Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Optimization Under Uncertainty”

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 1,261 records · Page 70

HYPPO: A Surrogate-Based Multi-Level Parallelism Tool for Hyperparameter Optimization

We present a new software, HYPPO, that enables the automatic tuning of hyperparameters of various deep learning (DL) models. Unlike other hyperparameter optimization (HPO) methods, HYPPO uses adaptive surrogate models and directly accounts for uncertainty in model predictions to find accurate and reliable models that make robust predictions. Using asynchronous nested parallelism, we are able to significantly alleviate the computational burden of training complex architectures and quantifying the uncertainty. HYPPO is implemented in Python and can be used with both TensorFlow and PyTorch libraries. We demonstrate various software features on time-series prediction and image classification problems as well as a scientific application in computed tomography image reconstruction. Finally, we show that (1) we can reduce by an order of magnitude the number of evaluations necessary to find the most optimal region in the hyperparameter space and (2) we can reduce by two orders of magnitude the throughput for such HPO process to complete.

adaptation models↗

SU-E-T-42: A Method to Determine the Optimal Proton Scanned Beam Angle to Ensure Robust Treatment Planning

Purpose: The ability of pencil beam scanning (PBS) to deliver highly conformal dose distributions may be affected by patient- and physics-related uncertainties. In clinical practice, selection of proton beam angles is determined qualitatively. This study investigates whether an optimal proton PBS beam angle could be quantitatively determined to ensure robust planning for pelvic targets. Methods: PBS beam angles were optimized based on two independent criteria; shortest and most homogeneous path from the patient surface to the distal edge of the target. The beam angle optimization criteria for gantry angles between 90°-270° were quantified in 10° increments for each ray, calculated as the straight line distance from the surface of the skin to the CTV’s distal edge. The goal was to minimize the path length of a proton PBS beam from the patient surface to the distal edge of the CTV, relative to the entry point, while minimizing HU inhomogeneity along the ray. HU homogeneity (i.e. HU variation) was quantitatively defined as the standard deviation of the average intra-ray HU intensity distribution of the rays comprising a single beam. This method was validated relative to inter-fraction changes on ten consecutive, locally advanced, rectal cancer patients, who underwent an average 4 verification CTs. The displacement of the 95–98% isodose lines was determined from forward calculated dose distributions on verification CTs. Results: The posterior beam (180°) had the average shortest path length, 132.7±17.2mm, and the most homogenous path, 31.9±4.3HU. The 95–98% isodose lines from all plans verified our path length to within 2.3±1.2% and HU homogeneity to within 1.2±0.5%. Conclusion: The proposed optimization algorithm determined the posterior beam dose distribution as the most robust relative to inter-fraction variation for large pelvic targets treated with PBS and was validated via verification CT for our patient cohort. Future work will focus on further algorithm development.

60 APPLIED LIFE SCIENCES↗

SU-F-T-192: Study of Robustness Analysis Method of Multiple Field Optimized IMPT Plans for Head & Neck Patients

Purpose: Proton therapy is more sensitive to uncertainties than photon treatments due to protons’ finite range depending on the tissue density. Worst case scenario (WCS) method originally proposed by Lomax has been adopted in our institute for robustness analysis of IMPT plans. This work demonstrates that WCS method is sufficient enough to take into account of the uncertainties which could be encountered during daily clinical treatment. Methods: A fast and approximate dose calculation method is developed to calculate the dose for the IMPT plan under different setup and range uncertainties. Effects of two factors, inversed square factor and range uncertainty, are explored. WCS robustness analysis method was evaluated using this fast dose calculation method. The worst-case dose distribution was generated by shifting isocenter by 3 mm along x,y and z directions and modifying stopping power ratios by ±3.5%. 1000 randomly perturbed cases in proton range and x, yz directions were created and the corresponding dose distributions were calculated using this approximated method. DVH and dosimetric indexes of all 1000 perturbed cases were calculated and compared with the result using worst case scenario method. Results: The distributions of dosimetric indexes of 1000 perturbed cases were generated and compared with the results using worst case scenario. For D95 of CTVs, at least 97% of 1000 perturbed cases show higher values than the one of worst case scenario. For D5 of CTVs, at least 98% of perturbed cases have lower values than worst case scenario. Conclusion: By extensively calculating the dose distributions under random uncertainties, WCS method was verified to be reliable in evaluating the robustness level of MFO IMPT plans of H&N patients. The extensively sampling approach using fast approximated method could be used in evaluating the effects of different factors on the robustness level of IMPT plans in the future.

