This paper explores how electricity systems may evolve in the 21st century. The paper focuses on some fundamental challenges facing the utilization of electricity today and for years to come. Paralleling the challenges, several directions of how new solutions may emerge are suggested. In this context, some new approaches to manage power system development and deployment are outlined.

The impacts of uncoordinated charging of electrical vehicles (EVs) under high penetration on distribution transformers is studied. It is shown that it may cause prolonged overload condition leading to accelerated assets loss of life and increased hazard of failure. To mitigate the impact, a fuzzy logic-based algorithm is devised. It provides a performance index for distribution grid operator using four main inputs: 1) EV battery state of charge; 2) required state of charge for the next trip; 3) estimated time of EV departure; and 4) customer comfort level. A decision-support tool is proposed by which the distribution grid operators can utilize the output of the fuzzy systems to receive a suggestion whether to delay the charging of each EV and pay the incentive to the EV owner. The data for the city of College Station, Texas, USA including temperature, price of electricity and load profile are collected from various sources to simulate different use cases. The example illustrates how the proposed EV management approach could mitigate the impact of EV charging on the transformer loss of life and hazard of failure. The main advantage of the proposed approach is the low cost and ease of design implementation. The information that needs to be sent from the consumer to the distribution system operator is minimized, which helps in maintaining costumers' privacy.

An end-to-end supervised learning method is proposed for fault detection in the electric grid using Big Data from multiple Phasor Measurement Units (PMUs). The approach consists of preprocessing steps aimed at reducing data noise and dimensionality, followed by utilization of six classification models considered for detecting faults. Three of the models were variants of Convolutional Neural Network (CNN) architectures that consider a single type of measurement (voltage, current or frequency) at all PMUs or all types together also at all PMUs. CNN based models were compared to traditional methods of Logistic Regression (LR), Multi-layer Perceptron (MLP) and Support Vector Machine (SVM). Evaluation was conducted on two-year data measured by PMUs at 37 locations in a large electric grid. The response variable for classification were extracted from the grid-wide outage event log. Experiments show that CNN-based models outperformed traditional methods on one year out-of-sample outage detection over the entire grid.

Life-cycle assessment of high voltage (HV) circuit breakers (CB) in transmission systems, if efficiently done, can lead to an optimal decision on when, where, and how to perform maintenance. This paper elaborates a new approach on the identification of CB's deterioration/recovery states, i.e., the so-called life-cycle assessment, using its control circuit condition monitoring data. Reliability-oriented performance indicators, which can assess the condition of different physical parts of an HV CB in real time, are introduced first. Then, a quantitative methodology to define the probability of the CB falling into each class of deterioration/recovery states i.e., healthy, vulnerable, troubled, and failed, is proposed. Using this approach, maintenance decisions can be effectively made on different parts of an HV CB, the impact of maintenance can be quantified, and system-wide maintenance optimization with respect to the condition-based distinction of CBs can be made possible. Field condition monitoring data recorded at different time intervals during the CB operation is utilized to evaluate the applicability and effectiveness of the proposed approach.

Event classification is one of the central components of automated disturbance analysis based on phasor measurement unit (PMU) measurements. Obtaining high-quality event labels remains a challenge for supervised learning-based classification of local and system-wide events in power grids due to its labor-intensive requirement. We present a sensitivity study considering rapidly refined, partially, and fully inspected event labels that leads to evidence that hierarchical convolutional neural networks (HCNNs) outperform traditional classification models regardless of the quality of the available event labels. It is demonstrated that performance similar to the one obtained using entirely domain-driven labeling can be achieved as long as the involved expert does not mislabel more than ~5% of the event data captured by PMU measurements.