Simplified Modeling of a Flapper-Nozzle Servo Valve for Electro-Hydraulic Actuators: Genetic Algorithms and Neural Networks

Despite the onset of new subsystem design philosophies, such as the More Electric Aircraft (MEA) one, which propose a shift towards aircraft electrification, hydraulically powered actuators still make up the backbone of flight control actuators in modern days airliners. Electro-Hydraulic Actuators (EHA), even though heavier and less flexible than more advanced actuators, represent a time-tested, reliable and mature technology. This paper proposes three methodologies to model an Electro-Hydraulic Actuator flapper-nozzle servo valve. While taking a computationally intensive physical-based model for reference, the authors developed three independent simpler models leveraging a look-up table, a genetic algorithm-based approach and an artificial neural network concept. The models take as input the same four main servo valve parameters: spool position, flow rate, covering and radial clearance. The neural network based model has been used for diagnostic purposes too, representing a first step towards the future development of more integrated prognostic frameworks, enhancing the component safety and reliability. The final outcomes show general positive results with errors of 5-10, 10-15 and 1-5 percent for the look-up table, genetic algorithm and neural network methods respectively.