SEARCH · Search NASA

Results for “Decentralized 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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4

All sources NASA NTRS DOE OSTI

DOE OSTI2022

Secure mmWave Spectrum Sharing with Autonomous Beam Scheduling for 5G and Beyond

Spectrum Sharing (SS) has seen a renewed set of initiatives in 5G with the availability of shared and unlicensed spectrum bands that can be used by multiple cellular service providers and private cellular networks. Beam based transmission, instead of the traditional sector based transmission in conjunction with the spectrum agility of the 5G New Radio (NR) has brought new opportunities to optimized sharing of spectrum. Currently in the U.S., a centralized Spectrum Access Server (SAS) is used to co-ordinate spectrum sharing among networks sharing the same spectrum band. However, SAS becomes a focal point for security attacks and a performance bottleneck. In addition, SAS relies on an Environmental Sensor Network (ESN), separate from the 5G network. Without trusted spectral occupancy information, false reporting of spectrum sensing data can create sub-optimal and unfair spectrum usage. This paper summarizes our recent research findings in using a decentralized scheme for multiple networks to securely share spectrum with autonomous beam scheduling : 1) A new stochastic network framework based on Lyapunov Optimization approach is developed to optimize scheduling at the base stations; 2) Game theoretic (GT) approach is used to formulate the distributed scheduler; 3) Another distributed scheduler with Q-learning is presented that utilizes the Reinforcement Learning (RL) approach; 4) The performance and convergence rate of these distributed solutions to use shared and unlicensed spectrum are compared with existing solutions. Conditions under which the performance of these schedulers approach the theoretical upper bound, which is the performance possible with no interference among the operators sharing the spectrum, are presented; 5) The ability of a base station to use its own user equipment as sensors, for optimal spectrum sharing with base stations in other operator networks, is demonstrated to be an effective approach.

5G↗

NASA NTRS1985

Robust detection-isolation-accommodation for sensor failures

The results of a one year study to: (1) develop a theory for Robust Failure Detection and Identification (FDI) in the presence of model uncertainty, (2) develop a design methodology which utilizes the robust FDI ththeory, (3) apply the methodology to a sensor FDI problem for the F-100 jet engine, and (4) demonstrate the application of the theory to the evaluation of alternative FDI schemes are presented. Theoretical results in statistical discrimination are used to evaluate the robustness of residual signals (or parity relations) in terms of their usefulness for FDI. Furthermore, optimally robust parity relations are derived through the optimization of robustness metrics. The result is viewed as decentralization of the FDI process. A general structure for decentralized FDI is proposed and robustness metrics are used for determining various parameters of the algorithm.

Weiss, J. L.↗

DOE OSTI2024

Towards Agentic AI on Particle Accelerators

As particle accelerators grow in complexity, traditional control methods face increasing challenges in achieving optimal performance. This paper envisions a paradigm shift: a decentralized multi-agent framework for accelerator control, powered by Large Language Models (LLMs) and distributed among autonomous agents. We present a proposition of a self-improving decentralized system where intelligent agents handle high-level tasks and communication and each agent is specialized control individual accelerator components. This approach raises some questions: What are the future applications of AI in particle accelerators? How can we implement an autonomous complex system such as a particle accelerator where agents gradually improve through experience and human feedback? What are the implications of integrating a human-in-the-loop component for labeling operational data and providing expert guidance? We show two examples, where we demonstrate viability of such architecture.

