T. A. Kuznetsova - Ph.D. (Eng.), Associate Professor, Head of Distance Education Technologies Centre, Perm National Research Polytechnic University. Е-mail: tatianaakuznetsova@gmail.com
E. A. Gubarev - Student, Perm National Research Polytechnic University. E-mail: eugenegubarev@gmail.com
Yu. V. Likhacheva - Student, Perm National Research Polytechnic University. E-mail: likhachevajul@rambler.ru

Aircraft engine as a control object is a complex dynamic system which parameters are characterized by random distribution over a wide range, caused by the internal and external interferences. Engine reliability is greatly determined by the quality of engine-s automatic control system (ACS GTE). For optimum realization of the automatic flight control functions the main objective is the obtaining a reliable real-time data shown engine-s current characteristics. The solution of this problem requires the identification methods - usage. Taking into account that the ACS operates under interferences in the model channel (due to inaccuracy of the model) and in the measuring channel (due to sensors - error), the ensuring of a high identification accuracy of engine-s parameters including current modeling data and airborne measurements is very important task. This work considers the usage of optimal Kalman filtering algorithms for signals at the input and output of the built-in multi-mode linearized mathematical model of the aircraft engine, based on its dynamic and throttle characteristics. Input filtering was performed in the fuel dosing channel. Multi-dimensional Kalman filter is connected to the output of the built-in mathematical dynamic model of GTE. The test of feasibility of Slutsky-s conditions and Pearson-s hypotheses of normality of real noise experimental characteristics distribution showed the possibility of Kalman filters - using for this class of processes. Optimal filtering task is the noise effects elimination in the channel model, and in the measurement channel at the current time. Filter adaptation algorithms are based on determining the optimal Kalman coefficient by solving the task of minimizing the expectation of the square of the error of identified parameter within the error and the optimal estimate in the previous moment. The task the optimal Kalman filtering algorithms - design has been solved with the usage of a model experiment in an environment Simulink. Results of Kalman filtering of the input signal of the built-in ACS model of GTE in fuel supply circuit for applying the control signal, which permits movement of the metering needle throughout the operating range, showed that the average relative error is 1.68%. Relative error of output signals of the built-in ACS model filtered by means of multi-dimensional Kalman filter for four parameters - high and low pressure turbines - speed, pressure, temperature does not exceed 0.5%, which corresponds to the specified technical requirements of identification algorithms - accuracy. Analyze of the results suggests that the use of one-dimensional and multi-dimensional matrix Kalman filter can achieve higher levels of accuracy than similar models developed without the usage of optimal filtering algorithms in solving the problems of parameter identification. The developed optimal filtering algorithms operate in both static and dynamic modes of the aircraft engine under the action of the "hard" external and internal interferences in a wide range of the engine-s operating.