• Energy Research

The protection of a microgrid containing inverter- based distributed generators (IBDGs) presents several problems if traditional techniques which rely on the current (fuses and overcurrent relays) are used. A possible solution to these problems is the use of a new type of the relay which takes advantage of the enhanced processing techniques and communication infrastructure, both of which are recently becoming available for power networks application. This paper proposes a new communication-based protection scheme for isolated microgrids where a data mining approach is used to identify the relay settings and parameters. A feature selection technique is implemented to help identify the most relevant electrical features required for the fault detection and to establish the best communication strategy to use between relays. The proposed approach is tested using a MATLAB simulation of a facility scale isolated microgrid embedded with IBDGs. The results show that a differential protection scheme that relies on symmetrical components is the most effective strategy for protecting microgrids with IBDGs.

Smart grids have become one of the important and challenging topics due to the numerous benefits it can bring to the power system. In this context, distributed generation (DG) is expected to play a significant role. The smart grid can have multiple configurations depending on the smart grid operating strategy and system conditions. In smart grids, DG could be operated either grid connected or islanded. Such flexible and variable configuration results in variable fault current levels which could impact the operation of the existing protective devices on the distribution system. In this paper, it is proposed to optimally size thyristor-controlled impedance (TCI) of both inductive and capacitive type to manage the fault current levels under different smart grid configurations. The salient benefit is to avoid damage and delayed operation of protective devices due to the variability in fault currents with synchronous-based DG. The problem is formulated as a nonlinear programming (NLP) problem and the optimum size and type of the TCI is determined using particle swarm optimization (PSO). Results show that by optimally locating and sizing TCI, fault current levels under various smart grid configurations can be managed and thus avoiding protective device coordination failure and damage.

Tree-based ensemble learning has received significant interest as one of the most reliable and broadly applicable classes of machine learning techniques. However, thus far, it has rarely been used to model and evaluate the drivers of energy consumption in buildings and as such its performance and accuracy in this field have yet to be properly tested or fully understood. The goal of this paper is to evaluate the performance of three ensemble learning algorithms in modelling and predicting the heating and cooling loads of buildings, namely (i) random forests, (ii) extremely randomized trees (extra-trees), and (iii) gradient boosted regression trees. Results show that the tested algorithms outperform the ones proposed in the recent literature, with gradient boosting improving on the prediction accuracy of the second best-performing algorithm by an average of 14% and 65% for the heating and cooling loads, respectively.

Abstract This paper presents an application of term frequency (TF) as a means of identifying useful trends from text documents. Of particular interest is the relationship between publication patterns, as characterized by TF, and the underlying technological developments. To demonstrate the usefulness of our approach, a case study on distributed generation (DG) was conducted. Important sub-domains of DG research were identified and the associated TF values were extracted using relevant keywords. The evolution of these values through time helps to highlight key trends in the development of DG-related technologies.

High concentrations of induction motor loads can impose stress on transmission and distribution systems, leading to voltage instability in some situations. Properly sized and coordinated reactive power sources will provide for improved operation. We present a strategy for finding an optimal mix (type, size, and location) of dynamic shunt reactive compensation devices. The planning strategy is subject to satisfying steady-state, dynamic and transient performance criteria such as fault-induced delayed voltage recovery limits, as well as criteria related to single ( N –1) contingency and load disturbance events. Shunt reactive power compensation devices considered include mechanically switched capacitor banks, static reactive power compensators and static synchronous compensators. The proposed strategy employs a large number of multitimescale time-domain simulations suitable for use with high performance computing clusters and a genetic algorithm to solve the mixed-integer nonlinear programming formulation using parallel computation capabilities. The method is applied to a New England IEEE 39-bus system with assumed high penetration of induction motors. A comprehensive study shows that performance enhancement and significant cost reduction can be achieved using an optimum combination of various shunt compensator technologies.

A deregulated electricity market is one of the keystones of up-and-coming smart grid deployments. In such a market, forecasting electricity prices is essential to helping stakeholders with the decision making process. Electricity price forecasting is an inherently difficult problem due to its special characteristics of dynamicity and nonstationarity. In our research, we use an Artificial Neural Network (ANN) model on carefully crafted input features for forecasting hourly electricity prices for the next 24 hours. The input features are selected from a pool of features derived from information such as past electricity price data, weather data, and calendar data. A wrapper method for feature selection is used in which the ANN model is continuously trained and updated in order to select the best feature set. The performance of the proposed method is evaluated and compared with the published results of the state-of-the-art Pattern Sequence-based Forecasting (PSF) method on the same data sets and our method is observed to provide superior results.

In this paper, the development of electricity price and demand forecasting, with the emergence of demand response programs, is investigated. Short Term Load/Price Forecasting (STL/PF) is performed for an electricity market that offers Demand Response (DR) Programs. The change in the forecasting errors, of both electricity price and demand, over years of inactive and active DR is monitored. Commonly used prediction methods, namely; Least Squares-Support Vector Machines (LS-SVM), and Random Forests (RF), are used for forecasting, to ensure the generality of the results. The Australian National Electricity Market (ANEM), specifically Victoria region, is used as a subject case study. It was concluded that adding DR programs decreases the volatility of electricity price, with no validated effect on demand.