Page 363 - From Smart Grid to Internet of Energy

Big data, privacy and security in smart grids Chapter 8 327

FIG. 8.7 Big data analytics requirement in smart grid applications.

distribution and consumption is equipped with IEDs and sensor networks in this
convergence. The big data analysis and decision support tools operated on smart
grid plane are illustrated in Fig. 8.7 where each requirement has been noted. The
big data analytics are operated at seven steps in smart grid applications as data
acquisition, data transmission, storage, cleaning, preprocessing, integration,
feature and model selection stages [3, 25].
The overall schematic of smart grid infrastructure shown in Fig. 8.7 is
completely depended on ICT operations since all sections should be instantly
monitored and controlled. The two-way communication is indispensable for
each application and Big Data analytic requirement depicted on the lower plane
of figure. Each application presented in Fig. 8.7 should be dealt with seven steps
of big data analysis framework. The generated data of thousands of smart
devices should be inherited and sampled in a few seconds. The generated data
could be brought by any section of generation, transmission, distribution or con-
sumption levels of smart grid network. The generation level data sources can be
distributed generation sources, forecasting measurements, power plants, RES
plants and so on. The enormous number of data sources are located at consumer
side since many nodes are comprised by each customer which have also differ-
ent IEDs in their reserves. The residential data sources are mostly generated by
AMIs, smart meters, home energy management systems, electric vehicles,
micro sources, and by many more sensors used for surveillance, forecasting,
and management demands.
The data acquisition process is performed in three steps as data capture, data
transmission and data preprocessing in smart grid big data analysis. The gener-
ated data of sensors and nodes are captured by centralized of distributed agents.
The acquired data are transmitted to a central Hadoop cluster for storing and
local master nodes are comprised. The stored raw data are transferred to data
storage system for initial processing. The data integration process is required
at step since the inherited data are provided by a wide diverse of devices and
different file formats or information types are generated. This stage is named