Search NASA⌕ Search

SEARCH · Search NASA

Results for “Model predictive control”

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 469 records · Page 26

Predictive drug dosage control through a Fokker–Planck observer

During the years, various techniques have been presented for optimal control of tumor growth for which its stochastic behavior has rarely been considered. This article uses the well-known Gompertz stochastic model, for describing tumor growth, and by receding horizon model predictive control (RH-MPC) scheme computes the drug dosage for minimizing the tumor-cell population. To do this, a predictive control problem is defined based on the difference between the probability density function of tumor-cell population and the desired probability density function. In the model, both the drug dosage limitation and the Fokker–Planck equation have been considered as constraints. By solving this problem, the drug dosage which is considered as an external input to tumor model, is computed. In this article, the Fokker–Planck equation is used (1) as a nonlinear observer of probability density function of tumor-cell population and (2) a mapping vehicle from stochastic to deterministic domain. The Fokker–Planck equation is written for the Gompertz stochastic model of tumor growth. The equation is then solved through the path integral method. In this way, the probability density function of tumor-cell population, which contains the entire stochastic characteristics of the tumor growth, is obtained for any instance of time. The simulation results have also been presented for the evaluation of the suggested approach. The results show that the tumor-cell population can be controlled within a number of time windows (15 time windows in our case study) if an appropriate desired PDF is selected.

97 MATHEMATICS AND COMPUTING↗

Deep Learning Explicit Differentiable Predictive Control Laws for Buildings

We present a differentiable predictive control (DPC) methodology for learning constrained control laws for unknown nonlinear systems. DPC poses an approximate solution to multiparametric programming problems emerging from explicit nonlinear model predictive control (MPC). Contrary to approximate MPC, DPC does not require supervision by an expert controller. Instead, a system dynamics model is learned from a small dataset of recorded observations of the perturbed system's dynamics and the control law is optimized offline by interaction with the learned system model. The DPC method is based on two sequential steps, i) system identification using a constrained neural state-space model, and ii) optimization of an explicit control law parametrized by another neural network in closed-loop simulation with the identified neural state-space model. The combination of a differentiable closed-loop system and penalty methods for constraint handling of system outputs and inputs allows us to optimize the control law's parameters directly by backpropagating economic MPC loss through the learned system model. By incorporating domain knowledge and leveraging established techniques from optimal control, our method leverages deep neural networks as nonlinear function approximators for system identification and control while avoiding concomitant costs of intractably large datasets, and computationally expensive over-parametrized models. The scalability, data efficiency, and constrained optimal control capability of the proposed DPC method are demonstrated in simulation using a multi-zone building emulator.

Drgona, Jan↗

Machine‐learning‐based predictive control of nonlinear processes. Part II: Computational implementation

Abstract Machine learning is receiving more attention in classical engineering fields, and in particular, recurrent neural networks (RNNs) coupled with ensemble regression tools have demonstrated the capability of modeling nonlinear dynamic processes. In Part I of this two‐article series, the Lyapunov‐based model predictive control (LMPC) method using a single RNN model and an ensemble of RNN models, respectively, was rigorously developed for a general class of nonlinear systems. In the present article, computational implementation issues of this new control method ranging from training of the RNN models, ensemble regression of the RNN models, and parallel computation for accelerating the real‐time model calculations are addressed. Furthermore, a chemical reactor example is used to demonstrate the implementation and effectiveness of these machine‐learning tools in LMPC as well as compare them with standard state‐space model identification tools.

