Abstract:Industrial, commercial, and residential facilities are progressively adopting automation and generation capabilities. By having flexible demand and renewable energy generation, traditional passive customers are becoming active participants in electric power system operations. Through profound coordination among grid operators and active customers, the facilities’ capability for demand response (DR) and distributed energy resource (DER) management will be valuable asset for ancillary services (ASs). To comply with the increasing demand and flexible energy, utilities urgently require standards, regulations, and programs to efficiently handle load-side resources without trading off stability and reliability. This study reviews different types of customers’ flexibilities for DR, highlighting their capabilities and limitations in performing local ancillary services (LASs), which should benefit the power grid by profiting from it through incentive mechanisms. Different financial incentives and techniques employed around the world are presented and discussed. The potential barriers in technical and regulatory aspects are successfully identified and potential solutions along with future guidance are discussed.

Abstract:The load demand and distributed generation (DG) integration capacity in distribution networks (DNs) increase constantly, and it means that the violation of security constraints may occur in the future. This can be further worsened by short-term power fluctuations. In this paper, a scheduling method based on a multi-objective chance-constrained information-gap decision (IGD) model is proposed to obtain the active management schemes for distribution system operators (DSOs) to address these problems. The maximum robust adaptability of multiple uncertainties, including the deviations of growth prediction and their relevant power fluctuations, can be obtained based on the limited budget of active management. The systematic solution of the proposed model is developed. The max term constraint in the IGD model is converted into a group of normal constraints corresponding to extreme points of the max term. Considering the stochastic characteristics and correlations of power fluctuations, the original model is equivalently reformulated by using the properties of multivariate Gaussian distribution. The effectiveness of the proposed model is verified by a modified IEEE 33-bus distribution network. The simulation result delineates a robust accommodation space to represent the adaptability of multiple uncertainties, which corresponds to an optional active management strategy set for future selection.

Abstract:Accurate and timely fault diagnosis is of great significance for the safe operation and power supply reliability of distribution systems. However, traditional intelligent methods limit the use of the physical structures and data information of power networks. To this end, this study proposes a fault diagnostic model for distribution systems based on deep graph learning. This model considers the physical structure of the power network as a significant constraint during model training, which endows the model with stronger information perception to resist abnormal data input and unknown application conditions. In addition, a special spatiotemporal convolutional block is utilized to enhance the waveform feature extraction ability. This enables the proposed fault diagnostic model to be more effective in dealing with both fault waveform changes and the spatial effects of faults. In addition, a multi-task learning framework is constructed for fault location and fault type analysis, which improves the performance and generalization ability of the model. The IEEE 33-bus and IEEE 37-bus test systems are modeled to verify the effectiveness of the proposed fault diagnostic model. Finally, different fault conditions, topological changes, and interference factors are considered to evaluate the anti-interference and generalization performance of the proposed model. Experimental results demonstrate that the proposed model outperforms other state-of-the-art methods.

Abstract:With the large-scale distributed generations (DGs) being connected to distribution network (DN), the traditional day-ahead reconfiguration methods based on physical models are challenged to maintain the robustness and avoid voltage off-limits. To address these problems, this paper develops a deep reinforcement learning method for the sequential reconfiguration with soft open points (SOPs) based on real-time data. A state-based decision model is first proposed by constructing a Marko decision process-based reconfiguration and SOP joint optimization model so that the decisions can be achieved in milliseconds. Then, a deep reinforcement learning joint framework including branching double deep Q network (BDDQN) and multi-policy soft actor-critic (MPSAC) is proposed, which has significantly improved the learning efficiency of the decision model in multi-dimensional mixed-integer action space. And the influence of DG and load uncertainty on control results has been minimized by using the real-time status of the DN to make control decisions. The numerical simulations on the IEEE 34-bus and 123-bus systems demonstrate that the proposed method can effectively reduce the operation cost and solve the overvoltage problem caused by high ratio of photovoltaic (PV) integration.

