The randomness, volatility, and intermittence of solar power generation make it difficult to achieve the desired accuracy PV output-power prediction. Therefore, time learning weight (TLW) proposed in this paper is used improve correlation LSTM network. Fusion Activation Function (FAF) resolve gradient disappearance. Learning Factor Adaptation (LFA) Momentum Resistance Weight Estimation (MRWE) are accelerate convergence global search capabilities. Finally, synthesizes improvement proposes...

Modeling accuracy of DC-DC converters may deviate largely in the presence different variation levels constant power loads (CPLs), hence is well acknowledged as a main hurdle for design advanced model-driven control strategies literature. Aiming to enhance bus voltage regulation performance buck converters, model-free deep reinforcement learning (DRL) strategy proposed this brief. Firstly, Markov Decision Process (MDP) model and Q network (DQN) algorithm are utilized stabilization issue...

In the primary control layer of DC microgrids, engineers usually select gains with a robust design strategy (i.e., worst case study), aiming to ensure stable operation system presence large-signal disturbances. However it will inevitably result in degraded nominal performance. To this end, compromise between dynamic performance and stability appears how reconcile is challenging task. context, we propose novel adaptive pursue balance above two properties. Firstly, higher-order sliding mode...

The reinforcement learning (RL) control approach with application to power electronics systems has become an emerging topic, while the sim-to-real issue remains a challenging problem as very few results can be referred in literature. Indeed, due inevitable mismatch between simulation models and real-life systems, offline-trained RL strategies may sustain unexpected hurdles practical implementation during transfer procedure. In this article, methodology via delicately designed duty ratio...

This paper is aiming to address a decentralized dynamic safety control issue for battery energy storage system in DC microgrids. A novel barrier function (DCBF) based on nonlinear disturbance observers devised the converter restrain current value while ensuring bus voltage regulation. First, DCBF dependent load variation built order of each unit triggered by sudden changes. Second, composite law, incorporating via quadratic program (QP), achieves objective constraint large-signal stability...

In high-power electronic industrial systems, constant power loads (CPLs) can affect the system performance or even cause instability due to its inherent negative impedance characteristics. Regarding this control issue, model predictive (MPC) methods are generally applied dc/dc converters feeding CPLs, aiming achieve satisfactory performance. However, they mostly based on a fixed horizon design, whose optimal may be far from desired one change of operating conditions, e.g., different...

The traditional photovoltaic (PV) forecasting method depends on sufficient historical data (PV power station generation and numerical weather prediction meteorological data), which is not suitable for a newly built PV plant. In order to calculate the array irradiance predict power, physical approach based solar inclined surfaces proposed. This selects three decomposition models four transposition be combined into 12 combination models. Furthermore, spectral response, incidence angle, soiling...

The negative impedance characteristic of constant power loads (CPLs) can cause DC bus voltage fluctuations or even microgrid instability. In this paper, a nonlinear synergetic control method is used to guarantee the stability microgrids with CPLs. First, regulate capacitor its desired value, macro-variable defined using theory. Then, due sensitivity system parameters input voltage, CPL, and resistive load, also enhance overall performance, an improved introduced by adding integral action....

For the dc/dc boost converter feeding constant power loads (CPLs), improving its dynamic characteristics and guaranteeing output voltage stability have aroused much attention from electronics society in recent years. This article proposes an optimal regulation strategy by fusing a robust stabilization with deep reinforcement learning (DRL). First, employing higher-order sliding mode observer (HOSMO) to estimate uncertainties of system, then we are able realize fast performance recovery...

The global robust output voltage regulation problem for dc-dc buck converter systems is investigated by means of designing a sampled-data almost disturbance decoupling (ADD) control methodology. Aiming to effectively restrain the fluctuation caused load variations, physically realizable controller proposed employing linear discrete-time observer generate reliable state-reconstructed information under conception sensorless control. By virtue feedback domination approach, strategy able achieve...

