Abstract
This paper deals with an optimal operation of a microgrid in the electricity market and presents the communication between the distribution market operator and microgrid operator. The distribution market operator controls and manages the electricity market established in the distribution level, determining the amount of both electricity price and power exchange between market participants. The microgrid operator is able to purchase active and reactive power from the local distribution market. An effective short-term scheduling of the microgrid is implemented to ensure optimal operation. A risk based stochastic model is used to model the prevailing uncertainties such as loads, wind generation, and main-grid availability in a market-based operation framework. Moreover, in this model, a linearized AC power flow is added to the mathematical formulations to offer a comprehensive solution to the security-constraint operation of the microgrid. The stochastic operation strategy is formulated as a mixed integer linear programming problem. Regarding the uncertainty modeling, the substation equipment failure is modeled with Monte-Carlo algorithm. The effectiveness of the risk-based stochastic method is demonstrated using a microgrid test bed in the presence of demand response resources, dispatchable and wind generation units as well as energy storage system. The results demonstrate demand response program can significantly reduce the operation cost in worst scenarios. Also, it is indicated that the risk averse decisions reduce the risk of experiencing costly scenario. The proposed framework incorporating the distribution market constraints reduces the uncertainty in real-time operation as it can specified the required energy before running the problem. The deviations from assigned energy to the microgrid are penalized through the distribution market operator.

Abstract
Wind power instability and inconsistency involve the reliability of renewable power energy, the safety of the transmission system, the electrical grid stability and the rapid developments of energy market. The study on wind power forecasting is quite important at this stage in order to facilitate maximum wind energy growth as well as better efficiency of electrical power systems. In this work, we propose a novel hybrid data driven model based on the concepts of deep learning-based convolutional-long short term memory (CLSTM), mutual information, evolutionary algorithm, neural architectural search procedure, and ensemble-based deep reinforcement learning (RL) strategies. We name this hybrid model as DOCREL. In the first step, the mutual information extracts the most effective characteristics from raw wind power time series datasets. Second, we develop an improved version of the evolutionary whale optimization algorithm in order to effectively optimize the architecture of the deep CLSTM models by performing the neural architectural search procedure. At the end, our proposed deep RL-based ensemble algorithm integrates the optimized deep learning models to achieve the lowest possible wind power forecasting errors for two wind power datasets. In comparison with fourteen state-of-the-art deep learning models, our proposed DOCREL algorithm represents an excellent performance seasonally for two different case studies.

Abstract
Natural gas will play a key role in the transition to a lower-carbon economy, constituting a natural alternative to coal and acting as a backup resource to the intermittent nature of renewable generation. These energy carriers can be structurally linked together by Power-to-X technologies because of their interaction to increase energy efficiency. For this purpose, this paper proposes an innovative model to optimally manage the electricity and natural gas grids in a cost-efficient manner. In this model, an energy hub has water, electricity, and gas oil as inputs, supplying electric and thermal loads. Besides, the energy hub uses the Power-to-gas (P2G) technology to produce natural gas, selling it to a gas network to reduce the congestion in gas pipelines and the energy hub owner's costs. A demand response program has been also applied in this model to shift the loads from on-peak times to off-peak ones. Various technologies such as energy storage and distributed generation have been used in the modeling to reach the goals targeted by operators. Furthermore, a scenario generation method has been applied to model the uncertainty of wind turbine output. The proposed problem has been finally formulated as mixed-integer linear programming that has been solved under GAMS software by using CPLEX solver to reach the global optimality. The results obtained from simulations demonstrate that the proposed model can significantly reduce the operation cost, while properly alleviating gas network congestion.

Abstract
Energy storage system (ESS) has great importance in saving energy in new power systems. Optimum selection of these elements poses a new challenge to improve the energy management and prevent cost increases in the system. Also, renewable energy resources have been increasingly used in microgrids. The uncertainty and variation of renewable distributed generation (DG) affect the performance of power systems. In this paper, ESS implementations and photovoltaic (PV) power prediction are used to improve voltage/power profile of the system and reduce the total cost of the microgrid. The purpose of this paper is the optimal installation of ESSs in a microgrid to minimize the total cost where quantile nearest neighbour forecasting is utilized for PV output power prediction as an efficient approach. Gathering data of the last samples in time duration can be used for an effective prediction of PV output in this method, which can overcome PV uncertainty due to changes in solar irradiation and other parameters. Artificial neural networks combined with multi-layer perceptron and genetic algorithm are used for optimizing the size and location of ESSs in the system. Simulation results show that the proposed method improves the power profile as 14%, 21% and 28%, relatively to the scenarios of optimal ESS installation without PV prediction, using PV prediction but with no optimal ESS implementation and not using PV-no ESS implementation, respectively. Moreover, the accuracy of the proposed prediction method is more than the gradient-descent and RNN methods by about 12% and 5%, respectively, as shown in the simulation results. Also, the cost reduction of proposed method is enhanced as 24% and 31% relatively to the cases of optimal ESS installation without PV prediction and PV prediction without optimal ESS implementation, respectively.

Abstract
Accurate household profiles (e.g., house type, number of occupants) identification is the key to the successful implementation of behavioral demand response. Currently, supervised learning methods are widely adopted to identify household profiles using smart meter data. Such methods could achieve promising performance in the case of sufficient labeled data but show low accuracy if labeled data is insufficient or even unavailable. However, the acquisition of accurately labeled data (usually obtained by survey) is very difficult, costly, and time-consuming in practice due to various reasons such as privacy concerns. To this end, a semi-supervised learning approach is proposed in this paper to address the above issues. Firstly, 78 preliminary features reflecting the household profiles information are extracted from both time and frequency domain. Secondly, feature selection methods are introduced to select more relevant ones as the input of the identification model from the preliminary features. Thirdly, a transductive support vector machine method is adopted to learn the mapping relation between the input features and the output household profile identification results. Case study on an Irish dataset indicates that the proposed approach outperforms supervised learning methods when only limited labeled data is available. Furthermore, the impacts of different feature selection methods (i.e., Filter, Wrapper and Embedding methods) are also investigated, among which the wrapper method performs best, and the identification accuracy improves with the increase of data resolution.

Abstract
This paper intends to give an effective hybrid planning of distributed generation and distribution automation in distribution networks aiming to improve the reliability and operation indices. The distribution automation platform consists of automatic voltage and VAR control and automatic fault management systems. The objective function minimizes the sum of the expected daily investment, operation, energy loss and reliability costs. The scheme is constrained by linearized AC optimal power flow equations and planning model of sources and distribution automation. A stochastic programming approach is also implemented in this paper based on a hybrid method of Monte Carlo simulation and simultaneous backward method to model uncertainty parameters of the understudy model including load, energy price and availability of network equipment. Finally, the proposed strategy is implemented on an IEEE 69-bus radial distribution network and different case studies are presented to demonstrate the economic and technical benefits of the investigated model. By allocating the optimal places for sources and distribution automation across the distribution network and extracting the optimal performance, the proposed scheme can simultaneously enhance economic, operation, and reliability indices in the distribution system compared to power flow studies.