Volt-var Control Research Articles

Safe deep reinforcement learning-assisted two-stage energy management for active power distribution networks with hydrogen fueling stations

In a power-hydrogen coupled integrated energy system (PHCIES), hydrogen fueling stations (HFSs) with solar photovoltaic (PV) systems are crucial devices to support hydrogen demand and maintain a stable and near-zero pollutant emission-based PHCIES operation by producing a pollutant-free hydrogen. However, optimal operation management of HFSs is challenging because the electrolyzers, compressor, hydrogen storage system (HSS), and fuel cell in HFSs are interconnected and change dynamically under various operational environments. To resolve this issue, this paper presents a safe deep reinforcement learning (DRL)-assisted two-stage framework that ensures a reliable and economical PHCIES operation. In the first stage, day-ahead operational schedules of stand-alone PV systems, PV-enabled HFSs, on-load tap changers, and capacitor banks with hourly resolution are coordinated to minimize the system operation cost, electricity arbitrage cost, PV curtailment cost, and real power loss by solving a Volt-VAR control (VVC) optimization problem. In the second stage, a safe DRL algorithm with a 15-min resolution is employed to minimize the real power loss and maximize the HFS profit by rescheduling the reactive power of stand-alone PV systems and real/reactive power and hydrogen of HFSs. The proposed self-tuning adaptive safety module integrated in the DRL method ensures no violations of HSS’ state-of-hydrogen (SOH) and voltage magnitude in the PHCIES during the training process. Numerical examples conducted on a PHCIES (IEEE 33-bus system coupled with two PV-enabled HFSs) demonstrate the effectiveness of the proposed framework in terms of training convergence, SOH/voltage violation, real power loss, and profit of HFSs.

Impact of Impedances and Solar Inverter Grid Controls in Electric Distribution Line with Grid Voltage and Frequency Instability

The penetration of solar energy into centralized electric grids has increased significantly during the last decade. Although the electricity from photovoltaics (PVs) can deliver clean and cost-effective energy, the intermittent nature of the sunlight can lead to challenges with electric grid stability. Smart inverter-based resources (IBRs) can be used to mitigate the impact of such high penetration of renewable energy, as well as to support grid reliability by improving the voltage and frequency stability with embedded control functions such as Volt-VAR, Volt–Watt, and Frequency–Watt. In this work, the results of an extensive experimental study of possible interactions between the unstable grid and two residential-scale inverters from different brands under different active and reactive power controls are presented. Two impedance circuits were installed between Power Hardware-in-the-loop (P-HIL) equipment to represent the impedance in an electric distribution line. Grid voltage and frequency were varied between extreme values outside of the normal range to test the response of the two inverters operating under different controls. The key findings highlighted that different inverters that have met the same requirements of IEEE 1547-2018 responded to grid instabilities differently. Therefore, commissioning tests to ensure inverter performance are crucial. In addition to the grid control, the residential PV installed capacity and physical distances between PV homes and the substation, which impacted the distribution wiring impedance which we characterized by the ratio of the reactive to real impedance (X/R), should be considered when assigning the grid-supporting control setpoints to smart inverters. A higher X/R of 3.5 allowed for more effective control to alleviate both voltage and frequency stability. The elimination of deadband in an aggressive Volt-VAR control also enhanced the ability to control voltage during extreme fluctuation. The analysis of sudden spikes in the grid responses to a large frequency drop showed that a shallow slope of 1.5 kW/Hz in the droop control resulted in a >65% lower sudden reactive power overshoot amplitude than a steeper slope of 2.8 kW/Hz.

Attention-Enhanced Multi-Agent Reinforcement Learning Against Observation Perturbations for Distributed Volt-VAR Control

Attention-Enhanced Multi-Agent Reinforcement Learning Against Observation Perturbations for Distributed Volt-VAR Control

Reactive compensation in distribution systems and volt/var control analysis: Mutual performance of inverters and curve slope effects

The high penetration of photovoltaic distributed generation is causing overvoltage in distribution networks due to the reverse active power flow. Proposals to mitigate this problem through Volt/Var control (VVC) by photovoltaic inverters are being discussed. However, the configuration of the appropriate Volt/Var curve depends on the topological factors of each system. In this context, this paper presents a practical methodology for calculating the reactive compensation of the local VVC. In addition, the impact of the slope of the curve on power quality and the problems of mutual actuation of inverters in the VVC are analyzed. The studies were conducted in quasistatic time series (QSTS) mode on the IEEE 33 bus system. The slope of the curve affects grid losses and the substation power factor, and can mitigate the negative effects of the mutual actuation of the inverters in the VVC. Based on this, guidelines were established to adapt the definition of the slope to the characteristics of each system.

