This paper proposes a heuristic planning energy management controller, based on Dyna agent of reinforcement learning (RL) approach, for real-time fuel saving optimization plug-in hybrid electric vehicle (PHEV). The presented method is referred to as the Dyna-H algorithm, which model-free online RL algorithm. First, case study, detailed powertrain modeling Chevrolet Volt built, where all control components have been experimentally validated. Four traction operation modes are allowed by...

The short-term load forecasting is crucial in the power system operation and control. However, due to its nonstationary complicated random features, an accurate forecast of behavior challenging. An improved method proposed this article. At first, decomposed into different frequency components varying from low high levels realized by ensemble empirical-mode decomposition algorithm. Then, smooth periodic low-frequency are predicted multivariable linear regression while maintaining efficient...

This paper proposes a multi-agent deep reinforcement learning-based approach for distribution system voltage regulation with high penetration of photovoltaics (PVs). The designed agents can learn the coordinated control strategies from historical data through counter-training local policy networks and centric critic networks. learned allow us to perform online control. Comparative results other methods show enhanced capability proposed method under various conditions.

This paper proposes a novel model-free/data-driven centralized training and decentralized execution multi-agent deep reinforcement learning (MADRL) framework for distribution system voltage control with high penetration of PVs. The proposed MADRL can coordinate both the real reactive power PVs existing static var compensators battery storage systems. Unlike DRL-based methods, our method does not rely on model during stages. is achieved by developing new interaction scheme between surrogate...

A time-variable time-of-use electricity price can be used to reduce the charging costs for electric vehicle (EV) owners. Considering uncertainty of fluctuation and randomness EV owner's commuting behavior, we propose a deep reinforcement learning based method minimization individual cost. The problem is first formulated as Markov decision process (MDP), which has unknown transition probability. modified long short-term memory (LSTM) neural network representation layer extract temporal...

Effective detection of fault in rolling bearings with a limited amount data is essential for the safe operation electric machines. This article proposes novel meta-learning-enabled method machines under varying working conditions data. The diagnosis various cast as few-shot classification problem, which solved using model-agnostic meta-learning-based model. Specifically, meta-learner first trained series interrelated fault-diagnosis tasks conditions. During this stage, gradient-by-gradient...

The accurate estimation of lithium battery state health (SOH) is very important for the safe and stable operation battery. Since user's charging process random, it difficult user to know SOH through segment. In this article, we proposed a method random based on convolutional gated recurrent unit (CNN-GRU). extracts key features adaptively from segments voltage, current temperature curves in CNN-GRU framework realize estimation. Compared with traditional methods, does not need manually select...

With the increasing size of wind farms, impact wake effect on farm energy yields become more and evident. The arrangement locations turbines (WTs) will influence capital investment contribute to losses, which incur reduction production. As a consequence, optimized placement WTs may be done by considering as well components cost within farm. In this paper, mathematical model includes variation both direction deficit is proposed. problem formulated using levelized production (LPC) objective...

The Danish power system has a large penetration of wind power. fluctuation causes high variation in the generation, which must be balanced by other sources. battery storage-based Plug-In Electric Vehicles (PEVs) may possible solution to balance variations systems with penetrations. In this paper, integration plug-in electric vehicles penetrations is proposed and discussed. Optimal operation strategies PEV spot market are order decrease energy cost for owners. Furthermore, application storage...

This paper proposes attention enabled multi-agent deep reinforcement learning (MADRL) framework for active distribution network decentralized Volt-VAR control. Using the unsupervised clustering, whole system can be decomposed into several sub-networks according to voltage and reactive power sensitivity relationships. Then, distributed control problem of each sub-network is modeled as Markov games solved by improved MADRL algorithm, where an adaptive agent. An mechanism developed help agent...

In this paper, an efficient methodology is proposed to deal with segmented-time reconfiguration problem of distribution networks coupled reactive power control distributed generators. The target find the optimal dispatching schedule all controllable switches and generators' powers in order minimize comprehensive cost. Corresponding constraints, including voltage profile, maximum allowable daily switching operation numbers (MADSON), limits, so on, are considered. strategy grouping branches...

An optimal reactive power dispatch strategy is proposed to minimize the total electrical losses of a wind farm (WF), including not only in transmission cables and turbine (WT) transformers, but also inside energy generation systems. The WT uses splitting over stator grid side converter (GSC), which aims loss system, generator, converters, filters. Optimization problems are formulated based on established models limits. A WF carefully designed used for case studies. Wake effect considered...