Search NASA⌕ Search

SEARCH · Search NASA

Results for “Differentiable 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 19 records

Learning Constrained Parametric Differentiable Predictive Control Policies With Guarantees

We present differentiable predictive control (DPC), a method for offline learning of constrained neural control policies for nonlinear dynamical systems with performance guarantees. We show that the sensitivities of the parametric optimal control problem can be used to obtain direct policy gradients. Specifically, we employ automatic differentiation (AD) to efficiently compute the sensitivities of the model predictive control (MPC) objective function and constraints penalties. To guarantee safety upon deployment, we derive probabilistic guarantees on closed-loop stability and constraint satisfaction based on indicator functions and Hoeffding’s inequality. We empirically demonstrate that the proposed method can learn neural control policies for various parametric optimal control tasks. In particular, we show that the proposed DPC method can stabilize systems with unstable dynamics, track time-varying references, and satisfy nonlinear state and input constraints. Our DPC method has practical time savings compared to alternative approaches for fast and memory-efficient controller design. Specifically, DPC does not depend on a supervisory controller as opposed to approximate MPC based on imitation learning. We demonstrate that, without losing performance, DPC is scalable with greatly reduced demands on memory and computation compared to implicit and explicit MPC while being more sample efficient than model-free reinforcement learning (RL) algorithms.

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↗

Neural Lyapunov Differentiable Predictive Control

We present a learning-based predictive control methodology using the differentiable programming framework with sufficient Lyapunov-based guarantees. The neural Lyapunov differentiable predictive control (NLDPC) learns the policy by constructing a computational graph encompassing the system dynamics, state, and input constraints, and the necessary Lyapunov certification constraints, and thereafter using the automatic differentiation to update the neural policy parameters. In conjunction, our approach jointly learns a Lyapunov function that certifies the regions of state-space with stable dynamics. We also provide an initial condition sampling-based statistical guarantee for the training of NLDPC. Our offline training approach provides a computationally efficient and scalable alternative to classical model predictive control solutions, which now directly comes with Lyapunov based guarantees as shown in this paper. We substantiate the advantages of the proposed approach with simulations to stabilize the double integrator model and on an example of controlling an aircraft model.

Mukherjee, Sayak↗

Differentiable Predictive Control with Safety Guarantees: A Control Barrier Function Approach

In this paper, we develop a novel form of differentiable predictive control (DPC) with safety and robustness guarantees. DPC is a form of approximate model predictive control (MPC), wherein the control policy is a neural network that learns a receding horizon, optimal control law. The proposed approach exploits a new form of sampled-data barrier function to enforce safety, while only interrupting the neural network-based controller near the boundary of the safe set. The effectiveness of the proposed approach is demonstrated in simulation.

Shaw Cortez, Wenceslao E.↗

Learning Stochastic Parametric Differentiable Predictive Control Policies

We present a scalable unsupervised learning-based method for obtaining explicit control policies for model predictive control problems for stochastic linear systems with additive uncertainties subject to nonlinear chance constraints. We call the proposed method stochastic parametric differentiable predictive control (SP-DPC), which extends the recently proposed deterministic DPC policy optimization algorithm. We formulate the SP-DPC as a deterministic approximation to the stochastic parametric constrained optimal control problem via independent sampling of the problem's parameters and uncertainties. This formulation allows us to directly compute the policy gradients via automatic differentiation of the problem's value function, evaluated over sampled parameters and uncertainties. In particular, the computed expectation of the problem's value function is backpropagated through the finite-time closed-loop system rollouts parametrized by a known nominal system dynamics model and neural control policy. We also provide theoretical probabilistic guarantees on closed-loop stability and chance constraints satisfaction for systems controlled by learned neural policies. We demonstrate the computational efficiency and scalability of the proposed policy optimization algorithm in three numerical examples, including systems with a large number of states or subject to nonlinear constraints.

Drgona, Jan↗

Koopman-based Differentiable Predictive Control for the Dynamics-Aware Economic Dispatch Problem

The dynamics-aware economic dispatch (DED) problem embeds low-level generator dynamics and operational constraints to enable near real-time scheduling of generation units in a power network. DED produces a more dynamic supervisory control policy than traditional economic dispatch (T-ED) that reduces overall generation costs. However, the incorporation of differential equations that govern the system dynamics makes DED an optimization problem that is computationally prohibitive to solve. In this work, we present a new data-driven approach based on differentiable programming to efficiently obtain offline parametric solutions to the underlying DED problem. In particular, we employ the recently proposed differentiable predictive control (DPC) for offline learning of explicit neural control policies based on identified Koopman operator (KO) model of the system dynamics. We demonstrate the high solution quality and five orders of magnitude computational-time savings of the DPC method over the original optimization-based DED approach on a 9-bus test power grid network.