43 PARTICLE ACCELERATORS↗

NASA NTRS1993

Application of controller partitioning optimization procedure to integrated flight/propulsion control design for a STOVL aircraft

A parameter optimization framework has earlier been developed to solve the problem of partitioning a centralized controller into a decentralized, hierarchical structure suitable for integrated flight/propulsion control implementation. This paper presents results from the application of the controller partitioning optimization procedure to IFPC design for a Short Take-Off and Vertical Landing (STOVL) aircraft in transition flight. The controller partitioning problem and the parameter optimization algorithm are briefly described. Insight is provided into choosing various 'user' selected parameters in the optimization cost function such that the resulting optimized subcontrollers will meet the characteristics of the centralized controller that are crucial to achieving the desired closed-loop performance and robustness, while maintaining the desired subcontroller structure constraints that are crucial for IFPC implementation. The optimization procedure is shown to improve upon the initial partitioned subcontrollers and lead to performance comparable to that achieved with the centralized controller. This application also provides insight into the issues that should be addressed at the centralized control design level in order to obtain implementable partitioned subcontrollers.

Garg, Sanjay↗

NASA NTRS1993

Application of controller partitioning optimization procedure to integrated flight/propulsion control design for a STOVL aircraft

Garg, Sanjay↗

DOE OSTI2021

Attention Enabled Multi-Agent DRL for Decentralized Volt-VAR Control of Active Distribution System Using PV Inverters and SVCs

This paper proposes attention enabled multi-agent deep reinforcement learning (MADRL) framework for active distribution network decentralized Volt-VAR control. Using the unsupervised clustering, the whole distribution system can be decomposed into several sub-networks according to the voltage and reactive power sensitivity relationships. Then, the distributed control problem of each sub-network is modeled as Markov games and solved by the improved MADRL algorithm, where each sub-network is modeled as an adaptive agent. An attention mechanism is developed to help each agent focus on specific information that is mostly related to the reward. All agents are centrally trained offline to learn the optimal coordinated Volt-VAR control strategy and executed in a decentralized manner to make online decisions with only local information. Compared with other distributed control approaches, the proposed method can effectively deal with uncertainties, achieve fast decision makings, and significantly reduce the communication requirements. Comparison results with model-based and other data-driven methods on IEEE 33-bus and 123-bus systems demonstrate the benefits of the proposed approach.

distribution network↗

NASA NTRS1992

A parameter optimization approach to controller partitioning for integrated flight/propulsion control application

A parameter optimization framework is presented to solve the problem of partitioning a centralized controller into a decentralized hierarchical structure suitable for integrated flight/propulsion control implementation. The controller partitioning problem is briefly discussed and a cost function to be minimized is formulated, such that the resulting 'optimal' partitioned subsystem controllers will closely match the performance (including robustness) properties of the closed-loop system with the centralized controller while maintaining the desired controller partitioning structure. The cost function is written in terms of parameters in a state-space representation of the partitioned sub-controllers. Analytical expressions are obtained for the gradient of this cost function with respect to parameters, and an optimization algorithm is developed using modern computer-aided control design and analysis software. The capabilities of the algorithm are demonstrated by application to partitioned integrated flight/propulsion control design for a modern fighter aircraft in the short approach to landing task. The partitioning optimization is shown to lead to reduced-order subcontrollers that match the closed-loop command tracking and decoupling performance achieved by a high-order centralized controller.

Schmidt, Phillip↗

NASA NTRS1993

A parameter optimization approach to controller partitioning for integrated flight/propulsion control application

Schmidt, Phillip H.↗

DOE OSTI2022

Decentralized Distribution System Restoration with Grid-Forming/Following Inverter-Based Resources

The high penetration of distributed energy resources (DERs) in active distribution systems has posed challenges to the centralized distribution system restoration (DSR) strategies in current practice. On the other hand, the advancement in smart inverter technologies enables the bottom-up restoration capability. This paper is motivated to develop a 3-layered hierarchical framework for decentralized DSR, based on the grid-forming (GFM) and grid-following (GFL) grid-edge inverters. The first layer presents the tertiary control, which determines the load pickup schedule and generation dispatch of DERs, using the alternating direction method of the multipliers algorithm. The second layer consists of two control functions: GFM control, which regulates voltage and frequency, establishing a stable grid for GFL inverters to follow; and GFL control, which regulates the real and reactive power. In the third layer, the primary control is proposed to regulate the inverter voltage and current, which is developed based on the virtual oscillator control (VOC). Furthermore, the developed framework is tested in the modified IEEE 13-node test feeder. Two scenarios of grid-connected and islanded operating modes are designed, and simulation results demonstrate the effectiveness of decentralized DSR strategies for controlling grid-edge inverters to enhance the distribution system resilience.

