Search NASA⌕ Search

SEARCH · Search NASA

Results for “hard constraint”

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 37 records · Page 2

A System for Automatically Generating Scheduling Heuristics

The goal of this research is to improve the performance of automated schedulers by designing and implementing an algorithm by automatically generating heuristics by selecting a schedule. The particular application selected by applying this method solves the problem of scheduling telescope observations, and is called the Associate Principal Astronomer. The input to the APA scheduler is a set of observation requests submitted by one or more astronomers. Each observation request specifies an observation program as well as scheduling constraints and preferences associated with the program. The scheduler employs greedy heuristic search to synthesize a schedule that satisfies all hard constraints of the domain and achieves a good score with respect to soft constraints expressed as an objective function established by an astronomer-user.

Morris, Robert↗

Real-time optimal guidance for orbital maneuvering.

A new formulation for soft-constraint trajectory optimization is presented as a real-time optimal feedback guidance method for multiburn orbital maneuvers. Control is always chosen to minimize burn time plus a quadratic penalty for end condition errors, weighted so that early in the mission (when controllability is greatest) terminal errors are held negligible. Eventually, as controllability diminishes, the method partially relaxes but effectively still compensates perturbations in whatever subspace remains controllable. Although the soft-constraint concept is well-known in optimal control, the present formulation is novel in addressing the loss of controllability inherent in multiple burn orbital maneuvers. Moreover the necessary conditions usually obtained from a Bolza formulation are modified in this case so that the fully hard constraint formulation is a numerically well behaved subcase. As a result convergence properties have been greatly improved.

Cohen, A. O.↗

Safe Reinforcement Learning-Based Transient Stability Control for Islanded Microgrids With Topology Reconfiguration

This paper proposes a safe reinforcement learning (RL)-based transient stability emergency control (TSEC) method for islanded microgrids. RL requires extensive interaction with the environment to learn control strategies, hence, a data-driven approach is used as a substitute for time-consuming time-domain simulation calculations. Deep sigma point processes (DSPP), which is a Gaussian process model, is utilized to predict the normal distribution of transient stability of microgrids and to construct a transient stability chance constraint. Reward-constrained policy optimization (RCPO) can simultaneously achieve objective prediction, policy learning, and constraint cost coefficient update across multiple timescales. RCPO interacts with the DSPP-based microgrid environment through a multi-process parallel manner, greatly increasing the training speed. Case studies on a real islanded microgrid demonstrate that the proposed method can efficiently and quickly obtain the optimal emergency control strategy while adhering to all hard constraints.

14 SOLAR ENERGY↗

SU-F-BRD-10: Improving Plan Delivery Efficiency of Intensity Modulated Proton Plans with Prioritized Optimization

Purpose: To integrate treatment delivery time into plan optimization in spot scanning intensity modulated proton therapy. Utilizing a dedicated research treatment planning system we present an optimization approach to explore the trade-off between the correlated parameters treatment time and plan quality on an astrocytoma patient case. Methods: The planning system is based on prioritized optimization, a stepwise approach of implementing clinical goals. After each optimization step, dosimetric achievements are turned into hard constraints to maintain the achieved plan quality. Prior achievements can be violated by a so-called slip-factor which allows to study possible trade-offs of conflicting goals. Plan quality is obtained in the first two steps, while the third step optimizes delivery efficiency by working on the spot weight distribution via four alternative Methods: elimination of low weighted spots (1), elimination of spots hardly contributing to PTV dose, followed by reoptimization of the resulting smaller optimization problem (2), reduction of spot weights variance within each energy layer (3), and reduction of the overall spot weight sum (4). Treatment times were calculated assuming either constant or variable beam current depending on the lowest spot weight. Results: Delivery efficiency can be improved remarkably without influencing the plan quality. Absolute time savings depend on the utilized method and facility properties. By varying slip-factor and spot reduction limits, a border of worsening quality is detectable for all methods.Deleting low weighted spots by 10% results in a noticeable decrease in minimum target dose. Further reduction results in more heterogeneous dose distributions and insufficient coverage. Option 2 showed constant plan quality for spot reductions of more than 10%. Conclusion: Including treatment time optimization as a final step into prioritized optimization allows for more efficient treatment plans by redistributing the spot patterns without compromising plan quality, which has to remain the main intent of radiotherapy treatment. The authors report grants from German Research Foundation (DFG) during the conduct of the study.

