Spanish energy company Endesa has launched the artificial intelligence-driven project PASTORA – Preventive Analysis of Smart Grids with Real Time Operation and Renewable Assets Integration, as part of its efforts to digitalise the distribution grid.
PASTORA draws on big data comprised of millions of data points gathered from smart meters and sensors installed at various locations in the grid. These will be analysed with artificial intelligence and machine learning algorithms to develop predictive models to anticipate the behaviour of the grid and thus to improve real-time control and the undertaking of preventive maintenance and ultimately to enhance the quality of service for customers.
With such tools the integration of intermittent renewable energies and distributed resources such as electric vehicles should be facilitated and investments and maintenance in the grid can be optimised.
“The aim is to try and anticipate any type of incident that may occur on the grid, based on all of the knowledge that big data offers,” explains PASTORA project director Susanna Carillo Aparicio.
Smartcity Málaga testbed
Artificial intelligence is finding increasing application in the energy sector for use cases ranging from automation and control to weather and demand forecasting. However, its full potential is generally being underestimated, according to a new study from Capgemini Research Institute.
PASTORA, a 31-month, €2.8m project was launched in February and is being run in the Smartcity Málaga Living Lab, which offers a real-life environment for innovation around smart grid and other technologies.
The project consortium headed by Endesa includes the technology providers Ayesa Advanced Technologies, Ormazabal Media Tensión and Ingelectus Innovative Electrical Solutions, the University of Seville research association AICIA and the University of Granada.
PASTORA builds on an earlier project, MONICA – Monitoring and Advanced Control of MV and LV distribution grids, which was launched in 2015 to develop technologies to improve situational awareness on the grid. Based on data gathered from the approximately 15,000 customer smart meters and 750 sensors deployed in 56 MV/LV transformer substations in Smartcity Málaga, the project resulted in the development and implementation of one of the first phasor measurement devices for the medium and low voltage grids globally – similar to those implemented at transmission level.
“MONICA has shown us how the medium and low voltage networks will operate in the smart cities of the future,” says Carillo. “For us the electricity network of the future is a digital network that gives us information on what is happening with the energy at every point.”
AI in customer service
Endesa is no stranger to the application of artificial intelligence and appears to be one of a few – 15% in the Capgemini study – that has been developing multiple use cases across the company. One of these is virtual voice assistants for customer service.
An application has been introduced with Amazon’s Alexa which enables customers with any of the Echo wireless speakers to obtain information about their energy use and tailored energy advice. This is combined in a customer service channel with a home automation control offering in an all-inclusive service aimed to empower customers and give them more decision-making capacity on energy and bill saving.
The company says it intends to continue advancing the development up to the level of customers being able to carry out the same dealings or queries as on its website or app. It is anticipated also to be able to open up communication with the contact centre if additional service is required.
Another related application is the use of IBM’s Watson artificial intelligence to assist customers both in the chat and on the telephone, with a “significant percentage” now attended to autonomously. Further customer facing applications using artificial intelligence that are under development include models to reduce non-payments, robotic process automation, a complaints classifier based on text recognition and time demand forecasting.
An application to detect fraud has been running since 2017.
Another area of the business in which artificial intelligence applications have been introduced is generation, particularly for predictive purposes. These include detection of faults in plant equipment and in low and medium voltage switches. Others include supervision cameras with artificial vision, a global reporting system and support for generation plant operation and maintenance.
The Capgemini study estimates potential savings in the energy sector from the use of intelligent automation at scale at between $237bn and $813bn over the next three years. However, the study found a key challenge has proven to be scaling up initiatives. Based on best practices of frontrunners in the field Capgemini recommends that use cases must be selected that fit the organisation’s business strategy, they should be optimised before being scaled and executed by a dedicated team.