Abstract:The increasing flexibility of active distribution systems (ADSs) coupled with the high penetration of renewable distributed generators (RDGs) leads to the increase of the complexity. It is of practical significance to achieve the largest amount of RDG penetration in ADSs and maintain the optimal operation. This study establishes an alternating current (AC)/direct current (DC) hybrid ADS model that considers the dynamic thermal rating, soft open point, and distribution network reconfiguration (DNR). Moreover, it transforms the optimal dispatching into a second-order cone programming problem. Considering the different control time scales of dispatchable resources, the following two-stage dispatching framework is proposed. ① The day-ahead dispatch uses hourly input data with the goal of minimizing the grid loss and RDG dropout. It obtains the optimal 24-hour schedule to determine the dispatching plans for DNR and the energy storage system. ② The intraday dispatch uses 15 min of input data for 1-hour rolling-plan dispatch but only executes the first 15 min of dispatching. To eliminate error between the actual operation and dispatching plan, the first 15 min is divided into three 5-min step-by-step executions. The goal of each step is to trace the tie-line power of the intraday rolling-plan dispatch to the greatest extent at the minimum cost. The measured data are used as feedback input for the rolling-plan dispatch after each step is executed. A case study shows that the comprehensive cooperative ADS model can release the line capacity, reduce losses, and improve the penetration rate of RDGs. Further, the two-stage dispatching framework can handle source-load fluctuations and enhance system stability.

Abstract:Direct current (DC) bus voltage stability is essential for the stable and reliable operation of a DC system. If an oscillation source can be quickly and accurately localized, the oscillation can be adequately eliminated. We propose a method based on the power spectral density for identifying the voltage oscillation source. Specifically, a DC distribution network model combined with the component connection method is developed, and the network is separated into multiple power modules. Compared with a conventional method, the proposed method does not require determining the model parameters of the entire power grid, which is typically challenging. Furthermore, combined with a novel judgment index, the oscillation source can be identified more intuitively and clearly to enhance the applicability to real power grids. The performance of the proposed method has been evaluated using the MATLAB/Simulink software and PLECS RT Box experimental platform. The simulation and experimental results verify that the proposed method can accurately identify oscillation sources in a DC distribution network.

Abstract:Most existing distribution networks are difficult to withstand the impact of meteorological disasters. With the development of active distribution networks (ADNs), more and more upgrading and updating resources are applied to enhance the resilience of ADNs. A two-stage stochastic mixed-integer programming (SMIP) model is proposed in this paper to minimize the upgrading and operation cost of ADNs by considering random scenarios referring to different operation scenarios of ADNs caused by disastrous weather events. In the first stage, the planning decision is formulated according to the measures of hardening existing distribution lines, upgrading automatic switches, and deploying energy storage resources. The second stage is to evaluate the operation cost of ADNs by considering the cost of load shedding due to disastrous weather and optimal deployment of energy storage systems (ESSs) under normal weather condition. A novel modeling method is proposed to address the uncertainty of the operation state of distribution lines according to the canonical representation of logical constraints. The progressive hedging algorithm (PHA) is adopted to solve the SMIP model. The IEEE 33-node test system is employed to verify the feasibility and effectiveness of the proposed method. The results show that the proposed model can enhance the resilience of the ADN while ensuring economy.

Abstract:This paper proposes a stochastic programming (SP) method for coordinated operation of distributed energy resources (DERs) in the unbalanced active distribution network (ADN) with diverse correlated uncertainties. First, the three-phase branch flow is modeled to characterize the unbalanced nature of the ADN, schedule DER for three phases, and derive a realistic DER allocation. Then, both active and reactive power resources are co-optimized for voltage regulation and power loss reduction. Second, the battery degradation is considered to model the aging cost for each charging or discharging event, leading to a more realistic cost estimation. Further, copula-based uncertainty modeling is applied to capture the correlations between renewable generation and power loads, and the two-stage SP method is then used to get final solutions. Finally, numerical case studies are conducted on an IEEE 34- bus three-phase ADN, verifying that the proposed method can effectively reduce the system cost and co-optimize the active and reactive power.

Abstract:The increased deployment of electricity-based hydrogen production strengthens the coupling of power distribution system (PDS) and hydrogen energy system (HES). Considering that power to hydrogen (PtH) has great potential to facilitate the usage of renewable energy sources (RESs), the coordination of PDS and HES is important for planning purposes under high RES penetration. To this end, this paper proposes a multi-stage co-planning model for the PDS and HES. For the PDS, investment decisions on network assets and RES are optimized to supply the growing electric load and PtHs. For the HES, capacities of PtHs and hydrogen storages (HSs) are optimally determined to satisfy hydrogen load considering the hydrogen production, tube trailer transportation, and storage constraints. The overall planning problem is formulated as a multi-stage stochastic optimization model, in which the investment decisions are sequentially made as the uncertainties of electric and hydrogen load growth states are revealed gradually over periods. Case studies validate that the proposed co-planning model can reduce the total planning cost, promote RES consumption, and obtain more flexible decisions under long-term load growth uncertainties.

