouse gas emissions. However, biomass sourced fuels are often produced as blends of hydrocarbons and their oxygenates. Such blending complicates the implementation of these fuels in combustion applications. Variations in a biofuel’s composition will dictate combustion properties such as auto ignition temperature, reaction delay time, and reaction pathways. A handful of novel drop-in replacement biofuels for conventional transportation fuels have recently been down selected from a list of over 10,000 potential candidates as part of the U.S. Department of Energy’s (DOE) Co-Optimization of Fuels and Engines (Co-Optima) initiative. Diisobutylene (DIB) is one such high-performing hydrocarbon which can readily be produced from the dehydration and dimerization of isobutanol, produced from the fermentation of biomass-derived sugars. The two most common isomers realized, from this process, are 2,4,4-trimethyl-1-pentene (α-DIB) and 2,4,4-trimethyl-2-pentene (β-DIB). Due to a difference in olefinic bond location, the α- and β- isomer exhibit dramatically different ignition temperatures at constant pressure and equivalence ratio. This may be attributed to different fragmentation pathways enabled by allylic versus vinylic carbons. For optimal implementation of these biofuel candidates, explicit identification of the intermediates formed during the combustion of each of the isomers is needed. To investigate the combustion pathways of these molecules, tunable vacuum ultraviolet (VUV) light (in the range 8.1–11.0 eV) available at the Lawrence Berkeley National Laboratory’s Advanced Light Source (ALS) has been used in conjunction with a jet stirred reactor (JSR) and time-of-flight mass spectrometry to probe intermediates formed. Relative intensity curves for intermediate mass fragments produced during this process were obtained. Several important unique intermediates were identified at the lowest observable combustion temperature with static pressure of 93,325 Pa and for 1.5 s residence time. As this relatively short residence time is just after ignition, this study is targeted at the fuels’ ignition events. Ignition characteristics for both isomers were found to be strongly dependent on the kinetics of C 4 and C 7 fragment production and decomposition, with the tert-butyl radical as a key intermediate species. However, the ignition of α-DIB exhibited larger concentrations of C 4 compounds over C 7 , while the reverse was true for β-DIB. These identified species will allow for enhanced engineering modeling of fuel blending and engine design.

09 BIOMASS FUELS↗

Isolating and quantifying cross-beam energy transfer in direct-drive implosions on OMEGA and the National Ignition Facility

The angularly resolved mass ablation rates and ablation-front trajectories for Si-coated CH targets were measured in direct-drive inertial confinement fusion experiments to quantify cross-beam energy transfer (CBET) while constraining the hydrodynamic coupling. A polar-direct-drive laser configuration, where the equatorial laser beams were dropped and the polar beams were repointed from a symmetric direct-drive configuration, was used to limit CBET at the pole while allowing it to persist at the equator. The combination of low- and high-CBET conditions observed in the same implosion allowed for the effects of CBET on the ablation rate and ablation pressure to be determined. Hydrodynamic simulations performed without CBET agreed with the measured ablation rate and ablation-front trajectory at the pole of the target, confirming that the CBET effects on the pole are small. The simulated mass ablation rates and ablation-front trajectories were in excellent agreement with the measurements at all angles when a CBET model based on Randall's equations [C. J. Randall et al., Phys. Fluids 24, 1474 (1981)] was included into the simulations with a multiplier on the CBET gain factor. These measurements were performed on OMEGA and at the National Ignition Facility to access a wide range of plasma conditions, laser intensities, and laser beam geometries. The presence of the CBET gain multiplier required to match the data in all of the configurations tested suggests that additional physics effects, such as intensity variations caused by diffraction, polarization effects, or shortcomings of extending the 1-D Randall model to 3-D, should be explored to explain the differences in observed and predicted drive.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Strong-field dipole resonance: Limiting analytical cases

We investigate population dynamics in N-level systems driven beyond the linear regime by a strong external field, which couples to the system through an operator with nonzero diagonal elements. As concrete example we consider the case of dipolar molecular systems. We identify limiting cases of the Hamiltonian leading to wave functions that can be written in terms of ordinary exponentials, and focus on the limits of slowly and rapidly varying fields of arbitrary strength. For rapidly varying fields we prove for arbitrary N that the population dynamics is independent of the sign of the projection of the field onto the dipole coupling. In the opposite limit of slowly varying fields the population of the target level is optimized by a dipole resonance condition. As a result population transfer is maximized for one sign of the field and suppressed for the other one, so that a switch based on flopping the field polarization can be devised. For significant sign dependence the resonance linewidth with respect to the field strength is small. In the intermediate regime of moderate field variation, the integral of lowest order in the coupling can be rewritten as a sum of terms resembling the two limiting cases, plus correction terms for N>2, so that a less pronounced sign-dependence still exists.

74 ATOMIC AND MOLECULAR PHYSICS↗

Entangling Quantum Generative Adversarial Networks

Generative adversarial networks (GANs) are one of the most widely adopted machine learning methods for data generation. In this work, we propose a new type of architecture for quantum generative adversarial networks (an entangling quantum GAN, EQ-GAN) that overcomes limitations of previously proposed quantum GANs. Leveraging the entangling power of quantum circuits, the EQ-GAN converges to the Nash equilibrium by performing entangling operations between both the generator output and true quantum data. In the first multiqubit experimental demonstration of a fully quantum GAN with a provably optimal Nash equilibrium, we use the EQ-GAN on a Google Sycamore superconducting quantum processor to mitigate uncharacterized errors, and we numerically confirm successful error mitigation with simulations up to 18 qubits. Finally, we present an application of the EQ-GAN to prepare an approximate quantum random access memory and for the training of quantum neural networks via variational datasets.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Adaptive PID Gain Scheduling Control for Hydropower Turbine Using Neural CDE and Stochastic Distribution Shaping

This paper introduces a gain-scheduling PID controller design strategy for hydroturbine frequency control mode. This scheme first uses real data to learn the nonlinear dynamics of the hydroturbine using neural controlled differential equations and then perturbs the obtained nonlinear system at different equilibrium points, based on which a static output feedback adaptive dynamic programming algorithm is then used to optimize the PID gains for each equilibrium point. Moreover, a continuous-time version of stochastic distribution control is proposed to further fine-tune the optimized PID gains. Finally, the controller is obtained by implementing linear interpolation between the optimized PID control gains. The simulation results show that the proposed gain-scheduling PID controller can control a larger range of operation points compared with the given fixed PID controller and the baseline method. Compared with the given fixed PID controller, the proposed gain-scheduling PID controller can regulate hydroturbine frequency against disturbances induced by power-load variation with over 50% less overshoot for some operation points.

13 HYDRO ENERGY↗