60 APPLIED LIFE SCIENCES↗

Effective data sampling strategies and boundary condition constraints of physics-informed neural networks for identifying material properties in solid mechanics

Material identification is critical for understanding the relationship between mechanical properties and the associated mechanical functions. However, material identification is a challenging task, especially when the characteristic of the material is highly nonlinear in nature, as is common in biological tissue. In this work, we identify unknown material properties in continuum solid mechanics via physics-informed neural networks (PINNs). To improve the accuracy and efficiency of PINNs, we develop efficient strategies to nonuniformly sample observational data. We also investigate different approaches to enforce Dirichlet-type boundary conditions (BCs) as soft or hard constraints. Finally, we apply the proposed methods to a diverse set of time-dependent and time-independent solid mechanic examples that span linear elastic and hyperelastic material space. The estimated material parameters achieve relative errors of less than 1%. As such, this work is relevant to diverse applications, including optimizing structural integrity and developing novel materials.

97 MATHEMATICS AND COMPUTING↗

TH-CD-209-04: Fuzzy Robust Optimization in Intensity-Modulated Proton Therapy Planning to Account for Range and Patient Setup Uncertainties

Purpose: We propose to apply a robust optimization model based on fuzzy-logic constraints in the intensity-modulated proton therapy (IMPT) planning subject to range and patient setup uncertainties. The purpose is to ensure the plan robustness under uncertainty and obtain the best trade-off between tumor dose coverage and organ-at-risk(OAR) sparing. Methods: Two IMPT plans were generated for 3 head-and-neck cancer patients: one used the planning target volume(PTV) method; the other used the fuzzy robust optimization method. In the latter method, nine dose distributions were computed - the nominal one and one each for ±3mm setup uncertainties along three cardinal axes and for ±3.5% range uncertainty. For tumors, these nine dose distributions were explicitly controlled by adding hard constraints with adjustable parameters. For OARs, fuzzy constraints that allow the dose to vary within a certain range were used so that the tumor dose distribution was guaranteed by minimum compromise of that of OARs. We rendered this model tractable by converting the fuzzy constraints to linear constraints. The plan quality was evaluated using dose-volume histogram(DVH) indices such as tumor dose coverage(D95%), homogeneity(D5%-D95%), plan robustness(DVH band at D95%), and OAR sparing like D1% of brain and D1% of brainstem. Results: Our model could yield clinically acceptable plans. The fuzzy-logic robust optimization method produced IMPT plans with comparable target dose coverage and homogeneity compared to the PTV method(unit: Gy[RBE]; average[min, max])(CTV D95%: 59 [52.7, 63.5] vs 53.5[46.4, 60.1], CTV D5% - D95%: 11.1[5.3, 18.6] vs 14.4[9.2, 21.5]). It also generated more robust plans(CTV DVH band at D95%: 3.8[1.2, 5.6] vs 11.5[6.2, 16.7]). The parameters of tumor constraints could be adjusted to control the tradeoff between tumor coverage and OAR sparing. Conclusion: The fuzzy-logic robust optimization generates superior IMPT with minimum compromise of OAR sparing. This research was supported by the National Cancer Institute Career Developmental Award K25CA168984, by the Fraternal Order of Eagles Cancer Research Fund Career Development Award, by The Lawrence W. and Marilyn W. Matteson Fund for Cancer Research, by Mayo Arizona State University Seed Grant, and by The Kemper Marley Foundation. eRA Person ID(s) for the Principal Investigator: 11017970 (Research Supported by National Institutes of Health)

60 APPLIED LIFE SCIENCES↗

Function Maximization with Dynamic Quantum Search

Finding the maximum value of a function in a dynamic model plays an important role in many application settings, including discrete optimization in the presence of hard constraints. We present an iterative quantum algorithm for finding the maximum value of a function in which prior search results update the acceptable response. Our approach is based on quantum search and utilizes a dynamic oracle function to mark items in a specified input set. As a realization of function optimization, we verify the correctness of the algorithm using numerical simulations of quantum circuits for the Knapsack problem. In conclusion, our simulations make use of an explicit oracle function based on arithmetic operations and a comparator subroutine, and we verify these implementations using numerical simulations up to 30 qubits.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Robust state estimation of feeding–blending systems in continuous pharmaceutical manufacturing

State estimation is a fundamental part of monitoring, control, and real-time optimization in continuous pharmaceutical manufacturing. For nonlinear dynamic systems with hard constraints, moving horizon estimation (MHE) can estimate the current state by solving a well-defined optimization problem where process complexities are explicitly considered as constraints. Traditional MHE techniques assume random measurement noise governed by some normal distributions. However, state estimates can be unreliable if noise is not normally distributed or measurements are contaminated with gross or systematic errors. In this paper, to improve the accuracy and robustness of state estimation, we incorporate robust estimators within the standard MHE skeleton, leading to an extended MHE framework. The proposed MHE approach is implemented on two pharmaceutical continuous feeding–blending system (FBS) configurations which include loss-in-weight (LIW) feeders and continuous blenders. Numerical results show that our MHE approach is robust to gross errors and can provide reliable state estimates when measurements are contaminated with outliers and drifts. Finally and moreover, the efficient solution of the MHE realized in this work, suggests feasible application of on-line state estimation on more complex continuous pharmaceutical processes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Modular machine learning-based elastoplasticity: Generalization in the context of limited data

