METHOD FOR DIAGNOSING A COMPLEX OF ON-BOARD EQUIPMENT OF AIRCRAFT BASED ON UNSUPERVISED MACHINE LEARNING WITH AUTOMATIC DETERMINATION OF MODEL TRAINING PARAMETERS Russian patent published in 2024 - IPC G06F15/00 G06F11/00 G06N20/00 

FIELD: diagnostic systems.

SUBSTANCE: method for diagnosing a complex of on-board equipment of aircraft based on unsupervised machine learning with automatic determination of training parameters of models, comprising steps of: collecting and preparing source data, as well as importing the required libraries of the programming language used in the algorithm, determining transmitted information data, performing automatic determination of optimal parameters of a one-dimensional Kalman filter, applying a modified Kalman filter for the collected data, automatic creation of K-means, SOM and DBSCAN machine learning models, automatic determination of models training parameters, models are trained on the data according to the criteria for obtaining a cluster describing the operable state of each information-converting element, modified Kalman filter for real-time data is applied.

EFFECT: increasing the accuracy of detecting faults in aircraft on-board equipment.

1 cl, 4 dwg