60 APPLIED LIFE SCIENCES↗

Efficient Evaluation and Optimization of Structural Reliability

Use of approximations for calculating the structural failure probability under uncertainties is a necessary aspect. But efficient use of suitable approximations at different stages of reliability prediction and design makes this valuable tool practical for many large scale structures. This research addresses the modern approximation concepts and their utility in structural reliability estimation. Emphasis is placed on building higher order approximations using intervening variables to closely represent the nonlinear limit state functions. Several test problems with highly nonlinear performance functions are used to demonstrate the accuracy of the proposed method. Many of the structural problems are modeled and simulated using the finite element methods (FEM) for obtaining a detailed structural response. FEM is a computationally intensive numerical procedure with a large number of degrees of freedom. With an increase in the complexity of the structural model and the multidisciplinary nature of analyses, the number of limit states and their computation increases very significantly. The requirement of frequent repetitive analysis of the structure hinders the utility of any tool and discourages that approach from being used. One particular tool which faces this difficulty is the computation of the structural reliability under parametric uncertainties.

Grandhi, Ramana V.↗

Multifidelity Analysis and Optimization for Supersonic Design

Supersonic aircraft design is a computationally expensive optimization problem and multifidelity approaches over a significant opportunity to reduce design time and computational cost. This report presents tools developed to improve supersonic aircraft design capabilities including: aerodynamic tools for supersonic aircraft configurations; a systematic way to manage model uncertainty; and multifidelity model management concepts that incorporate uncertainty. The aerodynamic analysis tools developed are appropriate for use in a multifidelity optimization framework, and include four analysis routines to estimate the lift and drag of a supersonic airfoil, a multifidelity supersonic drag code that estimates the drag of aircraft configurations with three different methods: an area rule method, a panel method, and an Euler solver. In addition, five multifidelity optimization methods are developed, which include local and global methods as well as gradient-based and gradient-free techniques.

Kroo, Ilan↗

RAVEN User Manual, Revision 4

RAVEN is a generic software framework to perform parametric and probabilistic analysis based on the response of complex system codes. The initial development was aimed to provide dynamic risk analysis capabilities to the Thermo-Hydraulic code RELAP-7, currently under development at the Idaho National Laboratory (INL). Although the initial goal has been fully accomplished, RAVEN is now a multi-purpose probabilistic and uncertainty quantification platform, capable to agnostically communicate with any system code. This agnosticism includes providing Application Programming Interfaces (APIs). These APIs are used to allow RAVEN to interact with any code as long as all the parameters that need to be perturbed are accessible by inputs files or via python interfaces. RAVEN is capable of investigating the system response, and investigating the input space using Monte Carlo, Grid, or Latin Hyper Cube sampling schemes, but its strength is focused to- ward system feature discovery, such as limit surfaces, separating regions of the input space leading to system failure, using dynamic supervised learning techniques. The development of RAVEN has started in 2012, when, within the Nuclear Energy Advanced Modeling and Simulation (NEAMS) program, the need to provide a modern risk evaluation framework became stronger. RAVEN principal assignment is to provide the necessary software and algorithms in order to employ the concept developed by the Risk Informed Safety Margin Characterization (RISMC) program. RISMC is one of the pathways defined within the Light Water Reactor Sustainability (LWRS) program. In the RISMC approach, the goal is not just the individuation of the frequency of an event potentially leading to a system failure, but the closeness (or not) to key safety-related events. Hence, the approach is interested in identifying and increasing the safety margins related to those events. A safety margin is a numerical value quantifying the probability that a safety metric (e.g. for an important process such as peak pressure in a pipe) is exceeded under certain conditions. The initial development of RAVEN has been focused on providing dynamic risk assess- ment capability to RELAP-7, currently under development at the INL and, likely, future replacement of the RELAP5-3D code. Most the capabilities that have been implemented having RELAP-7 as principal focus are easily deployable for other system codes. For this reason, several side activates are currently ongoing for coupling RAVEN with soft- ware such as RELAP5-3D, etc. The aim of this document is the explanation of the input requirements, focalizing on the input structure.

