Data-Driven Screening of Network Constraints for Unit Commitment

Pineda, Salvador; Morales, Juan M.; Jiménez-Cordero, Asunción

doi:10.1109/tpwrs.2020.2980212

Cited by 63 publications

(44 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We initialize the bounds using interval arithmetic (for details see [15]). Then, to compute tighter bounds, we minimize and maximize the output of each neuron z i k subject to the linear relaxation of the trained neural network (13), ( 14), ( 16)- (21), and subject to the restricted input domain in (11). Note that for the linear relaxation only we relax the binary variables b k to continuous variables between 0 and 1.…”

Section: A Mixed-integer Reformulation Of Trained Neural Networkmentioning

confidence: 99%

“…The focus of this work is on obtaining guarantees for machine learning approaches such as the ones in [5]- [9], which predict solutions to OPF problems and replace the use of conventional optimization solvers. These approaches can result to larger computational speed-ups compared to predicting inactive constraints [11] or warm-start points [12] that could accelerate conventional optimization solvers. The work in [5] trains neural networks to directly predict the solution to DC-OPF problems, achieving a speed-up of two orders of magnitude (i.e., 100 times faster).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Learning Optimal Power Flow: Worst-Case Guarantees for Neural Networks

Venzke

Low

et al. 2020

2020 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm)

View full text Add to dashboard Cite

This paper introduces for the first time a framework to obtain provable worst-case guarantees for neural network performance, using learning for optimal power flow (OPF) problems as a guiding example. Neural networks have the potential to substantially reduce the computing time of OPF solutions. However, the lack of guarantees for their worst-case performance remains a major barrier for their adoption in practice. This work aims to remove this barrier. We formulate mixed-integer linear programs to obtain worst-case guarantees for neural network predictions related to (i) maximum constraint violations, (ii) maximum distances between predicted and optimal decision variables, and (iii) maximum sub-optimality. We demonstrate our methods on a range of PGLib-OPF networks up to 300 buses. We show that the worst-case guarantees can be up to one order of magnitude larger than the empirical lower bounds calculated with conventional methods. More importantly, we show that the worst-case predictions appear at the boundaries of the training input domain, and we demonstrate how we can systematically reduce the worst-case guarantees by training on a larger input domain than the domain they are evaluated on.Index Terms-Neural networks, mixed-integer linear programming, optimal power flow.

show abstract

Section: A Mixed-integer Reformulation Of Trained Neural Networkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Learning Optimal Power Flow: Worst-Case Guarantees for Neural Networks

Venzke

Low

et al. 2020

2020 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm)

View full text Add to dashboard Cite

show abstract

“…This might be especially relevant for online applications or analysis in the transmission system. Due to its high accuracy and run-time, this topic is yet to be exploited [35].…”

Section: Discussionmentioning

confidence: 99%

Optimal Power Dispatch in Energy Systems Considering Grid Constraints

Rubio¹,

Schuldt²,

Klement³

et al. 2021

Preprint

View full text Add to dashboard Cite

As a consequence of the increasing share of renewable energies and sector coupling technologies, new approaches are needed for the study, planning, and control of modern energy systems. Such new structures may add extra stress to the electric grid, as is the case with heat pumps and electrical vehicles. Therefore, the optimal performance of the system must be estimated considering the constraints imposed by the different sectors. In this research, a dispatch optimization method with an iterative grid constraint generation, decoupled from the linear unit commitment problem, is employed. From the considered scenarios, it was found that in a typical German neighborhood with 150 households, PV penetration of &sim;5kWp per household can lead to curtailment of &sim;60MWh per year due to line loading. Furthermore, the proposed method eliminates grid violations due to the addition of new sectors reducing the curtailment up to 60%. With the optimization of the heat pump operation, an increase of 7% of the self-consumption was achieved with similar results for the combination of battery systems and electrical vehicles. In conclusion, a safer and optimal operation of a complex energy system is fulfilled. Safer control strategies and more accurate plant sizing could be derived from this work.

show abstract

“…The current learning-based work consists of two categories. The first category is a hybrid approach, which integrates the learning techniques into the conventional solution algorithm to solve challenging OPF problems [6]- [19]. However, the core of these methods is still the traditional solver, which may incur high computational costs for large-scale power systems.…”

Section: Related Workmentioning

confidence: 99%

DeepOPF+: A Deep Neural Network Approach for DC Optimal Power Flow for Ensuring Feasibility

Zhao

Pan

Chen

et al. 2020

2020 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm)

View full text Add to dashboard Cite

Deep Neural Networks (DNNs) approaches for the Optimal Power Flow (OPF) problem received considerable attention recently. A key challenge of these approaches lies in ensuring the feasibility of the predicted solutions to physical system constraints. Due to the inherent approximation errors, the solutions predicted by DNNs may violate the operating constraints, e.g., the transmission line capacities, limiting their applicability in practice. To address this challenge, we develop DeepOPF+ as a DNN approach based on the so-called "preventive" framework. Specifically, we calibrate the generation and transmission line limits used in the DNN training, thereby anticipating approximation errors and ensuring that the resulting predicted solutions remain feasible. We theoretically characterize the calibration magnitude necessary for ensuring universal feasibility. Our DeepOPF+ approach improves over existing DNN-based schemes in that it ensures feasibility and achieves a consistent speed up performance in both light-load and heavy-load regimes. Detailed simulation results on a range of test instances show that the proposed DeepOPF+ generates 100% feasible solutions with minor optimality loss. Meanwhile, it achieves a computational speedup of two orders of magnitude compared to state-of-the-art solvers. NOMENCLATUREVariable Definition N Set of buses, N |N |. G Set of generators. D Set of loads. E Set of branches. PG Power generation injection vector, [PG i , i ∈ N ]. P min G Minimum generator output vector, [P min G i , i ∈ N ]. P max G Maximum generator output vector, [P max G i , i ∈ N ]. PD Power load vector, [PD i , i ∈ N ]. Θ Voltage angle vector. θi Voltage angle for bus i. B Admittance matrix. xijLine reactance from bus i to j. P max T ij Line transmission limit from bus i to j. NhidThe number of hidden layers in the neural network.

show abstract

Data-Driven Screening of Network Constraints for Unit Commitment

Cited by 63 publications

References 19 publications

Learning Optimal Power Flow: Worst-Case Guarantees for Neural Networks

Learning Optimal Power Flow: Worst-Case Guarantees for Neural Networks

Optimal Power Dispatch in Energy Systems Considering Grid Constraints

DeepOPF+: A Deep Neural Network Approach for DC Optimal Power Flow for Ensuring Feasibility

Contact Info

Product

Resources

About