Diagnostics of an Aeroengine Automatic Control System under Conditions of Insufficient Apriorial Information

The optimality of an aero engine automatic control system (ACS) depends on the reliability of information about the current characteristics of the control object. It is proposed to increase the ACS robustness by creating an algorithmic information redundancy based on a built-in onboard mathematical model of the engine with the parametric gas- path diagnostics function. The method of the general matrix of influence coefficients (diagnostic matrix) is used. The study deals with the problem of incorrect conditionality of diagnostic matrices. The uncertainty of diagnostic systems is caused by the "noise" of the model. Two methods of reducing the model "stochastic noise" are considered. The first method is based on reducing undefined and ill-defined systems of equations to certain ones by selecting optimal sets of non-measurable parameters. The second approach is based on expanding the engine state space using numerical Monte Carlo methods. Based on the diagnostic results, the model is corrected in real time. "Random noise" in measured channels is compensated based on the Kalman filter. A semi-natural experiment on an industrial regulator with various engine power settings gave unsatisfactory results using the first method. The best accuracy is achieved when evaluating the high-pressure rotor speed, the worst-when evaluating the pressure behind the compressor. The second approach, based on statistical modeling, increases the identification accuracy by 1.5-4.7 times. Kalman filtering of input signals reduces the standard deviation of the obtained unmeasured parameters by 1.3-2.5 times.