Abstract:Due to the lack of flexible interconnection devices, power imbalances between networks cannot be relieved effectively. Meanwhile, increasing the penetration of distributed generators exacerbates the temporal power imbalances caused by large peak-valley load differences. To improve the operational economy lowered by spatiotemporal power imbalances, this paper proposes a two-stage optimization strategy for active distribution networks (ADNs) interconnected by soft open points (SOPs). The SOPs and energy storage system (ESS) are adopted to transfer power spatially and temporally, respectively. In the day-ahead scheduling stage, massive stochastic scenarios against the uncertainty of wind turbine output are generated first. To improve computational efficiency in massive stochastic scenarios, an equivalent model between networks considering sensitivities of node power to node voltage and branch current is established. The introduction of sensitivities prevents violations of voltage and current. Then, the operating ranges (ORs) of the active power of SOPs and the state of charge (SOC) of ESS are obtained from models between networks and within the networks, respectively. In the intraday corrective control stage, based on day-ahead ORs, a receding-horizon model that minimizes the purchase cost of electricity and voltage deviations is established hour by hour. Case studies on two modified ADNs show that the proposed strategy achieves spatiotemporal power balance with lower cost compared with traditional strategies.

Abstract:The emergence of prosumers in distribution systems has enabled competitive electricity markets to transition from traditional hierarchical structures to more decentralized models such as peer-to-peer (P2P) and community-based (CB) energy transaction markets. However, the network usage charge (NUC) that prosumers pay to the electric power utility for network services is not adjusted to suit these energy transactions, which causes a reduction in revenue streams of the utility. In this study, we propose an NUC calculation method for P2P and CB transactions to address holistically economic and technical issues in transactive energy markets and distribution system operations, respectively. Based on the Nash bargaining (NB) theory, we formulate an NB problem for P2P and CB transactions to solve the conflicts of interest among prosumers, where the problem is further decomposed into two convex subproblems of social welfare maximization and payment bargaining. We then build the NUC calculation model by coupling the NB model and AC optimal power flow model. We also employ the Shapley value to allocate the NUC to consumers fairly for the NUC model of CB transactions. Finally, numerical studies on IEEE 15-bus and 123-bus distribution systems demonstrate the effectiveness of the proposed NUC calculation method for P2P and CB transactions.

Abstract:High penetration of distributed renewable energy sources and electric vehicles (EVs) makes future active distribution network (ADN) highly variable. These characteristics put great challenges to traditional voltage control methods. Voltage control based on the deep Q-network (DQN) algorithm offers a potential solution to this problem because it possesses human-level control performance. However, the traditional DQN methods may produce overestimation of action reward values, resulting in degradation of obtained solutions. In this paper, an intelligent voltage control method based on averaged weighted double deep Q-network (AWDDQN) algorithm is proposed to overcome the shortcomings of overestimation of action reward values in DQN algorithm and underestimation of action reward values in double deep Q-network (DDQN) algorithm. Using the proposed method, the voltage control objective is incorporated into the designed action reward values and normalized to form a Markov decision process (MDP) model which is solved by the AWDDQN algorithm. The designed AWDDQN-based intelligent voltage control agent is trained offline and used as online intelligent dynamic voltage regulator for the ADN. The proposed voltage control method is validated using the IEEE 33-bus and 123-bus systems containing renewable energy sources and EVs，and compared with the DQN and DDQN algorithms based methods, and traditional mixed-integer nonlinear program based methods. The simulation results show that the proposed method has better convergence and less voltage volatility than the other ones.

Abstract:Scalable coordination of photovoltaic (PV) inverters, considering the uncertainty in PV and load in distribution networks (DNs), is challenging due to the lack of real-time communications. Decentralized PV inverter setpoints can be achieved to address this issue by capitalizing on the abundance of data from smart utility meters and the scalable architecture of artificial neural networks (ANNs). To this end, we first use an offline, centralized data-driven conservative convex approximation of chance-constrained optimal power flow (CVaR-OPF) in which conditional value-at-risk (CVaR) is used to compute reactive power setpoints of PV inverter, taking into account PV and load uncertainties in DNs. Following that, an artificial neural network (ANN) controller is trained for each PV inverter to emulate the optimal behavior of the centralized control setpoints of PV inverter in a decentralized fashion. Additionally, the voltage regulation performance of the developed ANN controllers is compared with other decentralized designs (local controllers) developed using model-based learning (regression-based controller), optimization (affine feedback controller), and case-based learning (mapping) approaches. Numerical tests using real-world feeders corroborate the effectiveness of ANN controllers in voltage regulation and loss minimization.