Improving Distribution System State Estimation by Including Volt-Var Control Information

Improving Distribution System State Estimation by Including Volt-Var Control Information

Two-Critic Deep Reinforcement Learning for Inverter-Based Volt-Var Control in Active Distribution Networks

Two-Critic Deep Reinforcement Learning for Inverter-Based Volt-Var Control in Active Distribution Networks

Volt-VAR Control in Active Distribution Networks Using Multi-Agent Reinforcement Learning

With the advancement of power systems, the integration of a substantial portion of renewable energy often leads to frequent voltage surges and increased fluctuations in distribution networks (DNs), significantly affecting the safety of DNs. Active distribution networks (ADNs) can address voltage issues arising from a high proportion of renewable energy by regulating distributed controllable resources. However, the conventional mathematical optimization-based approach to voltage reactive power control has certain limitations. It heavily depends on precise DN parameters, and its online implementation requires iterative solutions, resulting in prolonged computation time. In this study, we propose a Volt-VAR control (VVC) framework in ADNs based on multi-agent reinforcement learning (MARL). To simplify the control of photovoltaic (PV) inverters, the ADNs are initially divided into several distributed autonomous sub-networks based on the electrical distance of reactive voltage sensitivity. Subsequently, the Multi-Agent Soft Actor-Critic (MASAC) algorithm is employed to address the partitioned cooperative voltage control problem. During online deployment, the agents execute distributed cooperative control based on local observations. Comparative tests involving various methods are conducted on IEEE 33-bus and IEEE 141-bus medium-voltage DNs. The results demonstrate the effectiveness and versatility of this method in managing voltage fluctuations and mitigating reactive power loss.

A supervisory Volt/Var control scheme for coordinating voltage regulators with smart inverters on a distribution system

This paper concentrates on the efficient utilization of smart inverters for Volt/Var control (VVC) within a distribution system. Although new smart inverters possess Var support capability, their effective deployment necessitates coordination with existing Volt/Var schemes. To address this, a novel VVC scheme is proposed to facilitate such synchronization. The proposed scheme bifurcates the issue into two levels. The initial level involves utilizing Load Tap Changer (LTC) and Voltage Regulators (VRs), coordinating their control with smart inverters to regulate the circuit's voltage levels within the desired range. The subsequent level determines the Var support required from smart inverters to minimize overall power loss in the circuit. The supervisory control results are communicated to the respective devices equipped with their local controllers. To minimize frequent dispatch, smart inverters are supervised by adjusting their Volt/Var characteristics as necessary. This approach enables the smart inverters to operate near their optimal control while meeting the limited communication prerequisites in a distribution system. A case study employing the IEEE 34 bus system illustrates the efficacy of this supervisory control scheme in contrast to traditional Volt/Var schemes.

Distributed optimal Volt/Var control in power electronics dominated AC/DC hybrid distribution network

AbstractThe integration of large‐scale distributed power sources increases the voltage fluctuation in AC/DC hybrid distribution network (AD‐HDN). Power electronics devices such as photovoltaic (PV) inverters, soft open point (SOP), and voltage source converters (VSCs) can be utilized for voltage/var control (VVC) to alleviate the risk of voltage fluctuation and violation. This paper proposes a distributed optimal VVC method in power electronics dominated AD‐HDN. Firstly, the reactive power and voltage characteristics of PV inverters, SOP, and VSCs are analysed, and an optimal VVC optimization model for AD‐HDN to minimize node voltage deviation, PV curtailment, and network loss is proposed. Then, the second‐order cone (SOC) relaxation technique is used to re‐formulate the model into a convex optimization model. A distributed optimal VVC framework based on the alternating direction method of multipliers (ADMM) is constructed. Based on the residual balance principle and relaxation technique, an accelerated ADMM method is further proposed to solve the proposed model. Finally, case studies are conducted on the IEEE 33‐node and 85‐node systems to verify the superiority and effectiveness of the proposed method.