The development of highly accurate constitutive models for materials that undergo path-dependent processes continues to be a complex challenge in computational solid mechanics. Challenges arise both in considering the appropriate model assumptions and from the viewpoint of data availability, verification, and validation. Recently, data-driven modeling approaches have been proposed that aim to establish stress-evolution laws that avoid user-chosen functional forms by relying on machine learning representations and algorithms. However, these approaches not only require a significant amount of data but also need data that probes the full stress space with a variety of complex loading paths. Furthermore, they rarely enforce all necessary thermodynamic principles as hard constraints. Hence, they are in particular not suitable for low-data or limited-data regimes, where the first arises from the cost of obtaining the data and the latter from the experimental limitations of obtaining labeled data, which is commonly the case in engineering applications. In this work, we discuss a hybrid framework that can work on a variable amount of data by relying on the modularity of the elastoplasticity formulation where each component of the model can be chosen to be either a classical phenomenological or a data-driven model depending on the amount of available information and the complexity of the response. The method is tested on synthetic uniaxial data coming from simulations as well as cyclic experimental data for structural materials. The discovered material models are found to not only interpolate well but also allow for accurate extrapolation in a thermodynamically consistent manner far outside the domain of the training data. This ability to extrapolate from limited data was the main reason for the early and continued success of phenomenological models and the main shortcoming in machine learning-enabled constitutive modeling approaches. Training aspects and details of the implementation of these models into Finite Element simulations are discussed and analyzed.

42 ENGINEERING↗

Toward Fully Automated Multicriterial Plan Generation: A Prospective Clinical Study

Purpose: To prospectively compare plans generated with iCycle, an in-house-developed algorithm for fully automated multicriterial intensity modulated radiation therapy (IMRT) beam profile and beam orientation optimization, with plans manually generated by dosimetrists using the clinical treatment planning system. Methods and Materials: For 20 randomly selected head-and-neck cancer patients with various tumor locations (of whom 13 received sequential boost treatments), we offered the treating physician the choice between an automatically generated iCycle plan and a manually optimized plan using standard clinical procedures. Although iCycle used a fixed “wish list” with hard constraints and prioritized objectives, the dosimetrists manually selected the beam configuration and fine tuned the constraints and objectives for each IMRT plan. Dosimetrists were not informed in advance whether a competing iCycle plan was made. The 2 plans were simultaneously presented to the physician, who then selected the plan to be used for treatment. For the patient group, differences in planning target volume coverage and sparing of critical tissues were quantified. Results: In 32 of 33 plan comparisons, the physician selected the iCycle plan for treatment. This highly consistent preference for the automatically generated plans was mainly caused by the improved sparing for the large majority of critical structures. With iCycle, the normal tissue complication probabilities for the parotid and submandibular glands were reduced by 2.4% ± 4.9% (maximum, 18.5%, P=.001) and 6.5% ± 8.3% (maximum, 27%, P=.005), respectively. The reduction in the mean oral cavity dose was 2.8 ± 2.8 Gy (maximum, 8.1 Gy, P=.005). For the swallowing muscles, the esophagus and larynx, the mean dose reduction was 3.3 ± 1.1 Gy (maximum, 9.2 Gy, P<.001). For 15 of the 20 patients, target coverage was also improved. Conclusions: In 97% of cases, automatically generated plans were selected for treatment because of the superior quality. Apart from the improved plan quality, automatic plan generation is economically attractive because of the reduced workload.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Application of Quantum Annealing to Nurse Scheduling Problem

Quantum annealing is a promising heuristic method to solve combinatorial optimization problems, and efforts to quantify performance on real-world problems provide insights into how this approach may be best used in practice. We investigate the empirical performance of quantum annealing to solve the Nurse Scheduling Problem (NSP) with hard constraints using the D-Wave 2000Q quantum annealing device. NSP seeks the optimal assignment for a set of nurses to shifts under an accompanying set of constraints on schedule and personnel. After reducing NSP to a novel Ising-type Hamiltonian, we evaluate the solution quality obtained from the D-Wave 2000Q against the constraint requirements as well as the diversity of solutions. For the test problems explored here, our results indicate that quantum annealing recovers satisfying solutions for NSP and suggests the heuristic method is potentially achievable for practical use. Moreover, we observe that solution quality can be greatly improved through the use of reverse annealing, in which it is possible to refine returned results by using the annealing process a second time. We compare the performance of NSP using both forward and reverse annealing methods and describe how this approach might be used in practice.

