Data-Driven Local Control Design for Active Distribution Grids Using Off-Line Optimal Power Flow and Machine Learning Techniques

Abstract

The optimal control of distribution networks often requires monitoring and communication infrastructure, either centralized or distributed. However, most of the current distribution systems lack this kind of infrastructure and rely on suboptimal, fit-and-forget, local controls to ensure the security of the network. In this paper, we propose a data-driven algorithm that uses historical data, advanced optimization techniques, and machine learning methods, to design local controls that emulate the optimal behavior without the use of any communication. We demonstrate the performance of the optimized local control on a three-phase, unbalanced, low-voltage, distribution network. The results show that our data-driven methodology clearly outperforms standard industry local control and successfully imitates an optimal-power-flow-based control.

Active Distribution Grids Offering Ancillary Services in Islanded and Grid-Connected Mode

“…In order to consider the impact of generation uncertainty, we follow our previous work [23], [33] and we re-formulate the problem using chance constraints [34], [35]. We assume that the PV power injection is the only source of uncertainty (load uncertainty can be also included in a similar way) and we use as input forecast error distributions with different forecasting horizons (1 to 24 hours ahead).…”

“…Following [23], [33] we model the voltage and current constraints as chance constraints that will hold with a chosen probability 1 − ε, where ε is the acceptable violation probability. E.g., the voltage and current magnitude constraints are reformulated as P {V min ≤ |V j,t | ≤ V max } ≥ 1 − ε and P |I br i,t | ≤ I i max ≥ 1 − ε, respectively.…”

Designing Reactive Power Control Rules for Smart Inverters Using Support Vector Machines

“…Optimal rules however are not necessarily linear: If an apparent power constraint becomes active, reactive injections can become nonlinear functions of solar generation. To capture this nonlinearity, recent approaches engage learning models which are trained to optimize: Given pairs of grid conditions (load and solar generation) and their optimal inverter dispatches computed, the aforesaid approaches learn dispatch rules using linear or kernel-based regression [20], [21].…”

“…Comparison between the proposed single-step learning approach (rules R 1 ), and the two-step learning approach of [20]-[21] (rules R 2 ).+ N n=1 w n Z n λ n − λ n − M x n − u n + N n=1 b n λ n − λ n − M x n…”

Data-driven Control Design Schemes in Active Distribution Grids: Capabilities and Challenges

“…However, in both works only reactive power control is considered, neglecting possible combinations with other available controls, and a balanced operation is assumed. Reference [9] considered reactive control and active power curtailment, while [10], [12] extended the available measures to controllable loads and storage systems. Furthermore, while [8], [11] design local, open-loop controllers, [9], [10], [12] employ a feedback, closed-loop control scheme.…”