Scientists from Perm National Research Polytechnic University (PNRPU), together with specialists from UEC-Aviadvigatel, have developed a software adaptive observer capable of compensating for aircraft engine sensor failures in real time. The new technology increases the reliability of control systems without increasing the weight, energy consumption, and cost of the engine.
Traditional methods of ensuring reliability—duplicating measurement channels—do not exclude the risk of simultaneous failure of redundant sensors and are associated with an increase in structural complexity. In contrast, the proposed solution is based on an adaptive mathematical model of the engine, which is continuously adjusted taking into account current operating conditions, including temperature, modes, and component wear.
The algorithm operates in two stages: first, an engine model is formed and refined, as close as possible to the physical object, then the system suppresses measurement noise and interference. If one or more sensors fail, their readings are replaced with predicted values.
Tests were carried out on an industrial bench complex in observation mode—without interfering with engine control. Testing covered all major operating modes, from minimum to take-off. According to the results, the discrepancy between real measurements and algorithm calculations did not exceed 0.008%, which is ten times better than the requirements of current aviation standards. The system also demonstrated resistance to external disturbances.
Head of the Department of Applied Mathematics of PNRPU, Vladimir Pervadchuk, emphasized that the technology is especially relevant for promising engines, where the requirements for the reliability of control systems are growing.