Wu, Zhe↗

An Introduction to the Federated Architecture for Secure and Transactive Distributed Energy Management Solutions (FAST-DERMS)

Deployment and capability of distributed energy resources (DER) in power systems is growing rapidly. These resources present an opportunity for low-cost provision of energy and grid services. The Federal Energy Regulatory Commission recently provided rulings to enable market participation of these distribution-connected resources, but the prevailing strategies for their management may not scale well to meet future needs. This paper introduces the Federated Architecture for Secure and Transactive Distributed Energy Management Solutions (FAST-DERMS) which was designed to address this need. In it we describe the architectural features of the approach, and a reference controls implementation employing a hierarchical coordination that includes stochastic optimization, model predictive control, and a simple real-time management scheme. Sample results from simulation show firm transmission-level service provision measured at the distribution substation.

grid architecture↗

An Introduction to the Federated Architecture for Secure and Transactive Distributed Energy Management Solutions (FAST-DERMS): Preprint

Deployment and capability of distributed energy resources (DER) in power systems is growing rapidly. These resources present an opportunity for low-cost provision of energy and grid services. The Federal Energy Regulatory Commission recently provided rulings to enable market participation of these distribution-connected resources, but the prevailing strategies for their management may not scale well to meet future needs. This paper introduces the Federated Architecture for Secure and Transactive Distributed Energy Management Solutions (FASTDERMS) which was designed to address this need. In it we describe the architectural features of the approach, and a reference controls implementation employing a hierarchical coordination that includes stochastic optimization, model predictive control, and a simple real-time management scheme. Sample results from simulation show firm transmission-level service provision measured at the distribution substation.

DERMS↗

A New Proposal Generalized Predictive Control Algorithm With Polynomial Reference Tracking Applied for Sodium Fast Reactors

This paper proposes a generalized predictive control (GPC) with constraints and orthonormal Laguerre functions using the simplified model of the primary system (reactor core and intermediate heat exchanger (IHX)) of a prototypical sodium fast reactor (SFR). This paper develops a multiple-input multiple-output (MIMO) GPC with input constraints able to track polynomial references of any degree applied in coolant temperature difference across the core and fractional power. The manipulated variables of the GPC-SFR are the reactivity and the sodium flow rate of the primary and secondary pipes. Moreover, orthonormal Laguerre functions and step down condition number techniques were also applied to avoid the numerical ill-conditioning issue in quadratic programming of large systems. Thus, a GPC type-2 was designed to control fractional power, coolant temperature difference across the core and sodium tank temperature of the SFR primary system when temperature references change according to a linear ramp after reaching their steady-state operation, sustaining 100% power operation on the reactor. In order to analyze the load tracking capability of the GPC-SFR type-2, the load following from 100% fractional power (FP) to 60% FP at 0.8% FP/min rate is simulated. Constraints on the rate of coolant temperature difference across the core and reactivity were applied for the design safety. For comparison criteria, this paper compares the GPC-SFR type-2 with the GPC-SFR type-1, i.e., standard model predictive control (MPC), to verify the viability and superior performance of the proposal regarding: (a) ramp-tracking capability of temperature and load; (b) the rejections of a reactivity disturbance of -1 cent and a secondary sodium inlet temperature disturbance of +10°F; and (c) a simulation with uncertainty in reactor design. The simulations show that the GPC-SFR type-2 overcome the GPC-SFR type-1 robustness and performance.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Comparative analysis of model-free and model-based HVAC control for residential demand response

In this paper, we present a comparative analysis of model-free reinforcement learning (RL) and model predictive control (MPC) approaches for intelligent control of heating, ventilation, and air-conditioning (HVAC). Deep-Q-network (DQN) is used as a candidate for model-free RL algorithm. The two control strategies were developed for residential demand-response (DR) HVAC system. We considered MPC as our golden standard to compare DQN's performance. The question we tried to answer through this work was, What % of MPC's performance can be achieved by model-free RL approach for intelligent HVAC control?. Based on our test result, RL achieved an average of ≈ 62% daily cost saving of MPC. Considering the pure optimization and model-based nature of MPC methods, the RL showed very promising performance. We believe that the interpretations derived from this comparative analysis provide useful insights to choose from various DR approaches and further enhance the performance of the RL-based methods for building energy managements.

