Znak Politechniki Warszawskiej

SILO - artificial intelligence in the power sector

SILO has been implemented in 46 boilers to date.

Scientists of the Faculty of Power and Aeronautical Engineering, Warsaw University of Technology, have been working on applications of artificial intelligence in the power sector for 20 years. The product of their efforts, the SILO system, is in use both in Poland and abroad. As the only solution from Poland, the system was exhibited in the “Energy Best Practices Area” pavilion at EXPO 2017 in Kazakhstan.

“Viewed from the theory of control and regulation, power boiler combustion as a whole is a difficult process because it is highly non-linear, with many interacting variables,” explains Prof. dr hab. inż. Konrad Świrski [D.Sc. Eng]. “To align all classical regulation systems to make a boiler work very well under any load or on coal of any grade or in any configuration of the working devices is a great challenge indeed.”

From the perspective of a power station, two issues are of key importance in the process: achieving top boiler efficiency and leaving as little carbon footprint as possible. Solutions applied 20 years ago failed to fully live up to those standards. So there was plenty of room for new ideas.

Curing the “disease”

WUT scientists could see the shortcomings of the classical regulation. They could also see a growing interest in artificial intelligence. They decided it could be successfully applied in power boiler control.

Initially, back in the mid-1990s, they used artificial neural networks, or structures simulating the functions of the nervous system, in particular the brain. Nowadays, they use artificial immune systems. “We compare the object concerned to an organism that we want to keep in top condition,” says Prof. Świrski. “Any disturbances are like pathogens that an organism should respond to with lymphocytes. This is how it works in SILO. If anything changes, be it the power level or coal quality, our system responds with its ‘lymphocyte’, which is optimized control.”

Like with a human organism, artificial immune systems are able to learn new “diseases” and create new sets of “lymphocytes” against them. “This means that if a new process occurs in a boiler, leading to regulation issues with the device, the system starts to learn and generates a new solution,” says Prof. Świrski. “It will prevent future occurrences. It’s a great advantage.”

The scientist stresses that boiler control is only a part of a very complex and sophisticated process of overall power unit control. “We adjust our solution to fit that process,” says Prof. Świrski. “Certainly, we can adapt it to accommodate individual boilers or a particular power station too.”

The system was exhibited in the “Energy Best Practices Area” pavilion at EXPO 2017 in Kazakhstan.

Where science meets business

The most advanced version of the SILO system has been implemented in Rybnik. “We have used dedicated equipment to track not only hard physical parameters, such as temperature, but also more elusive factors, such as the flame shape, as we know it also affects the combustion efficiency,” explains Prof. Świrski.

SILO has been implemented in 46 boilers to date. It is used in Poland and beyond, in the US, South Korea, China and Taiwan. This is a model example of how science meets the needs of business. The system offers a solution to the two major issues: it improves boiler efficiency and helps curb the carbon footprint, which in turn allows to save time and money.

Agnieszka Kapela

Promotion and Information Office