1
Introduction
Ahmed F. Zobaa
Brunel University London
Trevor J. Bihl
Wright State University
Contents
1.1Introduction
1.2Big Data
1.3Future Power Systems
1.4Book Organization
1.4.1Overview
1.4.2Big Data Application and Analytics in a Large-Scale Power System
1.4.3The Role of Big Data Analytics in Smart Grid Communications
1.4.4Big Data Optimization in Electric Power Systems: A Review
1.4.5Security Methods for Critical Infrastructure Communications
1.4.6Data-Mining Methods for Electricity Theft Detection
1.4.7Unit Commitment Control of Smart Grids
1.4.8Data-Based Transformer Differential Protection
1.5Conclusions
References
1.1 Introduction
As a concept, big data and power systems might appear unrelated; however, the Smart Grid and advances in general computing power have made power systems a data-driven industry. The result of the ability to collect endless data is the emergence of big data. However, power systems are connected to physical devices and critical infrastructure (CI) and thus additional research problems and concerns exist in power system big data.
1.2 Big Data
Big data involves more than the size of the data itself and extends to the complexity and speed at which it is collected. The term big data is frequently defined with vague and self-referencing definitions and naturally big data logically extends from data (Bihl, Young II, & Weckman, 2016). While data are generally any sensed output, big data involves data that are too big, complex, or overwhelming to be analyzed by traditional methods (Bihl, Young II, & Weckman, 2016).
The primary attributes of big data are the 3 “V’s” of volume, variety, and velocity (Bihl, Young II, & Weckman, 2016). While more than 42 attributes have been defined by some researchers in describing big data, the 3 V’s capture the gist of the big data problem (see Shafer, 2017). As attributes, volume relates to the overall size of the data, variety indicates that big data can contain various types of data (text, strings, numbers, etc.) all within one dataset, and velocity indicates that big data is collected in real time (Bihl, Young II, & Weckman, 2016).
Critically, velocity is an attribute frequently associated with big data. Given enough time, any large volume and highly various dataset could eventually be analyzed using traditional methods. However, when these data are continuously being collected, a velocity problem exists whereby the growing size and complexity preclude traditional methods. Thus, advanced analytics and data management methods are both necessary (cf. Gutierrez, Boehmke, Bauer, Saie, & Bihl, 2018; Najafabadi et al., 2015).
1.3 Future Power Systems
Future power systems imply power systems that differ from today’s due to increased decentralization, expanded communication and monitoring abilities, and wider variety of sources (Hebner, 2017). Multiple thrusts exist in power system research to accommodate this future; these include expanding the Smart Grid, increasing penetration of the Internet of Things (IoT), expanding renewable sources, and microgrid considerations.
Expanding penetration of the Smart Grid is not only expected but already underway (Amin & Wollenberg, 2005). Along with the Smart Grid comes a multitude of logged customer and power grid data which can be analyzed to find power theft (Jiang et al., 2014) and improve operating conditions of the grid at large (Fan et al., 2013). The IoT further expands upon the Smart Grid by enabling communication with any and all devices (Gubbi, Buyya, Marusic, & Palaniswami, 2013). An IoT-enabled power grid thus allows the monitoring of the CI while posing both big data and security problems (Sajid, Abbas, & Saleem, 2016).
Increasing decentralization through more microgrids and nanogrids can be also expected in the future power grid. While these have the ability to provide local resiliency (Hebner, 2017), they introduce uncertainty in larger grid planning (Khodaei, Bahramirad, & Shahidehpour, 2015). Added to this is the expected increase in the use of renewables, which also increase power system planning problems due to their general availability uncertainty (Atwa, El-Saadany, Salama, & Seethapathy, 2010; Polatidis, Haralambopoulos, Munda, & Vreeker, 2006).
1.4 Book Organization
To examine these problems, this book examines various intersections of big data and future power systems. For this goal, this book provides nine chapters, including the introduction, which focuses on the primary themes of big data in future power systems. Overall, this book discusses big data analysis methods, big data problems in future power systems, IoT concerns, security concerns related to big data, and various associated complexities.
1.4.1 Overview
This book is organized as follows:
•Chapter 2 discusses analytics and machine-learning methods in general and those applicable to big data in power systems.
•Chapter 3 discusses additional big data analytics relative to Smart Grid components.
•Chapter 4 discusses optimization methods which are suitable for big data models in power systems.
•Chapter 5 extends the discussion of Chapter 4 by considering various cyber security issues that exist in IoT-enabled future power systems.
•Chapter 6 discusses electricity theft detection and mitigation which is enabled by big data collection from the Smart Grid.
•Chapter 7 discusses renewable energy planning concerns which are associated with planned future power systems that have high renewable penetration.
•Chapter 8 discusses transformer protection methods which are enabled by big data collection on transformers.
1.4.2Big Data Application and Analytics in a Large-Scale Power System
To analyze big data, a variety of machine-learning methods are generally employed. Machine learning is broadly synonymous with pattern recognition, statistics, and data mining (Hand, 1998; Mannila, 1996). However, due to the emergence of big data, a variety of new methods have recently emerged, e.g., large-scale neural network known as “deep learning,” which are capable to analyze and exploit the bigness of big data. While these methods have achieved significant advancements in image recognition, they have begun to see use in power system big data analysis (see LeCun, Bengio, & Hinton, 2015).
1.4.3The Role of Big Data Analytics in Smart Grid Communications
Because a Smart Grid can be described as a huge sensor network, with a lot of intelligent devices, the growth in the number of devices will produce a considerable amount of measured data. How to quantify and to analyze these data to enhance grid operation arises as one big concern. Advances of the Smart Grid promise to give operators and utilities a better understanding of customer behavior, demand consumption, weather forecast, power outages, and failures. However, it is vital to quantify the volume of sampled data to take advantage of them. Therefore, this chapter aims to characterize and to evaluate the emerging growth of data in communications network applied to Smart Grid scenario. A future active distribution system will serve as an example to demonstrate the data requirements for monitoring and controlling the grid.
1.4.4Big Data Optimization in Electric Power Systems: A Review
Traditional data-processing applications have difficulties operating effectively due to the complexity, velocity, and voluminosity of big data. This chapter presents a review of big data optimization problems in electric power systems. The chapter starts with scientometric mapping methods that show the variety and diversity of large-scale optimization problems in today’s power system networks. An electrical grid power system could be categorized into generators which provide the required electric power, transmission systems that carry the electricity from the generating units, and distribution systems that feed the power to nearby industries and homes. The optimization issues such as logistics optimization in power system, as well as some optimization techniques including non-smooth, nonconvex, and unconstrained large-scale optimization are presented. Add...