A Machine Learning-Assisted Distributed Optimization Method for Inverter-Based Volt-VAR Control in Active Distribution Networks

The number of smart inverters in active distribution networks is growing rapidly, making it challenging to realize a fast, distributed Volt/Var control (VVC). This work proposes a machine learning-assisted distributed algorithm to accelerate the solution of the VVC strategy. We first observe the convergence process of the Alternating Direction Method of Multipliers (ADMM)-based VVC problem and explore the potential relationships between the convergence and time-series regression. Then, the long short-term memory (LSTM) technique is applied to learn the convergence process and regress the converged values of the dual and global variables with previous ADMM observations. After that, the LSTM-assisted ADMM algorithm is proposed, where the regressions are used for ADMM parameter updates. In this algorithm, the inputs of the LSTM model are carefully designed since the complementary conditions implied in the conventional ADMM should be considered. Unlike existing methods, the proposed method does not use the LSTM to determine the VVC strategy directly, indicating that it is non-intrusive and can satisfy all safety constraints during operations. The proof of its optimality and convergence is also given. The numerical simulations on the 33-bus distribution system demonstrate the effectiveness and efficiency of the proposed method.

Volt–Var curve determination method of smart inverters by multi-agent deep reinforcement learning

Reactive power control of PV inverters can be applied to mitigate the voltage increase caused by reverse power flow and voltage fluctuations caused by PV output fluctuations in the distribution system. This paper focuses on the Volt–Var control of PV smart inverters to minimize power losses. It proposes a multi-agent type cooperative voltage control framework to optimize the blind band and slope of the VVC. The proposed method utilizes the optimized VVC to eliminate voltage deviations using only local information. The optimization problem is formed as a Markov decision process and solved using a deep reinforcement learning algorithm called multi-agent soft-actor critic, which achieves fast control that fully accounts for uncertainty while achieving global optimization. Therefore, the proposed method can contribute to reducing power loss, which is a system-wide problem, with only local information. Case studies were conducted using a small and medium-sized distribution network model with IEEE 33 buses and 12 demand cases. The comparison results demonstrate that the proposed method is superior to other benchmark methods in terms of minimizing power losses and mitigating voltage variations, as the proposed method can reduce power losses by about 9% while achieving avoidance of voltage deviations.

Allocation and smart inverter setting of ground-mounted photovoltaic power plants for the maximization of hosting capacity in distribution networks

As the integration of solar photovoltaic (PV) power plants into distribution networks grows, quantifying the amount of PV power that distribution networks can host without harmfully impacting power quality becomes critical. This work aims to determine the best number, location, and size of PV systems to be installed on a distribution feeder, as well as the best control set-points of the PV inverters, to maximize the PV hosting capacity (HC). Therefore, a simulation-optimization framework is proposed for siting and sizing ground-mounted PV power plants equipped with smart inverters (SIs). Single (decentralized) and multiple (distributed) allocations are analyzed by considering the connection of one, two, and three PV systems. Genetic algorithm (GA) and particle swarm optimization (PSO) metaheuristics are employed to solve the optimization problem. The simulation-optimization framework is tested on a real-world feeder model from an Ecuadorian utility. Installing two PV systems with their SIs operating with the Volt-VAr control function yields maximum PV HC, which is increased by 32.1 % compared to a single PV power plant operating at a unity power factor. Moreover, a comparative analysis of the two metaheuristic algorithms reveals that the PSO method provides better results than GA.

Multi-Agent Safe Graph Reinforcement Learning for PV Inverters-Based Real-Time Decentralized Volt/Var Control in Zoned Distribution Networks

To realize real-time voltage/var control (VVC) in active distribution networks (ADNs), this paper proposes a new multi-agent safe graph reinforcement learning method to optimize reactive power output from PV inverters. The network is divided into several zones, and a decentralized framework is proposed for coordinated control of reactive power output in each zone to regulate voltage profiles and minimize network energy loss. The VVC problem is formulated as a multi-agent decentralized partially observable constrained Markov decision process. Each zone has a central control agent that embeds graph convolution networks (GCNs) in the policy network to improve the decision-making capability. The GCN extracts graph-structured features from the ADN topology, reflecting the relationship between VVC and grid topology, and can filter noise and impute missing data. The training process includes primal-dual policy optimization to rigorously satisfy voltage safety constraints. Simulations on a 141-bus distribution system demonstrate that the proposed method can effectively minimize network energy loss and reduce voltage deviations, even in the presence of noisy or incomplete input measurements.

