Applications of artificial neuron networks for solution of an inverse problem of gas dynamics of compressor cas-cades

Abstract

This paper presents a solution of an inverse problem of the gas dynamics of compressor cascades using arti-ficial neural networks (ANN) by summarizing the experimental data resulting from plane cascades purging. The technique for determination of geometrical parameters of the cascade of profiles for given parameters of the flow at infinity in front and behind the cascade is developed using the solution proposed. The technique uses ANN, the architecture of which is a multilayer perceptron, for calculations of aerodynamic characteristics of the cascade of profiles. The modified model of a classical genetic algorithm is used for ANN designing. Network training is performed using the error backpropagation method. The effectiveness of the developed technique was evaluated by solving an inverse problem of the gas dynamics for given flow parameters and a subsequent determination of aerodynamic characteristics of the cascade, based on a numerical simulation of the turbulent gas flow. The results obtained confirm the robustness of the technique reported for solving inverse problems of the gas dynamics of compressor cascades using ANN for generalization of the experimental data. The results of this work can be used at the stage of a conceptual design of compressor wheels for aircraft gas turbine engines and various power plants. The application of these results will automate and improve the design process, energy characteristics of the produced prototypes

Authors and Affiliations

S. Melashich

Keywords

Related Articles

Mathematical modelling 3D oscillation of liquid launch vehicle upper stage with cruise gimbal-mounted engine

Determinations of natural oscillation parameters of the launch vehicle upper stage is of important for the theoretical stability analysis of the launch vehicle upper stage relative to its elastic longitudinal and lateral...

Chemical and explosion- fire safety of spillage of propellant components at launching rocket com-plees

It is shown that the improvement of chemical and explosion-fire safety of spillage of propellant components due to accidents at the launching complexes during the launch preparation is provided by dispersed flows of th...

Selection of design parameters and control programs at initial stage of designing launch vehicles

The problem of the optimization of the basic design parameters and programs of the motion control of liq-uid launch vehicles is formulated. Programs of the motion control of space launch vehicles are determined in the cl...

Determination of parameters of cavitation self-oscillation in hydraulic system under simultaneous longitudinal oscillation of pipe structure and fluid

The paper deals simultaneous longitudinal oscillation of feed pipes of liquid rocket engines (LRE) and flu-ids. Cavities, normally occurring in operations of LRE pumps, can result in self-excited low-frequency of the flu...

Effects of nonlinear characteristics of bellows on parameters of cavitation self-oscillation under combined longitudinal oscillation of line structure and fluid

The paper examines the combined longitudinal oscillation of the feed line structure of liquid rocket engines and the fluid considering nonlinearities of an elastic characteristic of the bellows. The limitations of the c...

Download PDF file
  • EP ID EP356657
  • DOI -
  • Views 65
  • Downloads 0

How To Cite

S. Melashich (2014). Applications of artificial neuron networks for solution of an inverse problem of gas dynamics of compressor cas-cades. Технічна механіка, Техническая механика, Technical Mechanics, 1(1), 46-51. https://europub.co.uk/articles/-A-356657