Neural networks that could control aircraft engines, improving their efficiency, are to be tested later this year following expressions of interest from propulsion system suppliers.

The neural network in the engine could learn in real time how the engine is performing and make adjustments immediately, according to its developers at Missouri-Rolla University (UMR). The network would learn the best time to inject fuel and calculate and adjust the quantity of fuel being delivered. The artificial intelligence (AI) work at UMR will focus on aircraft engines next year, following further AI testing this November. The building of an internal combustion engine for the testing of the AI gets under way this week.

Dr Jagannathan Sarangapani, associate professor of electrical engineering at UMR, says "neural networks are a maturing technology applicable to a variety of engines".

Source: Flight International