97 MATHEMATICS AND COMPUTING↗

Physics-constrained machine learning for electrodynamics without gauge ambiguity based on Fourier transformed Maxwell’s equations

We utilize a Fourier transformation-based representation of Maxwell’s equations to develop physics-constrained neural networks for electrodynamics without gauge ambiguity, which we label the Fourier–Helmholtz–Maxwell neural operator method. In this approach, both of Gauss’s laws and Faraday’s law are built in as hard constraints, as well as the longitudinal component of Ampère–Maxwell in Fourier space, assuming the continuity equation. An encoder–decoder network acts as a solution operator for the transverse components of the Fourier transformed vector potential, $\hat{A}_⟂(k,t)$, whose two degrees of freedom are used to predict the electromagnetic fields. This method was tested on two electron beam simulations. Among the models investigated, it was found that a U-Net architecture exhibited the best performance as it trained quicker, was more accurate and generalized better than the other architectures examined. We demonstrate that our approach is useful for solving Maxwell’s equations for the electromagnetic fields generated by intense relativistic charged particle beams and that it generalizes well to unseen test data, while being orders of magnitude quicker than conventional simulations. We show that the model can be re-trained to make highly accurate predictions in as few as 20 epochs on a previously unseen data set.

97 MATHEMATICS AND COMPUTING↗

Physics-constrained 3D convolutional neural networks for electrodynamics

We present a physics-constrained neural network (PCNN) approach to solving Maxwell’s equations for the electromagnetic fields of intense relativistic charged particle beams. We create a 3D convolutional PCNN to map time-varying current and charge densities J(r, t) and ρ(r, t) to vector and scalar potentials A(r, t) and φ(r, t) from which we generate electromagnetic fields according to Maxwell’s equations: B = ∇ × A and E = −∇φ − ∂A/∂t. Our PCNNs satisfy hard constraints, such as ∇ · B = 0, by construction. Soft constraints push A and φ toward satisfying the Lorenz gauge.

97 MATHEMATICS AND COMPUTING↗

FORECAST FOR THE PLANCK PRECISION ON THE TENSOR-TO-SCALAR RATIO AND OTHER COSMOLOGICAL PARAMETERS

The Planck satellite, successfully launched on 2009 May 14 to measure with unprecedented accuracy the primary cosmic microwave background (CMB) anisotropies, is operating as expected. The Standard Model of the Universe ('concordance' model) provides the current realistic context to analyze the CMB and other cosmological/astrophysical data, inflation in the early universe being part of it. The Planck performance for the crucial primordial parameter r, the tensor-to-scalar ratio related to primordial B-mode polarization, will depend on the quality of data analysis and interpretation. The Ginzburg-Landau (G-L) approach to inflation allows us to take high benefit of the CMB data. The fourth-degree double-well inflaton potential gives an excellent fit to the current CMB+LSS data. We evaluate the Planck precision to the recovery of cosmological parameters, taking into account a reasonable toy model for residuals of systematic effects of instrumental and astrophysical origin based on publicly available information. We use and test two relevant models: the {Lambda}CDMr model, i.e., the standard {Lambda}CDM model augmented by r, and the {Lambda}CDMrT model, where the scalar spectral index, n{sub s} , and r are related through the theoretical 'banana-shaped' curve r = r(n{sub s}) coming from the G-L theory with a double-well inflaton potential. In the latter case, the analytical expressions for n{sub s} and r are imposed as a hard constraint in a Monte Carlo Markov Chain (MCMC) data analysis. We consider two C{sub l}-likelihoods (with and without B modes) and take into account the white noise sensitivity of Planck (LFI and HFI) in the 70, 100, and 143 GHz channels as well as the residuals from systematic errors and foregrounds. We also consider a cumulative channel of the three mentioned. We produce the sky (mock data) for the CMB multipoles C{sup TT}{sub l} , C{sup TE}{sub l} , C{sup EE}{sub l} , and C{sup BB}{sub l} from the {Lambda}CDMr and {Lambda}CDMrT models and obtain the cosmological parameter marginalized likelihood distributions for the two models. Foreground residuals affect only the cosmological parameters sensitive to the B modes. As expected, the likelihood r distribution is more clearly peaked near the fiducial value (r = 0.0427) in the {Lambda}CDMrT model than in the {Lambda}CDMr model. The best value for r in the presence of residuals turns out to be about r {approx_equal} 0.04 for both the {Lambda}CDMr and the {Lambda}CDMrT models. The {Lambda}CDMrT model is very stable; its distributions do not change by including residuals and the B modes. For r we find 0.028 < r < 0.116 at a 95% confidence level (CL) with the best value r = 0.04. We also compute the B mode detection probability by the most sensitive HFI-143 channel. At the level of foreground residual equal to 30% of our toy model, only a 68% CL (1{sigma}) detection is very likely. For a 95% CL detection (2{sigma}), the level of foreground residual should be reduced to 10% or lower of the adopted toy model. The lower bounds (and most probable value) we infer for r support the searching of CMB B-mode polarization in the current data as well as the planned CMB missions oriented toward B polarization.