Kurte, Kuldeep↗

Automated Production of Optimization-Based Control Logics for Dynamic Facade Systems, with Experimental Application to Two-Zone External Venetian Blinds

The primary goal of this research is to devise a system that produces controllers for complex fenestration systems that perform nearly as well as Model Predictive Control but at a level of cost and implementation complexity that rivals simple heuristic controls. To this end, a cloud-based automated controller production system has been set up for a motorized external Venetian blind device, with a simple web interface that can be used by non-experts. The computation cost per controller is in the range of a few dollars, and the control logic is simple enough to be implemented on small and cheap distributed controllers. The web interface allows the user to specify some details of their particular building and window configuration, including orientation, latitude, interior geometries, and lighting and HVAC system parameters. Upon submittal, a cloud-based system configures the necessary files and commands, and then runs thousands of optimizations with them. Once the calculations are finished, the system produces a lookup table and interpolation-based controller scripts that can be used on a simple and cheap distributed controller. This paper describes the underlying models and optimization processes. It also describes the resulting control logics for two cases tested at Lawrence Berkeley National Laboratory’s Advanced Windows Testbed Facility: illuminance maximization subject to glare constraints; and lighting + HVAC energy minimization. The performance of the model-based controllers produced by the automated web-based system are compared to a heuristic ‘block beam’ controller in physical experiments at the Testbed. The experimental results are supplemented by simulation experiments with the same configuration as the Testbed. The results show the illuminance maximizing controller significantly outperforms the heuristic controller in terms of glare avoidance, and also outperforms it in terms of hours of daylight autonomy. The energy minimizing controller also outperforms the heuristic controller. This paper also discusses how the web-based system may be extended to consider other configurations, such as electrochromic windows and thermally massive HVAC systems. Potential roles for this type of system within the building design and construction industry are discussed.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Deep reinforcement learning control of hydraulic fracturing

Hydraulic fracturing is a technique to extract oil and gas from shale formations, and obtaining a uniform proppant concentration along the fracture is key to its productivity. Recently, various model predictive control schemes have been proposed to achieve this objective. But such controllers require an accurate and computationally efficient model which is difficult to obtain given the complexity of the process and uncertainties in the rock formation properties. In this article, we design a model-free data-based reinforcement learning controller which learns an optimal control policy through interactions with the process. Deep reinforcement learning (DRL) controller is based on the Deep Deterministic Policy Gradient algorithm that combines Deep-Q-network with actor-critic framework. In addition, we utilize dimensionality reduction and transfer learning to quicken the learning process. We show that the controller learns an optimal policy to obtain uniform proppant concentration despite the complex nature of the process while satisfying various input constraints.

42 ENGINEERING↗

Gnu-RL: A Precocial Reinforcement Learning Solution for Building HVAC Control Using a Differentiable MPC Policy

Reinforcement learning (RL) was first demonstrated to be a feasible approach to controlling heating, ventilation, and air conditioning (HVAC) systems more than a decade ago. However, there has been limited progress towards a practical and scalable RL solution for HVAC control. While one can train an RL agent in simulation, it is not cost-effective to create a model for each thermal zone or building. Likewise, existing RL agents generally take a long time to learn and are opaque to expert interrogation, making them unattractive for real-world deployment. To tackle these challenges, we propose Gnu-RL: a novel approach that enables practical deployment of RL for HVAC control and re- quires no prior information other than historical data from exist- ing HVAC controllers. To achieve this, Gnu-RL adopts a recently- developed Differentiable Model Predictive Control (MPC) policy, which encodes domain knowledge on planning and system dynam- ics, making it both sample-efficient and interpretable. Prior to any interaction with the environment, a Gnu-RL agent is pre-trained on historical data using imitation learning, which enables it to match the behavior of the existing controller. Once it is put in charge of controlling the environment, the agent continues to improve its policy end-to-end, using a policy gradient algorithm. We evaluate Gnu-RL on both an EnergyPlus model and a real- world testbed. In both experiments, our agents were directly de- ployed in the environment after offline pre-training on expert demonstration. In the simulation experiment, our approach saved 6.6% energy compared to the best published RL result for the same environment, while maintaining a higher level of occupant comfort. Next, Gnu-RL was deployed to control the HVAC of a real-world conference room for a three-week period. Our results show that Gnu-RL saved 16.7% of cooling demand compared to the existing controller and tracked temperature set-point better.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Predicting Initial Trans-Membrane Pressure for Optimized Operations in UF Unit Using Random Forest

