Toward data-driven predictive control of multi-energy distribution systems

Bilgic, Deborah; Koch, Alexander; Pan, Guanru; Faulwasser, Timm

doi:10.1016/j.epsr.2022.108311

Cited by 19 publications

(8 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An interesting observation is R proj (g * pinv ) = 0 and thus the SPC solution (9) will not be penalized by (14). This assures consistency, that is, under perfect data there is no bias in the regularized solution to (13).…”

Section: Remedies For Stochastic Lti Systemsmentioning

confidence: 99%

“…6) SPC [36], [37]: The classical SPC scheme as described in (12). 7) PBR [24]: Regularized DeePC in (13) with PBR (14), where λ is optimally selected from a grid of values within [10 −3 , 10 3 ].…”

Section: Online Predictive Control Performancementioning

confidence: 99%

“…Besides its easy implementation, there is an additional subtlety that the bias induced by inadequate model fitting can be effectively circumvented [10], leading to competitive performance with model predictive control (MPC). Thanks to these merits, DeePC has found miscellaneous applications across domains, such as power systems [11]- [14], motion control [15]- [17], and smart buildings [18]. On the theoretical side, rigorous analysis of closed-loop stability of DeePC was given by [19], [20].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Data-Driven Control of Stochastic Systems: An Innovation Estimation Approach

Wang¹,

Chen²,

Huang³

2022

Preprint

View full text Add to dashboard Cite

Recent years have witnessed a booming interest in the data-driven paradigm for predictive control. However, under noisy data ill-conditioned solutions could occur, causing inaccurate predictions and unexpected control behaviours. In this article, we explore a new route toward data-driven control of stochastic systems through active offline learning of innovation data, which gives an answer to the critical question of how to derive an optimal data-driven model from a noise-corrupted dataset. A generalization of the Willems' fundamental lemma is developed for non-parametric representation of input-outputinnovation trajectories, provided realizations of innovation are precisely known. This yields a model-agnostic unbiased output predictor and paves the way for data-driven receding horizon control, whose behaviour is identical to the "oracle" solution of certainty-equivalent model-based control with measurable states. For efficient innovation estimation, a new low-rank subspace identification algorithm is developed. Numerical simulations show that by actively learning innovation from input-output data, remarkable improvement can be made over present formulations, thereby offering a promising framework for data-driven control of stochastic systems.

show abstract

Section: Remedies For Stochastic Lti Systemsmentioning

confidence: 99%

Section: Online Predictive Control Performancementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Data-Driven Control of Stochastic Systems: An Innovation Estimation Approach

Wang¹,

Chen²,

Huang³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Given the challenges faced by model-based control, the idea of Data-Enabled Predictive Control (DeePC) [Coulson et al (2019)] has emerged in recent years. It has been adopted for the control of distribution networks [Bilgic et al (2022)], power plants [Huang et al (2021); Mahdavipour et al (2022)], and synchronous motor drives [Carlet et al (2022)]. Unlike MPC, DeePC is model-free, as it constructs a non-parametric representation of the system dynamics directly from raw data, which are a collection of input/output trajectories.…”

Section: Data-driven Predictive Controlmentioning

confidence: 99%

Data-Driven Load-Current Sharing Control for Multi-Stack Fuel Cell System with Circulating Current Mitigation

Xu¹,

Guo²,

Dong³

et al. 2023

Preprint

View full text Add to dashboard Cite

The global trend toward renewable power generation has drawn great attention to hydrogen Fuel Cells (FCs), which have a wide variety of applications, from utility power stations to laptops. The Multi-stack Fuel Cell System (MFCS), which is an assembly of FC stacks, can be a remedy for obstacles in high-power applications. However, the output voltage of FC stacks varies dramatically under variable load conditions; hence, in order for MFCS to be efficiently operated and guarantee an appropriate load-current sharing among the FC stacks, advanced converter controllers for power conditioning need to be designed. An accurate circuit model is essential for controller design, which accounts for the fact that the parameters of some converter components may change due to aging and repetitive stress in long-term operations. Existing control frameworks and parametric and non-parametric system identification techniques do not consider the aforementioned challenges. Thus, this paper investigates the potential of a data-driven method that, without system identification, directly implements control on paralleled converters using raw data. Based on pre-collected input/output trajectories, a nonparametric representation of the overall circuit is produced for implementing predictive control. While approaching equal current sharing within the MFCS, the proposed method considers the minimization of load-following error and mitigation of circulating current between the converters. Simulation results verify the effectiveness of the proposed method.

show abstract

“…In some power systems applications, forecasting can enhance the controller for better performance [17]. This paper extends forecasting for the M-ENMPC by imposing a certainty horizon N C to exploit forecasting information, which is often accurate for a short time horizon, see Fig.…”

Section: B Proposed Controller -Multi-stage Approachmentioning

confidence: 99%

Reference optimisation of uncertain offshore hybrid power systems with multi-stage nonlinear model predictive control *

Hoang

Knudsen

Imsland

2023

2023 American Control Conference (ACC)

View full text Add to dashboard Cite

This paper presents a modified multi-stage economic nonlinear model predictive controller (M-ENMPC) for reference optimisation of isolated, uncertain offshore hybrid power systems (OHPSs). These systems require control strategies that can handle significant stochastic disturbances in exogenous power demand and wind, given uncertain forecasts of the disturbances. An M-ENMPC modified with a certainty horizon is formulated to hande uncertain forecasts of these disturbances for reference optimisation of OHPSs. The certainty horizon models the increase in uncertainty of forecasts with time to decrease the cost in the M-ENMPC. Monte Carlo simulations with different realisations of the considered disturbances show that explicitly considering scenarios of the disturbances with the M-ENMPC can decrease greenhouse gas (GHG) emissions by operating the gas turbines in the hybrid power system more efficiently while achieving an acceptable satisfaction of the exogenous power demand. Furthermore, the Monte Carlo simulations show that using the modified M-ENMPC decreases the average computational time by 17% compared with the conventional M-ENMPC from literature.

show abstract

Toward data-driven predictive control of multi-energy distribution systems

Cited by 19 publications

References 14 publications

Data-Driven Control of Stochastic Systems: An Innovation Estimation Approach

Data-Driven Control of Stochastic Systems: An Innovation Estimation Approach

Data-Driven Load-Current Sharing Control for Multi-Stack Fuel Cell System with Circulating Current Mitigation

Reference optimisation of uncertain offshore hybrid power systems with multi-stage nonlinear model predictive control *

Contact Info

Product

Resources

About