Optimal Short-Term Charge/Discharge Operation for Electric Vehicles With Volt-Var Control in Day-Ahead Electricity Market

Optimal Short-Term Charge/Discharge Operation for Electric Vehicles With Volt-Var Control in Day-Ahead Electricity Market

Residual Deep Reinforcement Learning with Model-based Optimization for Inverter-based Volt-Var Control

Residual Deep Reinforcement Learning with Model-based Optimization for Inverter-based Volt-Var Control

Deterministic and Robust Volt-var Control Methods of Power System Based on Convex Deep Learning

Deterministic and Robust Volt-var Control Methods of Power System Based on Convex Deep Learning

A linear Distflow model considering line shunts for fast calculation and voltage control of power distribution systems

The line shunts are usually ignored by various linear power flow (PF) models in power distribution system analysis, planning and optimization. However, “charging effects” from line shunts of underground/submarine power cables would cause non-negligible model errors for these commonly used PF models. In this paper, we propose a modified Linear Distflow model (LinDist) with line shunts, i.e., LinDistS. We also further propose its extensions considering the ZIP load, weakly-meshed topology and unbalanced three-phase systems. Moreover, the linearization error of voltage component is theoretically analyzed. Case studies show that compared with other models, the proposed LinDistS achieves the descent calculation accuracy and efficiency. We also show the application scope of LinDistS by a Volt/VAr control (VVC) framework with distributed generators, mainly photovoltaic (PV), and the non-negligible “charging effect”. Simulation exhibits that with LinDistS, the VVC can optimally dispatch the shunt capacitors and also optimize the real-time reactive power output of PV. Moreover, with LinDistS, the VVC shows better solutions’ consistency and higher computing efficiency compared to traditional VVC methods.

Adaptable Volt-VAR control digital twinning for smart solar inverters

The high-level goal of this research work is to enhance solar integration into power distribution grids with a lower cost and less complexity. High penetration levels of solar can lead to power quality issues at the point of interconnection such as flicker and voltage violation. This paper presents a novel Delta-Q approach that determines the most appropriate amount of reactive power support independent of the time step. The proposed scheme is embodied in the solar smart inverters’ Volt-VAR control. It concurrently mitigates voltage violation and smoothens the voltage profile in areas where the solar irradiance is highly intermittent. The performance of the proposed control scheme is evaluated on the Typhoon HIL digital twining testbed along with the dSPACE MicroLabBox rapid control prototype unit. The real-time analysis shows the effectiveness of the proposed technique in lowering voltage violations and also mitigating voltage fluctuations.

Development approach of a volt-var control for inverter-coupled renewable energy plants using deep reinforcement learning

Development approach of a volt-var control for inverter-coupled renewable energy plants using deep reinforcement learning

Encryption-based Coordinated Volt/Var Control for Distribution Networks With Multi-Microgrids

This paper proposes a data-driven coordinated volt/var control (VVC) strategy for active distribution networks (ADN) with multi-microgrids, which can achieve online economic and secure operations under false data injection attacks (FDIA). Based on voltage and power measurements, the microgrid central controller (MGCC) obtains optimal reactive power supports at the point of common coupling (PCC) and sends them to the distribution system operator (DSO). The MGCC is formulated with a convolution neural network (CNN) to emulate the optimal behaviors in microgrids (MGs), which can reduce computational burdens and facilitate its online application. The DSO then performs centralized optimization to dispatch VVC devices and update voltages at PCC. A voltage sensitivity-based reactive power adjustment method is also developed to simplify the iterative optimization process between ADN and MGs without deteriorating the VVC performance in each MG. Finally, data integrity and privacy are protected through an encrypted communication process against FDIA. The GGH (Goldreich-Goldwasser-Halevi) encryption algorithm directly prevents attackers from accessing the original transmitted data, while the RSA (Rivest-Shamir-Adleman) digital signature algorithm helps detect malicious tampering with the ciphertext during communication. Numerical simulations on a modified IEEE 33-bus ADN with three EU 16-bus MGs verify the effectiveness of the proposed method in mitigating voltage violations, reducing voltage regulation costs and protecting data security.