With the growing scarcity of freshwater, innovative process design mechanisms like Reverse Osmosis (RO) are increasingly gaining attention among water treatment utilities to address the rising demand. Ensuring reliable water production necessitates efficient resource utilization, minimizing downtime in (ultra-filtration) UF systems. Recent advancements in machine learning (ML) have enabled the development of accurate data-driven models for Model Predictive Control (MPC), often requiring minimal prior knowledge of underlying physical processes. In this study, we present predictive regression models based on Random Forest (RF) and Auto-Regressive (AR) approaches to forecast the initial Trans-Membrane Pressure (TMP) for each filtration cycle in data generated by Direct Potable Reuse (DPR) systems. The proposed RF-based model demonstrates superior performance compared to baseline methods, including historical mean, Last Observation Carried Forward (LOCF), and naïve AR models, across various forecasting horizons in terms of root mean square error (RMSE) metric. To evaluate how different classes of process variables contribute to TMP dynamics over time, we examine the feature importance of independent covariates across multiple forecast horizons. This analysis provides insight into the temporal relevance of operational and sensor-derived features, guiding control and monitoring strategies. Additionally, the impact of hyperparameter tuning on TMP prediction performance is studied for both direct and recursive RF modelling approaches across increasing forecast horizons. Accurate prediction of initial TMP is critical for optimizing RO operations, as it enables the development of robust modelling frameworks by accurately estimating membrane fouling trends, thereby enhancing process efficiency and long-term reliability. The demonstrated efficacy of the RF-based approach highlights its potential as a tool for real-time decision-making in water treatment systems, paving the way for advanced process optimization and sustainable water resource management.

Mukherjee, Subrata [ORNL] (ORCID:0000000309930338)↗

LBC (Learning Building Control)

LBC encompasses the source code and data to reproduce and extend results for a manuscript that compares demand responsive control schemes for multi-zone buildings. It includes examples of model predictive control (MPC), value function approximation via CVXPYLAYERS, and differentiable predictive control (DPC). The goal of the study is to evaluate state-of-the-art controllers and establish the efficacy (if any) of learning-based approaches that leverage deep neural networks in one way or another.

Zhang, Xiangyu↗

EASY-SHIFT v Alpha

The software is a generic, price- and load-responsive control algorithm integrating heat pumps with thermal energy storage. The algorithm leverages simple models of the system and easily accessible data to schedule operation of heat pumps and thermal energy storage in ways that minimize the cost of operating the heating/cooling system. This tool is specifically designed to be easy to interact with, and something that industry partners are able to adopt. There are two current state of the art approaches. Industry tends to develop very simple algorithms, with predetermined schedules that are not capable of changing operation in response to changes in operating environment. For example, a control designed to avoid high-price electricity from 5-8 PM will not be able to adapt if the high-price period changes to 4-9 PM. Academia commonly develops algorithms called Model predictive control (MPC). MPC requires extensive data and highly trained staff to develop a specific type of simulation model of the building, connect the building to optimization algorithms, and leverage powerful computers. Industry, with limited time/finance budgets for any project, is resistant to adopting MPC due to the associated high complexity and cost.

