Short-Term Load Forecasting (STLF) is the most appropriate type of forecasting for both electricity consumers and generators. In this paper, STLF in a Microgrid (MG) is performed via the hybrid applications of machine learning. The proposed model is a modified Support Vector Regression (SVR) and Long Short-Term Memory (LSTM) called SVR-LSTM. In order to forecast the load, the proposed method is applied to the data related to a rural MG in Africa. Factors influencing the MG load, such as various household types and commercial entities, are selected as input variables and load profiles as target variables. Identifying the behavioral patterns of input variables as well as modeling their behavior in short-term periods of time are the major capabilities of the hybrid SVR-LSTM model. To present the efficiency of the suggested method, the conventional SVR and LSTM models are also applied to the used data. The results of the load forecasts by each network are evaluated using various statistical performance metrics. The obtained results show that the SVR-LSTM model with the highest correlation coefficient, i.e., 0.9901, is able to provide better results than SVR and LSTM, which have the values of 0.9770 and 0.9809, respectively. Finally, the results are compared with the results of other studies in this field, which continued to emphasize the superiority of the SVR-LSTM model.

Nowadays, since energy management of buildings contributes to the operation cost, many efforts are made to optimize the energy consumption of buildings. In addition, the most consumed energy in the buildings is assigned to the indoor heating and cooling comforts. In this regard, this paper proposes a heating and cooling load forecasting methodology, which by taking this methodology into the account energy consumption of the buildings can be optimized. Multilayer perceptron (MLP) and support vector regression (SVR) for the heating and cooling load forecasting of residential buildings are employed. MLP and SVR are the applications of artificial neural networks and machine learning, respectively. These methods commonly are used for modeling and regression and produce a linear mapping between input and output variables. Proposed methods are taught using training data pertaining to the characteristics of each sample in the dataset. To apply the proposed methods, a simulated dataset will be used, in which the technical parameters of the building are used as input variables and heating and cooling loads are selected as output variables for each network. Finally, the simulation and numerical results illustrates the effectiveness of the proposed methodologies.

Power transformers usually confront various mechanical and electromagnetic stresses during an operation that may lead to defects in their windings. The short circuit in the windings is one of those severe defects. Early detection of short-circuits is necessary as extra heating in the shorted location can lead to progressive damage in windings insulation. Frequency response analysis (FRA) is a well-known method to diagnose short-circuits in transformers. Despite the accuracy of FRA, the interpretation of the obtained frequency response traces (FRTs) is still an intricate task. Due to the unknown impact of faults on FRTs, extracting efficient features from such traces is necessary for the interpretation of transformer's frequency response. In this article, an isometric feature mapping (Isomap) is used as a nonlinear dimensionality reduction technique to locate interturn faults in transformer windings due to its capability of capturing the nonlinear phenomena in FRT of power transformers. It is revealed that, after constructing the isometric mapping for a transformer, there is no need for any expertise to detect fault location even in nondirect (high impedance) short-circuits. In other words, it can be the first step for the automated interpretation of FRA of power transformers.

Accurate and fast fault detection in transmission lines is of high importance to maintain the reliability of power systems. Most of the existing methods suffer from false detection of high-impedance faults. In this paper, the transfer function (TF) method is introduced to evaluate the effect of impedance and location of faults by analyzing the voltage and current signals in the frequency domain. Interpretation of the results of the TF method is considered as a weakness of this method. In order to alleviate this problem, a convolutional neural network (CNN) and the hybrid model of deep reinforcement learning (DRL) are utilized to identify and locate single-phase to ground short circuit faults in transmission lines. Single-phase to ground short circuit faults with various fault impedances are applied on an IEEE standard transmission line system. Then, the TF traces are calculated and are collected as input datasets for the proposed models. The fault location results for each network are evaluated via various statistical performance metrics such as correlation coefficient (R), mean squared error (MSE), and root mean squared error (RMSE). The R-value of the CNN and DRL models in fault identification is presented as 96.12% and 98.04%, respectively. Finally, in the early detection of single-phase to ground short circuit fault location (high impedance), the results revealed the efficiency of the DRL model with R=96.61% compared to CNN with R=95.21%.

Early fault detection in power electronic systems (PESs) to maintain reliability is one of the most important issues that has been significantly addressed in recent years. In this article, after reviewing various works of literature based on fault detection in PESs, data mining-based techniques including artificial neural network, machine learning, and deep learning algorithms are introduced. Then, the fault detection routine in PESs is expressed by introducing signal measurement sensors and how to extract the feature from them. Finally, based on studies, the performance of various data mining methods in detecting PESs faults is evaluated. The results of evaluations show that the deep learning-based techniques given the ability of feature extraction from measured signals are significantly more effective than other methods and as an ideal tool for future applications in the power electronics industry are introduced.

Prediction of building energy consumption plays an important role in energy conservation, management, and planning. Continuously improving and enhancing the performance of forecasting models is the key to ensuring the performance sustainability of energy systems. In this connection, the current paper presented a new improved hybrid model of machine learning application for forecasting the cooling load (CL) and the heating load (HL) of residential buildings after studying and analyzing various types of CL and HL forecasting models. The proposed hybrid model, called group support vector regression (GSVR), is a combination of group method of data handling (GMDH) and support vector regression (SVR) models. To forecast CL and HL, this study also made use of base methods such as back-propagation neural network (BPNN), elastic-net regression (ENR), general regression neural network (GRNN), k-nearest neighbors (kNN), partial least squares regression (PLSR), GMDH, and SVR. The technical parameters of the building were utilized as input variables of the forecasting models, and the CL and HL were adopted as the output variables of each network. All models were saved in the form of black box after training and initial testing. Finally, comparative analysis was performed to assess the predictive performance of the suggested model and the well-known basic models. Based on the results, the proposed hybrid method with high correlation coefficient (R) (e.g. R=99.92% for CL forecasting and R=99.99% for HL forecasting) and minimal statistical error values provided the most optimal prediction performance.