97 MATHEMATICS AND COMPUTING↗

Spatial Statistical Data Fusion (SSDF)

As remote sensing for scientific purposes has transitioned from an experimental technology to an operational one, the selection of instruments has become more coordinated, so that the scientific community can exploit complementary measurements. However, tech nological and scientific heterogeneity across devices means that the statistical characteristics of the data they collect are different. The challenge addressed here is how to combine heterogeneous remote sensing data sets in a way that yields optimal statistical estimates of the underlying geophysical field, and provides rigorous uncertainty measures for those estimates. Different remote sensing data sets may have different spatial resolutions, different measurement error biases and variances, and other disparate characteristics. A state-of-the-art spatial statistical model was used to relate the true, but not directly observed, geophysical field to noisy, spatial aggregates observed by remote sensing instruments. The spatial covariances of the true field and the covariances of the true field with the observations were modeled. The observations are spatial averages of the true field values, over pixels, with different measurement noise superimposed. A kriging framework is used to infer optimal (minimum mean squared error and unbiased) estimates of the true field at point locations from pixel-level, noisy observations. A key feature of the spatial statistical model is the spatial mixed effects model that underlies it. The approach models the spatial covariance function of the underlying field using linear combinations of basis functions of fixed size. Approaches based on kriging require the inversion of very large spatial covariance matrices, and this is usually done by making simplifying assumptions about spatial covariance structure that simply do not hold for geophysical variables. In contrast, this method does not require these assumptions, and is also computationally much faster. This method is fundamentally different than other approaches to data fusion for remote sensing data because it is inferential rather than merely descriptive. All approaches combine data in a way that minimizes some specified loss function. Most of these are more or less ad hoc criteria based on what looks good to the eye, or some criteria that relate only to the data at hand.

Braverman, Amy J.↗

A forecast-driven decision-making model for long-term operation of a hydro-wind-photovoltaic hybrid system

Hydro-wind-photovoltaic (PV) hybrid system has the potential to increase the integration of renewable energy sources into an existing grid. For the long-term operation of the system, due to the non-storable nature of wind and PV power, it is essentially to decide the optimal long-term carryover storage of cascade reservoirs. However, it remains a challenge due to high uncertainties of long-term forecasts and complicated hydraulic/electrical relationships between cascade reservoirs. Here in this study, a forecast-driven decision-making model is proposed for the hybrid system, which converts the multi-stage long-term operation process into a two-stage operation problem (including current stage and carryover stage) to avoid using longer-horizon forecast information with lower accuracy. First, the carryover stage energy surfaces (CESs) considering the forecast uncertainties of wind, solar and hydro resources are proposed to characterize carryover stage benefit quantitatively. Then a CESs-based forward decision-making optimization model is developed to guide the long-term operation of a hydro-wind-photovoltaic hybrid system. The applications in a hydro-wind-PV hybrid system of Yalong River basin results show that: compared with conventional operation, 1) power generation increases 9.03%; 2) in terms of the carryover storages control, the reservoir impounding and drawdown timing are delayed, and the drawdown depth is increased, which can be used to formulate better reservoir operation rules.

13 HYDRO ENERGY↗

Discrete-time entropy formulation of optimal and adaptive control problems

The discrete-time version of the entropy formulation of optimal control of problems developed by G. N. Saridis (1988) is discussed. Given a dynamical system, the uncertainty in the selection of the control is characterized by the probability distribution (density) function which maximizes the total entropy. The equivalence between the optimal control problem and the optimal entropy problem is established, and the total entropy is decomposed into a term associated with the certainty equivalent control law, the entropy of estimation, and the so-called equivocation of the active transmission of information from the controller to the estimator. This provides a useful framework for studying the certainty equivalent and adaptive control laws.

Tsai, Yweting A.↗

Microgrid energy scheduling under uncertain extreme weather: Adaptation from parallelized reinforcement learning agents

Microgrids are useful solutions for integrating renewable energy resources and providing seamless green electricity to minimize carbon footprint. In recent years, extreme weather events happened often worldwide and caused significant economic and societal losses. Such events bring uncertainties to the microgrid energy scheduling problems and increase the challenges of microgrid operation. Traditional optimization approaches suffer from the inaccuracy of the uncertain microgrid model and the unseen events. Existing reinforcement learning (RL) - based approaches are also hampered by the limited generalization and the increasing computational burden when stochastic formulations are required to accommodate the uncertainties. This paper proposes a new parallelized reinforcement learning (PRL) method based on the probabilistic events to handle the microgrid energy uncertainties. Specifically, several local learning agents are employed to interact with pertinent microgrid environments in a distributed manner and report outcomes to the global agent, which will optimize microgrid energy resources online during extreme events. The stochastic microgrid energy optimization problem is reformulated to include all possible scenarios with probabilities. The advantage estimate functions of learning agents are designed with a backward sweep to transfer the outcomes to the value function updating process. Two simulation studies, stochastic optimization and online testing, are performed to compare with several existing RL approaches. Results substantiate that the proposed PRL method can achieve up to 20% optimization performance improvement with 4 and 28 times less computation cost than Q-learning with experience replay and multi-agent Q-learning approaches, respectively.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Some Issues in Uncertainty Quantification and Parameter Tuning: A Case Study of Convective Parameterization Scheme in the WRF Regional Climate Model

