Press Releases

News

BegiPlant technology, developed with Artificial Intelligence, in use at the Muskiz Petronor plant

15/02/2018

BegiPlant, the groundbreaking fire and smoke automatic detection system developed by the Artificial Vision area of DOMINION, Begirale, has been implemented and is in operation at Petronor's facilities in Muskiz (Bizkaia) since December 2017.

Thanks to this system, which uses Artificial Intelligence (Deep Learning) methods, effective fire and smoke detection is possible thanks to the video surveillance cameras installed in the plant. This is achieved by analyzing in real time the video signals coming from the closed-circuit television and processing them in search of any incident autonomously and in real time, generating alarms instantly.

 

After a first phase of technological demonstration in various fire-fighting fields, as well as the realization of a pilot in several chambers of the Petronor plant, the BegiPlant system was deployed in a total of 60 chambers distributed throughout the production area of the Muskiz plant. The launch of the project in late 2017 coincided with the remodelling of the new control room at plants 1 and 2 of the refinery, as well as the installation of the new Fire Protection System (PCI) which integrates the BegiPlant system.

 

The way in which BegiPlant's tool is developed allows it to operate autonomously, without any interaction with the rest of the plant's systems (except for the CCTV System), as well as being integrated with the plant's fire-fighting mechanisms, protocols and tools.

 

The application of Intelligent Video Content Analysis (VCA) technologies makes it possible to optimize investments in video surveillance and CCTV systems to increase security, thus considering innovative and highly advanced new business models aimed at increasing the reliability and efficiency of critical processes with low investment and reduced operating costs”, explains Javier Ortuondo, Technical Manager of Begirale. “The enormous amount of existing and underutilized information coming from cameras in industrial installations is used to a great extent" Javier concludes.

 

 

 

 

A continuous learning system

 

One of the key points of the tool is its adaptation to the particular environmental variables of each plant, and this is achieved thanks to the learning capacity of the BegiPlant system by continuously increasing its knowledge base. To do this, the system has a quick and easy tool that allows the operator or person responsible for monitoring to review the images and videos of the detected alarms and catalog them so that they can be applied in the continuous learning of the system.

 

“This first implementation of the system has an enormous value for us because, in addition to be an ideal test bench for our technology, it has become the perfect showcase to present our solution in the Oil & Gas market. This “demo” effect has already begun to bear fruit and in recent months other refineries of the Repsol group and plants of other companies in the sector in several countries have shown interest in BegiPlant. All this makes us very positive when it comes to deploying our solution over the next few years," says Iñigo Zorriketa, director of Begirale. This technology, which DOMINION already successfully applies in the railway sector to avoid accidents at level crossings, allows us to improve response times when dealing with an emergency. In addition, another advantage of these tools for early detection of incidents is that it does not require the installation of additional or specific video cameras, but rather provides intelligence to existing infrastructures.

 

"Begirale's main objective is now to enhance the value of the solutions already developed and at the same time to continue designing and developing new products that allow increasing the security of highly critical processes through the use of Intelligent Video Analysis (VCA) technologies; a technology of very high value and an enormous future potential", explains Iñigo.

image

Begiplant detects fire on site.