Multiagent Control of an Air-Jet Engine

Abstract

When an air-jet (turbojet) engine is operated, continuous parameter monitoring is required—specifically, of the gas temperature behind the combustion chamber (gas chamber, rotational frequencies of the free pressure turbine, rotational frequencies of a turbocharger rotor, etc.). The operation of the circuits is regulated by a selective controller, which asynchronously logically continuously controls the flow of fuel (kerosene, gas) through the metering device and analyzes the deviations of parameters of a turbojet engine. All circuits have software control and implement the principle of regulation by deviation. The software implementation of the control circuit operation of the turbojet engine is performed by the engine control unit (ECU). The calculations of optimal settings of regulators in the circuits are performed according to the technical optimum, provided that the static characteristics are constant. However, the actual static characteristics of a turbojet engine change during operation and the previously calculated settings of regulators in the circuits may not be optimal. Software control of the system enables the appropriate selection of coefficients from the ECU memory without taking into account the changes in the static characteristics of the turbojet engine at the current time. This method of calculating control system parameters is not effective, since it requires an exact mathematical description of the object and its changes at a given instant in time. Other uncertainties involve the conditions of the dispenser elements, fuel quality, etc. Currently, the aforementioned parameters are all regulated separately, which is a disadvantage. It has been proposed to replace regulators of the turbojet engine parameters with one regulator of state that would control the dispenser in order to eliminate the selector from turbojet engine control. A fuzzy state controller made on the base of adaptive fuzzifier with a fuzzy implication is considered. To adapt the fuzzifier, the objective function is used, with the help of which the coefficients of first-order polynomials are calculated.