The current tuning process of parameters in global climate models is often performed subjectively, or treated as an optimization procedure to minimize the difference between model fields and observations. The later approach may be generating a set of tunable parameters that approximate the observed climate but via an unrealistic balance of physical processes and/or compensating errors over different regions in the globe. In this study, we run the Weather Research and Forecasting (WRF) regional model constrained by the reanalysis data over the Southern Great Plains (SGP) where abundant observational data from various resources are available for calibration of the input parameters and validation of the model results. Our goal is to quantify the uncertainty ranges and identify the optimal values of five key input parameters in a new Kain-Frisch (KF) convective parameterization scheme incorporated in the WRF model. A stochastic sampling algorithm, Multiple Very Fast Simulated Annealing (MVFSA), is employed to efficiently sample the input parameters in KF scheme based on the skill score so that the algorithm progressively moves toward regions of the parameter space that minimize model errors. The results based on the WRF simulations with 25-km grid spacing over the SGP show that the model bias for precipitation can be significantly reduced by using five optimal parameters identified by the MVFSA algorithm. The model performance is very sensitive to downdraft and entrainment related parameters and consumption time of Convective Available Potential Energy (CAPE). Simulated convective precipitation decreases as the ratio of downdraft to updraft flux increases. Larger CAPE consumption time results in less convective but more stratiform precipitation. The simulation using optimal parameters obtained by only constraining precipitation generates positive impact on the other output variables, such as temperature and wind. By using the optimal parameters obtained at 25 km simulation, both the magnitude and spatial pattern of simulated precipitation are improved at 12-km spatial resolution. The optimal parameters identified from the SGP region have also improved the simulation of precipitation when moving model domain to another region with a different climate regime (i.e. North America monsoon region). These results suggest the improvement of precipitation simulation by using the optimal parameters remains when the model domain or spatial resolution is changed.

54 ENVIRONMENTAL SCIENCES↗

Atikokan Digital Twin: Machine learning in a biomass energy system

The Atikokan Generating Station, operated by Ontario Power Generation, has a 200 MW, biomass-fired tower boiler that operates on a dispatch schedule with a five-minute cycle. The boiler is generally operated in the range of 40–100 MW using two of five burner levels. In order to optimize boiler performance, we propose the implementation of a unique digital twin. Our digital twin abstraction couples Bayesian inference from science-based models and from observations (machine learning) with decision theory to predict operating-variable set points that optimize the physical asset (the boiler) in the presence of uncertainty (artificial intelligence). We focus this paper on the continuous Bayesian machine learning part of the Atikokan Digital Twin; we discuss decision theory in a companion paper. We identify and learn about 12 operational, model, and measured-output parameters and their uncertainties from high-fidelity, science-based simulations of the Atikokan boiler and from the observed measurements at the power plant. Since the goal of the Atikokan Digital Twin is to implement it online in real time, we require fast function evaluations for the quantities of interest extracted from the simulations in the Bayesian analysis. We use Gaussian process regression/interpolation to create accurate, robust surrogate models. We define the Bayesian priors and likelihood function and solve for the posterior distributions of the 12 parameters. Here we then propagate these distributions (i.e., parameters with uncertainty) into the predicted distributions of 790 quantities of interest to learn about the relative importance of various sources of error including experimental, model, and operating-parameter errors.

09 BIOMASS FUELS↗

A Multi-Parametric optimization approach for bilevel mixed-integer linear and quadratic programming problems

Optimization problems involving two decision makers at two different decision levels are referred to as bi-level programming problems. Herein, we present novel algorithms for the exact and global solution of two classes of bi-level programming problems, namely (i) bi-level mixed-integer linear programming problems (B-MILP) and (ii) bi-level mixed-integer convex quadratic programming problems (B-MIQP) containing both integer and bounded continuous variables at both optimization levels. Based on multi-parametric programming theory, the main idea is to recast the lower level problem as a multi-parametric programming problem, in which the optimization variables of the upper level problem are considered as bounded parameters for the lower level. The resulting exact multi-parametric mixed-integer linear or quadratic solutions are then substituted into the upper level problem, which can be solved as a set of single-level, independent, deterministic mixed-integer optimization problems. Extensions to problems including right-hand-side uncertainty on both lower and upper levels are also discussed. Finally, computational implementation and studies are presented through test problems.

