Big Data Analytics in Smart Grids

Big data analytics is the process of assessing big data to acquire information such as customer preferences and market trends that enable an organization to make better business decisions. There is more data now due to the rapid development of digital technology, with data on the internet now being measured in exabytes and zettabytes. Howard Dresner proposed the idea of acquiring variable information from immense collected data in commercial operations in 1989. However, he termed it ‘business intelligence’ (BI). Engineers now have a new vision to perceive the traditional electrical system and make it smart due to the enormous progress of information and communication technology (ICT). The paper discusses the big data analysis concepts and their application in smart grids.

In smart grids, big data is characterized by various features such as volume, which is the huge amount of data generated, making data sets too large for traditional database technology. Distributed systems in different locations connected by networks and brought together by software could, however, be used. Another characteristic of big data in the smart grid is velocity. Velocity is the speed of the generation of new data and the speed of data movement. There are also various data types one can now use, unlike before, when the focus was made on structured data that neatly fits into rational databases. Veracity is another characteristic of big data in the smart grid and is defined as the data’s trustworthiness or messiness. With large amounts of big data, the data’s accuracy and quality are less trustworthy, challenging the outcome data analysis. The last characteristic is value, which is the ability to get valuable information from the enormous data.

The smart grid is the superabundant source of information that covers data from electricity generation, transmission, distribution, and consumption. The data sources include electrical information from distribution stations, distribution switch stations, electricity meters, and non-electrical information. The data sources may be categorized into measurement data, business data, and external data. Installed sensors and smart meters that indicate the system’s status measure most of the power systems’ operational parameters (Sagiroglu et al., 2016). Smart meters cannot measure external data such as weather conditions and social events even though they affect the operation and planning in power systems. Electric vehicles and plug-in electric vehicles are an emerging component in the electricity market and smart grid. They have become more popular with the electrification movement in the transportation sector.

Wired and wireless infrastructures are the basic types of communication technologies for smart meters. Even though it may face the electromagnetic problem, wireless communication technology enables gathering measurement information from smart meters. Being one of the countries to develop the smart metering infrastructure, Italy has provided smart meters to almost all powerline communication technology customers. The country has installed about 30 million meters and 400,000 secondary substation concentrators. Data analysis is the most crucial stage of the big data processing system. It is the foundation for discovering valuable information and supporting decision making. Data analytics, also known as data mining, is from a general viewpoint the computation process showing relations among variables. However, collected data sets vary in quality in terms of noise, consistency, and redundancy due to the diverse sources.

The common data mining or machine learning algorithms are normally categorized as supervised or unsupervised learning based on whether there is an attached label to all items in the data sets. Data analytics models in the supervised learning algorithm can be trained depending on the given data to analyze how data attributes relate to the corresponding categories of values. For unsupervised learning algorithms, the data analytics model is normally created to realize the possible groups among all the items. The main course of action for data analytics in the smart grid is to derive valuable information for steering the performance and maintenance compared to real time data. The data collected is presented and reserved with data management techniques, and after preparation, the mathematical model is then set up through data mining procedures on the clean data.

The driving forces of the smart grid are carbon emission reduction and environmental stability. The introduction of distributed generator units provides a successful means of utilizing renewable energy, including wind and solar energy. The irregular features of renewable energy raise the power grid’s uncertainty, whose only solution is to use inverter interfaced distributed generators (IIDG) for quality power. In a grid-connected microgrid, a grid blackout or severe weather conditions may trigger an islanding accident, causing technical issues. Like Facebook and Twitter, social media may contain valuable information showing the location and occurrence of power outages. Distribution automation focuses on the system operation at the distribution level, where under the above concept, volumes of operational data are collected for monitoring by the state.

As a result of development in ICT in power systems, there is a collection of large volumes of data through communication infrastructures. Traditional statistical methods such as logistical regression and linear discriminant are considered for mining the link between power system falls and the features derived from raw data (Zhang et al., 2018). The galloping of power lines can be a major threat to the security transmission system since it causes electrical and structural failures. An issue closely related to the safety of the power system is transient stability. Power swing occurs when the oscillation of power flow on transmission lines advance to each other, causing a large disturbance. Power transformers may fail, causing blackouts in power systems. A potential threat to safe operation is the increasing number of aging assets in power systems, which lead to a lot of failure models.

Electric power quality is the frequency, magnitude, and waveform of voltage in power systems and closely associated with the power grid’s safe operation and customer satisfaction. Renewable energy forecasting should be the dominant source of energy for the next generation of power grid but is hindered by the randomness and intermittent features of large-scale utilization of renewable energy sources in a firm way. Load profiling is a way to express electrical consumption behavior, which represents the time domain for capital planning and load forecasting. Non-intrusive load monitoring, also referred to as load disaggregation, aims to sort out the overall load profiles into individual appliances’ energy consumption. There are increasing researches on big analytics, and smart grids through the applications that have been deployed are less. The techniques can generate implications in reality, but for this to take place, many issues have to be addressed.

The big data in the smart grid and corresponding analysis methods have been discussed and examined where data containing valuable information is collected from various systems. The advancement of ICT in power systems leads to fast and well-organized data analysis. The interactive communication with customers leads to a potential solution for more exact demand response though it raises difficulties in consumption behavior analysis. Data analytics applications require synergy among experts from various fields and strategic visions for the top designs. That is especially since the application is a comprehensive and complicated field involving computer science, mathematics, ICT, etc.

References

Sagiroglu, S., Terzi, R., Canbay, Y., and Colak, I. (2016). Big Data Issues in Smart Grid Systems. In: 2016 IEEE International Conference on Renewable Energy Research and Applications (ICRERA), Birmingham, pp 20-23

Zhang, Y., Huang, T. and Bompard, E. (2018, August 13) Big data analytics in smart grids: a review https://link.springer.com/article/10.1186/s42162-018-0007-5