24 POWER TRANSMISSION AND DISTRIBUTION↗

DOE OSTI2023

Traffic Signal Optimization by Integrating Reinforcement Learning and Digital Twins

Machine learning (ML) methods, especially reinforcement learning (RL), have been widely considered for traffic signal optimization in intelligent transportation systems. Most of these ML methods are centralized, lacking in scalability and adaptability in large traffic networks. Further, it is challenging to train such ML models due to the lack of training platforms and/or the cost of deploying and training in a real traffic networks. This paper presents an approach for the integration of decentralized graph-based multi-agent reinforcement learning (DGMARL) with a Digital Twin (DT) to optimize traffic signals for the reduction of traffic congestion and network-wide fuel consumption related to stopping. Specifically, the DGMARL agents learn traffic state patterns and make decisions regarding traffic signal control with assistance from a Digital Twin module, which simulates and replicates the traffic behaviors of a real traffic network. The proposed approach was evaluated using PTV-Vissim [1], a microscopic traffic simulation platform. PTV-Vissim is also the simulation engine of the DT, enabling emulation and optimization of the traffic signals on the MLK Smart Corridor in Chattanooga, Tennessee. Compared to an actuated signal control baseline approach, experiment results show that Eco_PI, a developed performance measure capturing the impact of stops on fuel consumption, was reduced by 44.27% in a 24-hour and an average of 29.88% in a PM-peak-hour scenario.

Multi-Agent Reinforcement Learning, Digital Twin, ↗

DOE OSTI2022

Roadmap for Deployment of Modularized Hydrothermal Liquefaction: Understanding the Impacts of Industry Learning, Optimal Plant Scale, and Delivery Costs on Biofuel Pricing

Hydrothermal liquefaction (HTL) is a promising technology for converting abundant organic wastes into fuels. Previous techno-economic analyses (TEAs) of HTL have been used to estimate the minimum fuel selling price (MFSP) of biofuel products, but these analyses often assume a bespoke plant design where each plant operates under unique process conditions and neglect transportation costs. However, transportation costs must be included in realistic TEAs, and further, a mass-produced fixed-scale modular plant design approach may be more effective than case-by-case plant design, provided that there is sufficient market capacity to benefit from modularization. This study estimates fuel price behavior in the presence of transportation costs and benefits stemming from modular plant design. This analysis indicates that a modular process capable of handling 60 dry tons per day (DTPD) is optimal, resulting in a ~25% reduction in MFSP (from $4.70/GGE, fully upgraded) at complete market feedstock utilization compared with case-by-case design. The associated cost reductions are attributable to learning benefits and modularization. Several HTL deployment “roadmaps” are then explored, with each roadmap consisting of different periods of case-by-case design followed by adoption of a modularized approach. A period of nonmodular industry growth up to market saturation of ~7% followed by implementation of modular plant design strikes a balance between the investment risk and learned cost reductions associated with modular plant design. However, if bespoke plants built during this period of nonmodular growth saturate more than 23% of available feedstock, learned cost reductions are significantly diminished. Here, this study points to the potential benefits of modularized and decentralized waste-to-energy processes when the modularization follows an optimal deployment strategy.