Grant, Peter [Lawrence Berkeley National Laborator↗

Offline Arterial Signal Timing Optimization for Closely Spaced Intersections

The stop-and-go driving behavior at a busy arterial causes a significant amount of fuel waste and emissions that could be reduced. For closely spaced intersections, the traditional bottom-up signal timing approach could fail under high volumes when the queues reach the upstream intersections. A well-coordinated arterial should manage the queue length to prevent the control from failing. Traditionally, coordination among signalized intersections for an arterial means finding the good offsets for each intersection. The control of the intersections along an arterial can also be coordinated by optimizing splits. Splits impact the capacity of an intersection. Similar to the highway, the capacity drop could cause growing queues. This work is based on the Virtual Phase-Link (VPL) model, a street traffic model designed for online traffic model predictive control, to obtain a top-down offline arterial signal timing. The VPL-based model can guarantee capacity consistency in intersections along an arterial. We, therefore, adopted the VPL-based model and developed an offline signal timing optimization approach. The proposed timing derived from the VPL-based offline signal timing optimization showed very good results in simulation. We also collected field experiment data, which demonstrated overall energy reductions and speed improvements on some sections of the study arterial.

47 OTHER INSTRUMENTATION↗

Efficient and Collision-Free Anticipative Cruise Control in Randomly Mixed Strings

Connected intelligent vehicle following can improve safety and efficiency compared to today's road transport, but real traffic in the near future will not provide an ideal setting for its deployment. Unconnected human-driven vehicles will follow variable behavior patterns that combine with differences in dynamic capability to create heterogeneous scenarios. In this paper, connected automated vehicles employ model predictive control for following in traffic that may include both heavy and passenger vehicles at quasi-random positions. Adding further realism, some quasi-randomly mixed vehicles are completely unconnected and equipped with the reactive intelligent driver model using pseudorandom parameters. A mixed-integer quadratic programming formulation adapts the predictive algorithm to diverse powertrain operating point constraints. The preceding vehicle's control input is estimated from velocity and brake light observations and used to probabilistically generate a preview for the ego vehicle. Here, a terminal constraint designed using particle kinematics prevents collisions due to shortsightedness. Results are simulated at various heavy and predictive vehicle concentrations in the presence of packet loss. Linear fuel economy improvements between 1.4% and 1.9% per 10% increase in predictive vehicle penetration rate are shown.

33 ADVANCED PROPULSION SYSTEMS↗

Preliminary Development of Heat Transfer Model-Based Control Algorithms of Liquid Sodium Purification System: Advanced Sensors and Instrumentation Advanced Controls

Monitoring the operation of sodium purification system is essential for efficient operation of sodium fast reactors. In this work, a heat transfer model has been developed for monitoring the plugging meter and cold trap systems at the Mechanisms Engineering Test Loop (METL) liquid sodium facility at Argonne National Laboratory. The model of the purification system was developed by treating the respective aspects of the cold trap purification loop and plugging meter diagnostic loop as two separate control volumes using information from the METL piping and instrumentation diagram (P&ID). A model predictive controller was designed using first order differential equations with the specified boundary conditions. The system behavior was studied with a tuned optimized procedure using the internal cold trap temperature and plugging meter outlet temperature as control variables, and the air blower temperature as an independent variable respectively. Results of computer simulations obtained in this study compared favorably with experimental data showing very good reference tracking response with negligible overshoot as both plugging meter and cold trap physical models approach the setpoint.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Coordination of Energy Storage and Distributed Generation for Voltage Control and Peak-valley Filling

The increasing penetration of distributed energy resources (DERs) in distribution network (DN) poses challenges on voltage control. In addition, the growing integration of DERs also reshapes the traditional load profile. To comprehensively address the voltage control and peak-valley filling in DN, this paper proposes a model predictive control (MPC) based optimization framework. The proposed method can achieve coordinated voltage control and peak-valley filling by adjusting the reactive power output from distributed generations (DGs) and the charging/discharging of energy storage systems (ESS). The performance of the proposed method is demonstrated by simulations on a modified IEEE-123 bus system.

Zhang, Zhengfa [University of Tennessee, Knoxville↗