Modification of an aircraft engine free turbine contour

Abstract

Multimode control of a turbofan is carried out by an engine control unit (ECU). ECU programs provide flight dynamic control under changes in controlled fluctuations within a given range. The effect of uncontrolled disturbances (changes in air parameters exceeding limits, fuel consumption due to leakage in the dispenser, changes in engine parameters, etc.) may disrupt the normal flight conditions of the aircraft. To counter this, it is proposed to correct ECU programs by adding adaptive neural control in all contours. Replacement of the standard free turbine speed control contour by an adaptive neural controller (ANC), which includes an adaptive fuzzifier and defuzzification unit, was shown as an example. A choice of the fuzzifier term set in relation to the properties of a dispenser which excludes a statistic error in the free turbine speed control contour was shown. A neuron with sequential learning provides adaptation of the fuzzifier. An algorithm for correction of activated grades of membership of the fuzzifier at a particular time changes the numerical value of synapses per iteration. Given waveforms confirm the possibility of using the adaptive neural control for stabilizing the free turbine fan speed. However, stabile maintenance of the free turbine fan speed causes self-oscillations due to imbalance of generated and consumed power of the aircraft engine. This disadvantage can be eliminated with the aid of a built-in rotation torque sensor in a free turbine, which can be used to determine the power consumption of the aircraft engine. Introduction of power feedback in the energy load and generation contours eliminates self-oscillations. The simultaneous use of adaptive neural control and power feedback makes this paper topical.