​Because the innovation project uses artificial intelligence, it is expected to be in a constant process of improvement and improvement.

Neoenergia is developing an innovative project in partnership with the Innovation Center of Iberdrola – the company's parent company – located in Qatar, in the Middle East. The initiative uses artificial intelligence and data analytics in the application of an algorithm capable of predicting possible failures in telecommunications and smart grid equipment in the power distribution network. The result has an analysis accuracy of more than 90% in occurrences that can occur within 24 hours, ensuring greater availability of devices and enhancing reliability in power supply.

The innovation is being applied in the Intelligent Network Management Center (Cegri), a site that manages all telecommunications and smart grid operations of Neoenergia distributors,in addition to monitoring the digital equipment used in the operations of power networks, such as reconnecters, meters, sensors and intelligent transformers. Cegri monitors more than 75,000 devices and collects data that reaches 10 gigabytes per day. What the algorithm does is turn data into information to contribute to decision making.

For the development of the technology, four parameters were used. The first is packet loss - every five minutes, the system sends a message to the devices, to check the functioning of the quality standards. This information exchange cycle between the system and the assets is continuously tracked and, in case of failure, registration is made. Similarly, the response time is another evaluated parameter. It allows you to check the interval between sending and receiving the response from the remote command that was given to the equipment.

There are also parameters that check the availability and connectivity of network assets, which assess whether they are operating normally. "These criteria are the basis for the operation of the algorithm. With the analysis of the four, the technology can evaluate the operation of the equipment and indicate if any of them are likely to fail," explains João Vyctor Garcia,project manager and engineer of Planning and Standardization of Smart Grids at Neoenergia. The advance prediction varies from one hour, when the accuracy in identifying the failure can reach 98%, and 12 to 24 hours, when the algorithm can be accurate 90% of the time.

The information allows the Neoenergia team to pre-cause possible failures and make the appropriate interventions before they happen. So you can send teams to the field to check the situation or act remotely when possible. "This is important in allowing the constant availability of equipment and contributing significantly to investments in predictive maintenance, which reinforces the quality of deliveries to customers," says Daniel Picchi, Manager of Standardization and Implementation of Smart Grids at Neoenergia.

For the technical director of the Iberdrola Middle East Innovation Center, Javier Hernandez, the project stands out for its ability to develop custom fault analysis algorithms. "This first stage of the project was able to help in understanding the relationships between the different devices of the network, being able to unveil connections not evident, sometimes hidden, behind the networks of telecommunications service providers. Predictive maintenance of Smart Grid systems enables companies to proactively manage impending incidents, reduce operating expenses, and increase network performance,"he says.

The innovation brought by the algorithm presents yet another differential, allowing the separate analysis of each equipment and the relationship between them. Thus, if an asset essential to the functioning of others is at risk of failure, the system can make the correlation and notify the situation. All this happens automatically and has an integrated dashboard, in which you can view all occurrences in a practical way. The management and monitoring panel is in the implementation phase with Cegri and, by the end of May, will already be fully in use by the team.

Because the innovation project uses artificial intelligence, it is expected to be in a constant process of improvement and improvement. In addition to being used in the Cegri database, the algorithm has the potential to be incorporated into other areas of Neoenergia, needing to only suit the new environment.

Initially operated on the supercomputers of the Iberdrola Innovation Center in Qatar, the algorithm now runs on Neoenergia's servers. This was possible due to the installation of Nutanix's hyperconvergence tool, which allows unifying the company's technological structures. Thus, it is possible to reduce by up to 60% the infrastructure that is within a conventional data center, optimizing and automating certain technology processes.

In the Innovation Center of Iberdrola are carried out several projects of interest to the company, focusing on three priority areas: Smart Grids, which develops advanced measurement projects, monitoring and sensing, and cybersecurity; Integration with renewables, responsible for the control of microgrids, distributed power generation and power system modeling; and Energy Management, which accounts for energy efficiency and data analysis.