09 BIOMASS FUELS↗

DOE OSTI2022

DLMP of Competitive Markets in Active Distribution Networks: Models, Solutions, Applications, and Visions

Traditionally, the electric distribution system operates with uniform energy prices across all system nodes. However, as the adoption of distributed energy resources (DERs) propels a shift from passive to active distribution network (ADN) operation, a distribution-level electricity market has been proposed to manage new complexities efficiently. In addition, distribution locational marginal price (DLMP) has been established in the literature as the primary pricing mechanism. The DLMP inherits the LMP concept in the transmission-level wholesale market but incorporates characteristics of the distribution system, such as high $R/X$ ratios and power losses, system imbalance, and voltage regulation needs. The DLMP provides a solution that can be essential for competitive market operation in future distribution systems. This article first provides an overview of the current distribution-level market architectures and their early implementations. Next, the general clearing model, model relaxations, and DLMP formulation are comprehensively reviewed. The state-of-the-art solution methods for distribution market clearing are summarized and categorized into centralized, distributed, and decentralized methods. Then, DLMP applications for the operation and planning of DERs and distribution system operators (DSOs) are discussed in detail. Finally, visions of future research directions and possible barriers and challenges are presented.

42 ENGINEERING↗

NASA NTRS2004

A Decision-Theoretic Approach to Autonomous Planetary Rover Control

The report discusses the: Decentralized Control of Markov Decision Processes. Study the complexity of decentralized control of Markov decision processes, and develop algorithms for finding optimal control policies. Scheduling Contract Algorithms. Develop an optimal method for scheduling runs of a contract anytime algorithm (one that takes the deadline as input) in situations where the deadline is unknown, multiple problem instances must be solved, and a multi-processor machine is available. Planetary Rover Control as a Markov Decision Process.Use the Markov decision process framework to formalize and solve problems in planetary rover control. Adaptive Peer Selection. Use reinforcement learning to maximize the expected down-load speed for a client in a peer-to-peer file sharing system.

Zilberstein, Shlomo↗

NASA NTRS2004

Masked Proportional Routing

Masked proportional routing is an improved procedure for choosing links between adjacent nodes of a network for the purpose of transporting an entity from a source node ("A") to a destination node ("B"). The entity could be, for example, a physical object to be shipped, in which case the nodes would represent waypoints and the links would represent roads or other paths between waypoints. For another example, the entity could be a message or packet of data to be transmitted from A to B, in which case the nodes could be computer-controlled switching stations and the links could be communication channels between the stations. In yet another example, an entity could represent a workpiece while links and nodes could represent, respectively, manufacturing processes and stages in the progress of the workpiece towards a finished product. More generally, the nodes could represent states of an entity and the links could represent allowed transitions of the entity. The purpose of masked proportional routing and of related prior routing procedures is to schedule transitions of entities from their initial states ("A") to their final states ("B") in such a manner as to minimize a cost or to attain some other measure of optimality or efficiency. Masked proportional routing follows a distributed (in the sense of decentralized) approach to probabilistically or deterministically choosing the links. It was developed to satisfy a need for a routing procedure that 1. Does not always choose the same link(s), even for two instances characterized by identical estimated values of associated cost functions; 2. Enables a graceful transition from one set of links to another set of links as the circumstances of operation of the network change over time; 3. Is preferably amenable to separate optimization of different portions of the network; 4. Is preferably usable in a network in which some of the routing decisions are made by one or more other procedure(s); 5. Preferably does not cause an entity to visit the same node twice; and 6. Preferably can be modified so that separate entities moving from A to B do not arrive out of order.

Wolpert, David↗

DOE OSTI2026

Signal Whisperers: Enhancing Wireless Reception Using DRL-Guided Reflector Arrays

This paper presents a multi-agent reinforcement learning (MARL) approach for controlling adjustable metallic reflector arrays to enhance wireless signal reception in non-line-of-sight (NLOS) scenarios. Unlike conventional reconfigurable intelligent surfaces (RIS) that require complex channel estimation, our system employs a centralized training with decentralized execution (CTDE) paradigm where individual agents corresponding to reflector segments autonomously optimize reflector element orientation in three-dimensional space using spatial intelligence based on user location information. Through extensive ray-tracing simulations with dynamic user mobility, the proposed multi-agent beam-focusing framework demonstrates substantial performance improvements over single-agent reinforcement learning baselines, while maintaining rapid adaptation to user movement within one simulation step. Comprehensive evaluation across varying user densities and reflector configurations validates system scalability and robustness. The results demonstrate the potential of learning-based approaches for adaptive wireless propagation control.