Power electronics, a critical component in modern power systems, face several challenges control design, including model uncertainties, and lengthy costly design cycles. This paper is aiming to propose Large Language Models (LLMs) based multi-agent framework for objective-oriented electronics. The leverages the reasoning capabilities of LLMs workflow develop an efficient autonomous controller process. LLM agent able understand respond high-level instructions natural language, adapting its...

Based on the ratio of size feasible region constraints to a constrained optimization problem, we propose new constraint handling approach improve efficiency heuristic search methods in solving problems. In traditional classification solution candidate, it is either or an infeasible solution. To refine this classification, concept about relative feasibility degree candidate proposed represent amount by which ‘feasibility’ exceeds that another candidate. Relative based selection rules are also...

Abstract High PV penetration into DC microgrids could bring serious stabilization challenges for power electronics engineers, as renewables are accessible to bus voltage oscillation, hence leading degradation of quality and even system collapse. This article addresses the issue a microgrid with high rate, in which decentralized composite generalized predictive control strategy is designed. First, disturbance observer designed deal lumped uncertainties both intermittency variation constant...

Accurately forecasting photovoltaic (PV) power is a prerequisite for dispatching of systems, which also crucial to grid stability. In this study, an evaluation different solar irradiance on inclined surfaces models in the short-term PV was reported. First, combinations decomposition and transposition were used estimate array from measured global horizontal irradiance. Then model proposed according performance inverter, applied obtain final forecasts. Finally, numerous data variety climates...

As a typical switching power supply, the DC-DC converter has been widely applied in DC microgrid. Due to variation of renewable energy generation, research and design control algorithm with outstanding dynamic characteristics significant theoretical practical application value. To mitigate bus voltage stability issue microgrid, an innovative intelligent strategy for buck constant loads (CPLs) via deep reinforcement learning is constructed first time. In this article, Markov Decision Process...

For enhancing the capacity to optimize control performance of generalized predictive (GPC) method, this paper proposes an adaptive horizon mechanism design methodology based on deep reinforcement learning (DRL). To handle systems with presence mismatched disturbances, a baseline and secure environment for DRL training is established by constructing offset-free GPC framework. Furthermore, proposed method incorporates multi-objective reward function into twin-delayed deterministic policy...

Aiming at the problems of traditional maximum power point tracking (MPPT) algorithm for photovoltaic systems, such as poor accuracy, long convergence time, large steady-state oscillation, and difficulty converging to global point. This paper presents a reinforcement learning (RL) method based on deep Q network (DQN) improve optimize MPPT algorithm. In this method, problem systems is transformed into Markov decision process (MDP), circuit composed cell DC-DC boost converter set training...

To accommodate constant power loads (CPLs) with varying degrees of disturbances levels in dc microgrid systems, the adaptability existing robust control strategies should be guaranteed. Regarding this issue, article proposes a novel parameter self-configuration mechanism based on deep reinforcement learning (DRL) an effort to achieve convincing balance between robustness and large-signal stabilizer. First, recent developed nonsmooth composite controller fractional order fine-tuning factor is...

A memory based differential evolution algorithm is adapted in this paper to solve Dynamic Constrained Optimization Problems. The approach on a mechanism utilize the useful past information problem special characteristics. hybrid scheme combined short-term and long term adopted reuses best feasible individual relaxed found before changes. Moreover, an IATM method balancing feasibility, change detection trial vector, update operations are handling changes constraint or objective. Finally, was...

In this paper, a scenario generation method based on Kriging model is proposed to describe the wind uncertainty in Energy Internet. It uses "forecast bin" generate large number of samples. used estimate operational cost corresponding scenario. Then, an importance sampling select This considered power and influence objective function. Finally, Internet with ten units verify effectiveness approach. Compared probability analysis method, it can see that has better economy, stability reliability problem

The energy storage system (ESS) is used to limit the impact of renewable source (RES) volatility in DC microgrids. Fluctuations occur on both and load sides actual working conditions. simplified model a typical autonomous microgrid structure introduced, considering ESS as dispatchable units system. A finite-time cooperative control strategy proposed this paper for large-signal stability by coordinating convergence time primary layer secondary layer. controller also dynamically adjusts power...