King, Ethan↗

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↗

Dynamic Decarbonization through Autonomous Physics-Centric Deep Learning and Optimization of Building Operations (Abstract only)

This project directly addresses the primary goal of Area of Interest 2 in the CRADA call: to advance optimization-based integrated energy management systems in commercial and residential buildings. Pacific Northwest National Laboratory (PNNL) and its industry partner PassiveLogic aim to accomplish this by reaching three key objectives. First, to ensure a broad impact in the building controls industry, PNNL will extend its open-source library for predictive control synthesis by augmenting its capabilities with data-driven self-learning of building models and auto-calibration of predictive controllers. The effort will focus on building use cases selected in collaboration with PassiveLogic. The team will specifically address the development of methods for data-driven adaptation of building models, investigation of model architectures that best address specific building types, and automated synthesis of differentiable predictive controllers that optimize diverse objectives. Second, PNNL will collaborate with PassiveLogic to integrate the aforementioned methods with PasiveLogic’s advanced controls platform. The collaborative integration effort will inform the developments under the first objective by providing specific data on the attainable performance of model learning on resource-constrained edge computing platforms. This software integration effort will increase the technical maturity of the developed libraries by exploring the use of software integration tools and methods. Third, PNNL and PassiveLogic will work to improve the technology readiness of the developed predictive controllers by testing their performance in relevant test environments, such as high-fidelity simulation, hardware in the loop, and actual test buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

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↗

Computational model for predicting the performance of fluid-powered longiudinal piston modules and motors.

A Matlab-based computational model line code has been developed for predicting the performance of fluid-powered longitudinal piston modules and motors. A system of equations has been developed based upon applying the conservation of mass to a control volume defined by a reciprocating piston. The method results in a differential equation that predicts the pressure gradient in the control volume as a function of rotor rotation angle. The model uses an integration scheme to predict the pressure state in the control volume. The orifice equation is used to predict the flowrate in and out of the control volume. From the resulting solution the torque and speed characteristics of the motor configuration can be predicted, optimized, or tailored for a specific requirement.SAND2019-13983 M Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Taylor, Michael↗

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↗

Multi-mode evaluation of power-maximizing cross-flow turbine controllers

A general method for predicting and evaluating the performance of three candidate cross-flow turbine power-maximizing controllers is presented in this paper using low-order dynamic simulation, scaled laboratory experiments, and full-scale field testing. For each testing mode and candidate controller, performance metrics quantifying energy capture (ability of a controller to maximize power), variation in torque and rotation rate (related to drive train fatigue), and variation in thrust loads (related to structural fatigue) are quantified for two purposes. First, for metrics that could be evaluated across all testing modes, we considered the accuracy with which simulation or laboratory experiments could predict performance at full scale. Second, we explored the utility of these metrics to contrast candidate controller performance. For these turbines and set of candidate controllers, energy capture was found to only differentiate controller performance in simulation, while the other explored metrics were able to predict performance of the full-scale turbine in the field with various degrees of success. Finally, effects of scale between laboratory and full-scale testing are considered, along with recommendations for future improvements to dynamic simulations and controller evaluation.

13 HYDRO ENERGY↗

NSLS-II Control of Dynamic Aperture

We have outlined how, by an intuitive approach, the on- and off-momentum dynamic aperture for a synchrotron light source can be estimated from a nonlinear system of algebraic equations for the sextupole/multipole strengths. The approach has only two free parameters: the relative weight for resonance vs. tune shift terms and the tune footprint for stable trajectories in a modern third generation synchrotron light source. In other words, we have established a control theory approach for the medium term (10{sup 3} turns) stability for a dynamic system described by a nonlinear system of ordinary differential equations. Equipped with a predictive, quantitative model for stability, we have then evaluated how to improve the control of the dynamics by analyzing and modifying the properties of the corresponding algebraic system. In particular, by changing the number- and characteristics of the parameters, i.e., we have not evaluated how the underlying (linear) optics could be improved. We have also validated our conjectures by numerical simulations with a realistic model. Presumably, our conclusions, summarized in Section 1.0, are a direct result of the presented analysis and observations.

43 PARTICLE ACCELERATORS↗