Experimental Validation of the Algorithm for Processing Diagnostic Parameters of Aviation GTE on the Basis of Multilayer Neural Networks

The paper presents a new approach to diagnosing the technical condition of aircraft gas turbine engines based on multilayer neural networks. The developed algorithm provides comprehensive processing of diagnostic parameters taking into account their mutual influence and temporal dynamics. Experimental studies on real operational data from CFM56 and PS-90A engines showed a significant increase in the accuracy of technical condition classification (95.5%) and a reduction in fault detection time by 69.7% compared to traditional methods. An optimized neural network architecture and a method for forming diagnostic features have been proposed, providing high efficiency in processing heterogeneous data. The results can be used to improve reliability, cost-effectiveness, and safety of aircraft operation.