Abstract:Owing to potential regulation capacities from flexible resources in energy coupling, storage, and consumption links, central energy stations (CESs) can provide additional support to power distribution network (PDN) in case of power disruption. However, existing research has not explicitly revealed the emergency response of PDN with leveraging multiple CESs. This paper proposes a decentralized self-healing strategy of PDN to minimize the entire load loss, in which multi-area CESs’ potentials including thermal storage and building thermal inertia, as well as the flexible topology of PDN, are reasonably exploited for service recovery. For sake of privacy preservation, the co-optimization of PDN and CESs is realized in a decentralized manner using adaptive alternating direction method of multipliers (ADMM). Furtherly, bilateral risk management with conditional value-at-risk (CVaR) for PDN and risk constraints for CESs is integrated to deal with uncertainties from outage duration. Case studies are conducted on a modified IEEE 33-bus PDN with multiple CESs. Numerical results illustrate that the proposed strategy can fully utilize the potentials of multi-area CESs for coordinated load restoration. The effectiveness of the performance and behaviors’ adaptation against random risks is also validated.

Abstract:The purpose of active distribution networks (ADNs) is to provide effective control approaches for enhancing the operation of distribution networks (DNs) and greater accommodation of distributed generation (DG) sources. With the integration of DG sources into DNs, several operational problems have drawn attention such as overvoltage and power flow alteration issues. These problems can be dealt with by utilizing distribution network reconfiguration (DNR) and soft open points (SOPs). An SOP is a power electronic device capable of accurately controlling active and reactive power flows. Another significant aspect often overlooked is the coordination of protection devices needed to keep the network safe from damage. When implementing DNR and SOPs in real DNs, protection constraints must be considered. This paper presents an ADN reconfiguration approach that includes DG sources, SOPs, and protection devices. This approach selects the ideal configuration, DG output, and SOP placement and control by employing particle swarm optimization (PSO) to minimize power loss while ensuring the correct operation of protection devices under normal and fault conditions. The proposed approach explicitly formulates constraints on network operation, protection coordination, DG size, and SOP size. Finally, the proposed approach is evaluated using the standard IEEE 33-bus and IEEE 69-bus networks to demonstrate the validity.

Abstract:A photovoltaic (PV)-rich low-voltage (LV) distribution network poses a limit on the export power of PVs due to the voltage magnitude constraints. By defining a customer export limit, switching off the PV inverters can be avoided, and thus reducing power curtailment. Based on this, this paper proposes a mixed-integer nonlinear programming (MINLP) model to define such optimal customer export. The MINLP model aims to minimize the total PV power curtailment while considering the technical operation of the distribution network. First, a nonlinear mathematical formulation is presented. Then, a new set of linearizations approximating the Euclidean norm is introduced to turn the MINLP model into an MILP formulation that can be solved with reasonable computational effort. An extension to consider multiple stochastic scenarios is also presented. The proposed model has been tested in a real LV distribution network using smart meter measurements and irradiance profiles from a case study in the Netherlands. To assess the quality of the solution provided by the proposed MILP model, Monte Carlo simulations are executed in OpenDSS, while an error assessment between the original MINLP and the approximated MILP model has been conducted.

Abstract:Nowadays, grid-connected renewable energy resources have widespread applications in the electricity market. However, providing household consumers with photovoltaic (PV) systems requires bilateral interfaces to exchange energy and data. In addition, residential consumers’ contribution requires guaranteed privacy and secured data exchange. Day-ahead dynamic pricing is one of the incentive-based demand response methods that has substantial effects on the integration of renewable energy resources with smart grids and social welfare. Different metering mechanisms of renewable energy resources such as feed-in tariffs, net metering, and net purchase and sale are important issues in power grid operation planning. In this paper, optimal condition decomposition method is used for day-ahead dynamic pricing of grid-connected residential renewable energy resources under different metering mechanisms: feed-in-tariffs, net metering, and net purchase and sale in conjunction with carbon emission taxes. According to the stochastic nature of consumers’ load and PV system products, uncertainties are considered in a two-stage decision-making process. The results demonstrate that the net metering with the satisfaction average of 68% for consumers and 32% for the investigated electric company leads to 28% total load reduction. For the case of net purchase and sale mechanism, a satisfaction average of 15% for consumers and 85% for the electric company results in 11% total load reduction. In feed-in-tariff mechanism, in spite of increased social welfare, load reduction does not take place.