42 ENGINEERING↗

Fast strength model characterization using Bayesian statistics

A variety of flow stress models exist with new models constantly being developed. These models aim to approximate the strength of materials in a variety of regimes from quasistatic loading through shock scenarios. All models contain an array of parameters which need to be tuned to the material under study. Some models perform well under limited conditions, requiring adjustment of the parameters when venturing outside of those predefined ranges. Other models perform well over a wide range of conditions with a set of parameters, but may be outperformed by other models optimized on a tighter range of conditions. Recent research by Los Alamos demonstrated the ability to optimize the Johnson Cook (JC) model using a set of 3 plate-impact experiments on Aluminum. They utilized Bayesian statistics and emulation to determine optimal parameters for the model with a quantification of parameter uncertainty. We present an extension of this capability to incorporate velocimetry from plate-impact tests, stress-strain data from split Hopkinson pressure bar and quasistatic compression tests, plus profiles from Taylor cylinders in a unified fashion. Statistically robust comparisons of the performance and uncertainty of different realizations of the JC flow stress model were carried out based on calibration to several possible combinations of these three different experiment types.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

An open source analysis framework for large-scale building energy modeling

Full integration of building energy modelling into the design and retrofit process has long been a goal of building scientists and practitioners. However, significant barriers still exist. Among them are the lack of available: (1) configurable technology stacks for performing both small- and large-scale analyses, (2) different classes of algorithms compatible with common design workflows, and (3) analysis tools for effectively visualizing large-scale simulation results. This article discusses the OpenStudio® Analysis Framework: a scalable analysis framework for building energy modelling that was developed to overcome the three barriers listed above. The framework is open-source and scalable to facilitate wider adoption and has a clearly defined application programming interface upon which other applications can be built. It runs on high-performance computing systems, within cloud infrastructure, and on laptops, and uses a common workflow to enable different classes of algorithms. Lessons learned from previous development efforts are also discussed.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Machine-learning-assisted automation of single-crystal neutron diffraction

Neutron scattering is a powerful but expensive technique to study materials and discover new matter. Advanced detector technology has significantly improved the efficiency of neutron experiments, increasing the complexity of neutron data reduction and analysis. Machine learning (ML) brings new directions for neutron diffraction data reduction and experiment operation. Here, this work presents an ML-assisted data reduction and analysis method for precise recognition of Bragg peaks and the corresponding regions of interest; it can then automatically screen and align a measured crystal using the recognized peaks, and subsequently plan and optimize the data collection with user-provided information and uncertainty quantification values of detected peaks. This method shows robust performance in different complex sample environments and enables automated single-crystal neutron diffraction.

47 OTHER INSTRUMENTATION↗

An Assessment of the Impact of Stochastic Day-Ahead SCUC on Economic and Reliability Metrics at Multiple Timescales

This paper incorporates the stochastic day-ahead security-constrained unit commitment (DASCUC) within a multi-timescale, multi-scheduling application with commitment, dispatch, and automatic generation control. The stochastic DASCUC is solved using a progressive hedging algorithm with constrained ordinal optimization to accelerate the individual scenario solution. Sensitivity studies are performed in the RTS-96 system, and the results show how this new scheduling application would impact costs and reliability with a closer representation of timescales of system operations in practice.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Identification of Critical Switches for Integrating Renewable Distributed Generation by Dynamic Network Reconfiguration

With growing penetration of renewable distributed generations (DGs) in distribution systems, effective integration of DGs has become a major concern. Distribution system dynamic reconfiguration (DSDR), which relies on real-time operations of remote-controlled switches, is potentially an efficient strategy receiving inadequate attention. Moreover, in most DSDR-related publications, normally all switches are assumed remotely controllable, which is not practical. Here we borrow the concept of critical switches to denote the switches that are most effective in accommodating DGs by DSDR. In this regard, the problem of identifying critical switches is not well investigated, although in several related publications, selected switches are assumed remote-controlled based on experience. In this work, we study the application of DSDR for DG integration. Critical switches, which optimally enable intra-day DSDR to minimize DG curtailments, are identified by limiting the number of switches to be operated and the switch-type-dependent operation constraints. Considering uncertainties of loads and DGs, the problem is formulated as a two-stage robust optimization model solved by a nested column-and-constraint generation algorithm. Illustrative cases show that DG curtailments can be significantly reduced by a small number of critical switches that operate only several times in intra-day DSDR. The proposed method can be used to provide insights for switch allocation, maintenance, and operation

Lei, Shunbo (ORCID:0000000181690821)↗