deep reinforcement learning↗

NASA NTRS1996

Use of the Collaborative Optimization Architecture for Launch Vehicle Design

Collaborative optimization is a new design architecture specifically created for large-scale distributed-analysis applications. In this approach, problem is decomposed into a user-defined number of subspace optimization problems that are driven towards interdisciplinary compatibility and the appropriate solution by a system-level coordination process. This decentralized design strategy allows domain-specific issues to be accommodated by disciplinary analysts, while requiring interdisciplinary decisions to be reached by consensus. The present investigation focuses on application of the collaborative optimization architecture to the multidisciplinary design of a single-stage-to-orbit launch vehicle. Vehicle design, trajectory, and cost issues are directly modeled. Posed to suit the collaborative architecture, the design problem is characterized by 5 design variables and 16 constraints. Numerous collaborative solutions are obtained. Comparison of these solutions demonstrates the influence which an priori ascent-abort criterion has on development cost. Similarly, objective-function selection is discussed, demonstrating the difference between minimum weight and minimum cost concepts. The operational advantages of the collaborative optimization

Braun, R. D.↗

DOE OSTI2021

A stochastic biomass blending problem in decentralized supply chains

Blending biomass materials of different physical or chemical properties provides an opportunity to adjust the quality of the feedstock to meet the specifications of the conversion platform. We propose a model which identifies the right mix of biomass to optimize the performance of the thermochemical conversion process at the mini-mum cost. This is a chance-constraint programming (CCP) model which takes into account the stochastic nature of biomass quality. The proposed CCP model ensures that process requirements, which are impacted by physical and chemical properties of biomass, are met most of the time. We consider two problem settings, a centralized and a decentralized supply chain. We propose a mixed-integer linear program to model the blending problem in the centralized setting and a bilevel program to model the blending problem in the decentralized setting. We use the sample average approximation method to approximate the chance constraints, and propose solution algorithms to solve this approximation. We develop a case study for South Carolina using data provided by the Billion Ton Study. Based on our results, the blends identified consist mainly of pine and softwood residues. The blends identified and the suppliers selected by both models are different. The cost of the centralized supply chain is 2%–6% lower. The implications of these results are twofold. First, these results could lead to improved collaborations in the supply chain. Second, these results provide an estimate of the approximation error from assuming centralized decision making in the supply chain.

09 BIOMASS FUELS↗

NASA NTRS1994

A mission planning concept and mission planning system for future manned space missions

The international character of future manned space missions will compel the involvement of several international space agencies in mission planning tasks. Additionally, the community of users requires a higher degree of freedom for experiment planning. Both of these problems can be solved by a decentralized mission planning concept using the so-called 'envelope method,' by which resources are allocated to users by distributing resource profiles ('envelopes') which define resource availabilities at specified times. The users are essentially free to plan their activities independently of each other, provided that they stay within their envelopes. The new developments were aimed at refining the existing vague envelope concept into a practical method for decentralized planning. Selected critical functions were exercised by planning an example, founded on experience acquired by the MSCC during the Spacelab missions D-1 and D-2. The main activity regarding future mission planning tasks was to improve the existing MSCC mission planning system, using new techniques. An electronic interface was developed to collect all formalized user inputs more effectively, along with an 'envelope generator' for generation and manipulation of the resource envelopes. The existing scheduler and its data base were successfully replaced by an artificial intelligence scheduler. This scheduler is not only capable of handling resource envelopes, but also uses a new technology based on neuronal networks. Therefore, it is very well suited to solve the future scheduling problems more efficiently. This prototype mission planning system was used to gain new practical experience with decentralized mission planning, using the envelope method. In future steps, software tools will be optimized, and all data management planning activities will be embedded into the scheduler.

Wickler, Martin↗