Abstract:The penetration of distributed energy resources (DERs) and energy-intensive resources is gradually increasing in active distribution networks (ADNs), which leads to frequent and severe voltage violation problems. As a densely distributed flexible resource in the future distribution network, 5G base station (BS) backup battery is used to regulate the voltage profile of ADN in this paper. First, the dispatchable potential of 5G BS backup batteries is analyzed. Considering the spatial-temporal characteristics of electric load for 5G BS, the dispatchable capacity of backup batteries at different time intervals is evaluated based on historical heat map data. Then, a voltage profile optimization model for ADN is established, consisting of 5G BS backup batteries and other voltage regulation resources. In this model, the charging/discharging behavior of backup batteries is based on its evaluation result of dispatchable capacity. Finally, the range of charging/discharging cost coefficients of 5G BS that benefits ADN and 5G operators are analyzed respectively. Further, an incentive policy for 5G operators is proposed. Under this policy, the charging/discharging cost coefficients of 5G BS can achieve a win-win situation for ADN and 5G operators. As an emerging flexible resource in ADN, the effectiveness and economy of 5G BS backup batteries participating in voltage profile optimization are verified in a test distribution network.

Abstract:The volatile and intermittent nature of distributed generators (DGs) in active distribution networks (ADNs) increases the uncertainty of operating states. The introduction of distribution phasor measurement units (D-PMUs) enhances the monitoring level. The trade-offs of computational performance and robustness of state estimation in monitoring the network states are of great significance for ADNs with D-PMUs and DGs. This paper proposes a second-order cone programming (SOCP) based robust state estimation (RSE) method considering multi-source measurements. Firstly, a linearized state estimation model related to the SOCP state variables is formulated. The phase angle measurements of D-PMUs are converted to equivalent power measurements. Then, a revised SOCP-based RSE method with the weighted least absolute value estimator is proposed to enhance the convergence and bad data identification. Multi-time slots of D-PMU measurements are utilized to improve the estimation accuracy of RSE. Finally, the effectiveness of the proposed method is illustrated in the modified IEEE 33-node and IEEE 123-node systems.

Abstract:In the compression of massive compound power quality disturbance (PQD) signals in active distribution networks, the compression ratio (CR) and reconstruction error (RE) act as a pair of contradictory indicators, and traditional compression algorithms have difficulties in simultaneously satisfying a high CR and low RE. To improve the CR and reduce the RE, a hybrid compression method that combines a strong tracking Kalman filter (STKF), sparse decomposition, Huffman coding, and run-length coding is proposed in this study. This study first uses a sparse decomposition algorithm based on a joint dictionary to separate the transient component (TC) and the steady-state component (SSC) in the PQD. The TC is then compressed by wavelet analysis and by Huffman and run-length coding algorithms. For the SSC, values that are greater than the threshold are reserved, and the compression is finally completed. In addition, the threshold of the wavelet depends on the fading factor of the STKF to obtain a high CR. Experimental results of real-life signals measured by fault recorders in a dynamic simulation laboratory show that the CR of the proposed method reaches as high as 50 and the RE is approximately 1.6%, which are better than those of competing methods. These results demonstrate the immunity of the proposed method to the interference of Gaussian noise and sampling frequency.

Abstract:This paper presents a properly designed branch current based state estimator (BCBSE) used as the main core of an accurate fault location approach (FLA) devoted to distribution networks. Contrary to the approaches available in the literature, it uses only a limited set of conventional measurements obtained from smart meters to accurately locate faults at buses or branches without requiring measurements provided by phasor measurement units (PMUs). This is possible due to the methods used to model the angular reference and the faulted bus, in addition to the proper choice of the weights in the state estimator (SE). The proposed approach is based on a searching procedure composed of up to three stages ① the identification of the faulted zones; ② the identification of the bus closest to the fault; and ③ the location of the fault itself, searching on branches connected to the bus closest to the fault. Furthermore, this paper presents a comprehensive assessment of the proposed approach, even considering the presence of distributed generation, and a sensitivity study on the proper weights required by the SE for fault location purposes, which can not be found in the literature. Results show that the proposed BCBSE-based FLA is robust, accurate, and aligned with the requirements of the traditional and active distribution networks.

