Improving the Accuracy of Aeroengine State Identification Using Artificial Intelligence Technologies

The study is devoted to the development of an adaptive on-board model of a gas turbine aeroengine, built into its automatic control system. The adaptability of the aeroengine mathematical model to a possible change in its state is achieved based on the diagnostic matrix method, allowing defining immeasurable engine parameters in terms of the physically measurable parameters. Immeasurable parameters refer to changes in the efficiency of the main engine components, air bleeds and leakage. The identification process is based on solving a diagnostic system of linear algebraic equations, which the main determinant is close to zero. The accuracy and speed of the on-board built-in model are determined by choice of mathematical method for solving obtained diagnostic system. The problem of the linear algebraic equations system uncertainty is solved based on the exact (Cramer, Gaussian), iteration (fixed-point, Jacobi, Seidel) and optimization (numerical Monte Carlo and a genetic algorithm) methods. The effectiveness of the methods was assessed based on the average relative error. The bench testing of developed algorithms on the simulator of a real industrial engine in the loop of an automatic control system was performed. The promise of using the genetic algorithms to solve the considered problem, taking into account all the limitations, has been proven.