Multiscale model predictive control of battery systems for frequency regulation markets using physics-based models

Cao, Yankai; Lee, Seong Beom; Subramanian, Venkat R.; Zavala, Ví­ctor M.

doi:10.1016/j.jprocont.2020.04.001

Cited by 19 publications

(24 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Following the results given by Cao et al 11 and Moura et al, 17 we assume a side reaction leading to the formation of a resistive SEI film in the negative electrode which causes the loss of active material and battery capacity fade. The side reaction and the SEI film growth are governed by the following equations:…”

Section: Single Particle Battery Modelmentioning

confidence: 99%

“…The voltage V , current I app , power P , and energy E of the battery are computed as therein. 8,11 The above SP model provides a high-fidelity prediction of battery states and captures the capacity fade incurred by charging/discharging cyclings. This SP model will be utilized in our paper to serve as a high-fidelity battery simulator for simulating and testifying our proposed FR policies.…”

Section: Single Particle Battery Modelmentioning

confidence: 99%

“…The MPC-based FR strategy aims at determining the optimal commitments F t , O t and L t at each hour by maximizing the objective function (5) while guaranteeing constraints ( 6)- (11) and the satisfaction of battery models. does not reflect the evolution of battery degradation, and consequently, the computed FR commitment decisions can be aggressive and myopic.…”

Section: Low-fidelity Mpc Strategymentioning

confidence: 99%

“…where the last term is to force the battery SOC at the end of each hour close to half of the remaining capacity for the preparation of next hour FR participation. The parameter π C f , which represents the long-term valuation of battery capacity, is set as $12, 000 based on the study of Cao et al 11. Since the electricity prices follow a weekly pattern, we select one week as one episode.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Deep Learning-based Predictive Control of Battery Management for Frequency Regulation

Li¹,

Chen²,

Hua³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

This paper proposes a deep learning-based optimal battery management scheme for frequency regulation (FR) by integrating model predictive control (MPC), supervised learning (SL), reinforcement learning (RL), and high-fidelity battery models. By taking advantage of deep neural networks (DNNs), the derived DNN-approximated policy is computationally efficient in online implementation. The design procedure of the proposed scheme consists of two sequential processes: (1) the SL process, in which we first run a simulation with an MPC embedding a low-fidelity battery model to generate a training data set, and then, based on the generated data set, we optimize a DNN-approximated policy using SL algorithms; and (2) the RL process, in which we 1

show abstract

Section: Single Particle Battery Modelmentioning

confidence: 99%

Section: Single Particle Battery Modelmentioning

confidence: 99%

Section: Low-fidelity Mpc Strategymentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Deep Learning-based Predictive Control of Battery Management for Frequency Regulation

Li¹,

Chen²,

Hua³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Subramanian [129,131,132,[165][166][167][168][169][170][171] has spent decades focused on modeling methods, and has more recently focused on reformulating [129,131,132,169] models for increased computational efficiency, developing battery management systems [168,171] using physics based reformulated models, and applying nonlinear model predictive control [170,171] to Li ion batteries. Subramanian's work has illustrated the tradeoff between computational complexity and accuracy, the need for reduced order models in production in both the automotive and grid storage space, and has provided ample proof that physics based models are the next step in bridging the gap between scientific understanding of aging phenomena and the tools that are used for controls every day.…”

Section: Plating Modelingmentioning

confidence: 99%

Lithium Plating Detection Methods in Lithium-Ion Batteries

2020

View full text Add to dashboard Cite

Lithium-ion batteries provide a low-cost, long cycle-life and high energy density solution to the expediting energy requirement of the automotive industry. There is a growing need of fast charging batteries for commercial application. However, large charging currents may cause Lithium plating which describes the deposition of metallic Lithium at the anode surface. It takes place at conditions of high currents and/or low temperatures because of kinetic limitations. The main reason behind plating is the slow solid-state diffusion of Lithium ions inside the active material. If the anode surface potential falls below 0 V versus Li/Li+, the formation of metallic Lithium is thermodynamically feasible. To avoid or reduce the amount Lithium plating, it is essential to detect its onset during a charging event. Determination of accurate Lithium plating curve is crucial in estimating the boundary conditions for battery operation without compromising life and safety. There are various data analysis methods involved in deriving the Lithium plating curve: anode potential using a three-electrode cell, variation of relaxation voltage after charging (dV/dt), variation of accumulated charge with voltage (dQ/dV) and coulombic efficiency of charge/discharge with %SOC (state of charge) are more commonly employed techniques. In addition to these methods, estimation of its occurrence is typically underpinned by electrochemical models where the negative (anode) electrode potential is expressed by a set of partial differential equations based on the electrochemical and physical properties of the battery components such as the electrodes and electrolyte. The present paper reviews the common test methods and analysis that are currently utilized in Lithium plating determination. Knowledge gaps are identified, and recommendations are made for the future development in the determination and verification of Lithium plating curve in terms of modelling and analysis.

show abstract

A robust MPC design concerning on battery variables for frequency regulation and saving battery life collaborating with demand response for a multi-source integrated power system

2023

View full text Add to dashboard Cite

Multiscale model predictive control of battery systems for frequency regulation markets using physics-based models

Cited by 19 publications

References 33 publications

Deep Learning-based Predictive Control of Battery Management for Frequency Regulation

Deep Learning-based Predictive Control of Battery Management for Frequency Regulation

Lithium Plating Detection Methods in Lithium-Ion Batteries

A robust MPC design concerning on battery variables for frequency regulation and saving battery life collaborating with demand response for a multi-source integrated power system

Contact Info

Product

Resources

About