Abstract:To optimize the placement of soft open points (SOPs) in active distribution networks (ADNs), many aspects should be considered, including the adjustment of transmission power, integration of distributed generations (DGs), coordination with conventional control methods, and maintenance of economic costs. To address this multi-objective planning problem, this study proposes a multi-stage coordinated robust optimization model for the SOP allocation in ADNs with photovoltaic (PV). First, two robust technical indices based on a robustness index are proposed to evaluate the operation conditions and robust optimality of the solutions. Second, the proposed coordinated allocation model aims to optimize the total cost, robust voltage offset index, robust utilization index, and voltage collapse proximity index. Third, the optimization methods of the multi- and single-objective models are coordinated to solve the proposed multi-stage problem. Finally, the proposed model is implemented on an IEEE 33-node distribution system to verify its effectiveness. Numerical results show that the proposed index can better reveal voltage offset conditions as well as the SOP utilization, and the proposed model outperforms conventional ones in terms of robustness of placement plans and total cost.

Abstract:With the booming of electric vehicles (EVs) across the world, their increasing charging demands pose challenges to urban distribution networks. Particularly, due to the further implementation of time-of-use prices, the charging behaviors of household EVs are concentrated on low-cost periods, thus generating new load peaks and affecting the secure operation of the medium- and low-voltage grids. This problem is particularly acute in many old communities with relatively poor electricity infrastructure. In this paper, a novel two-stage charging scheduling scheme based on deep reinforcement learning is proposed to improve the power quality and achieve optimal charging scheduling of household EVs simultaneously in active distribution network (ADN) during valley period. In the first stage, the optimal charging profiles of charging stations are determined by solving the optimal power flow with the objective of eliminating peak-valley load differences. In the second stage, an intelligent agent based on proximal policy optimization algorithm is developed to dispatch the household EVs sequentially within the low-cost period considering their discrete nature of arrival. Through powerful approximation of neural network, the challenge of imperfect knowledge is tackled effectively during the charging scheduling process. Finally, numerical results demonstrate that the proposed scheme exhibits great improvement in relieving peak-valley differences as well as improving voltage quality in the ADN.

Abstract:With the large-scale integration of distributed renewable generation (DRG) and increasing proportion of power electronic equipment, the traditional power distribution network (DN) is evolving into an active distribution network (ADN). The operation state of an ADN, which is equipped with DRGs, could rapidly change among multiple states, which include steady, alert, and fault states. It is essential to manage large-scale DRG and enable the safe and economic operation of ADNs. In this paper, the current operation control strategies of ADNs under multiple states are reviewed with the interpretation of each state and the transition among the three aforementioned states. The multi-state identification indicators and identification methods are summarized in detail. The multi-state regulation capacity quantification methods are analyzed considering controllable resources, quantification indicators, and quantification methods. A detailed survey of optimal operation control strategies, including multiple state operations, is presented, and key problems and outlooks for the expansion of ADN are discussed.

Abstract:The increasing integration of intermittent renewable energy sources (RESs) poses great challenges to active distribution networks (ADNs), such as frequent voltage fluctuations. This paper proposes a novel ADN strategy based on multi-agent deep reinforcement learning (MADRL), which harnesses the regulating function of switch state transitions for the real-time voltage regulation and loss minimization. After deploying the calculated optimal switch topologies, the distribution network operator will dynamically adjust the distributed energy resources (DERs) to enhance the operation performance of ADNs based on the policies trained by the MADRL algorithm. Owing to the model-free characteristics and the generalization of deep reinforcement learning, the proposed strategy can still achieve optimization objectives even when applied to similar but unseen environments. Additionally, integrating parameter sharing (PS) and prioritized experience replay (PER) mechanisms substantially improves the strategic performance and scalability. This framework has been tested on modified IEEE 33-bus, IEEE 118-bus, and three-phase unbalanced 123-bus systems. The results demonstrate the significant real-time regulation capabilities of the proposed strategy.