79 ASTRONOMY AND ASTROPHYSICS↗

3-regular three-XORSAT planted solutions benchmark of classical and quantum heuristic optimizers

With current semiconductor technology reaching its physical limits, special-purpose hardware has emerged as an option to tackle specific computing-intensive challenges. Optimization in the form of solving quadratic unconstrained binary optimization problems, or equivalently Ising spin glasses, has been the focus of several new dedicated hardware platforms. These platforms come in many different flavors, from highly-efficient hardware implementations on digital-logic of established algorithms to proposals of analog hardware implementing new algorithms. In this work, we use a mapping of a specific class of linear equations whose solutions can be found efficiently, to a hard constraint satisfaction problem (three-regular three-XORSAT, or an Ising spin glass) with a 'golf-course' shaped energy landscape, to benchmark several of these different approaches. We perform a scaling and prefactor analysis of the performance of Fujitsu's digital annealer unit (DAU), the D-Wave advantage quantum annealer, a virtual MemComputing machine, Toshiba's simulated bifurcation machine (SBM), the SATonGPU algorithm from Bernaschi et al, and our implementation of parallel tempering. We identify the SATonGPU and DAU as currently having the smallest scaling exponent for this benchmark, with SATonGPU having a small scaling advantage and in addition having by far the smallest prefactor thanks to its use of massive parallelism. Furthermore, our work provides an objective assessment and a snapshot of the promise and limitations of dedicated optimization hardware relative to a particular class of optimization problems.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Testing of inductively coupled Eddy current position sensor of diverse safety rod in sodium

Prototype Fast Breeder Reactor (PFBR) is 500 MWe sodium cooled reactor under construction at Kalpakkam (India)). To improve the reliability of shutdown, Diverse Safety Rods (DSRs) are used in-addition to normal Control and Safety rods. During reactor operating condition, the DSR is parked above the active core and held in its top position by an electromagnet. In the event of a scram signal from the safety logic, the electromagnet holding the DSR is de-energised. Hence the DSR is released into the active core and at the end of travel DSR gets deposited in its bottom position. Because of the mechanical constraints, hard wired connectivity is not permitted from the DSR subassembly to the instrumentation outside the reactor. Hence an inductively coupled Eddy Current Position Sensor (ECPS) has been conceptualized to detect that the DSR has reached its bottom most position and to measure the drop time. Results of feasibility study on laboratory model have been reported earlier. Testing of a 1:1 scale engineering model of ECPS is reported in this paper. Results obtained from the high temperature sodium testing of ECPS indicate a clearly measurable change in pick up voltage with sensitivity of 11 % at 675 Hz. The ECPS is in advanced stage of implementation in DSRDM of PFBR. (authors)

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

A Reference Governor for Plasma-Shape Control in Tokamaks

Tokamaks are torus-shaped devices where a high-temperature, ionized gas (i.e. a plasma) is confined by means of helical magnetic fields. The ultimate goal of tokamaks is producing net electrical energy by means of nuclear-fusion reactions within the plasma. The shape of the plasma inside the tokamak is closely related with its confinement characteristics and, therefore, with the produced fusion power. In order to achieve and maintain plasma shapes that maximize the fusion yield, feedback controllers have been successfully developed during the years. However, machine-protection and plasma-stability requirements impose hard constraints on the plasma-shape targets that can be safely achieved by a feedback controller. In this work, a reference governor is proposed to provide, in real time, plasma-shape targets that fulfill the aforementioned safety requirements. The reference governor calculates such plasma-shape targets in response to changes in the state of the closed-loop system, as well as disturbances in the plasma energy and current. Here, the reference-governor design is based on a dynamical model of the plasma electromagnetic response. Simulation tests of the reference governor have been carried out for different plasma scenarios in the DIII-D tokamak to emulate realistic cases where safety may be compromised.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