Investigation of suddenly expanded flows at subsonic Mach numbers using an artificial neural networks approach.

The purpose of this study is to explore two concepts: first, the use of artificial neural networks (ANN) to forecast the base pressure (β) and wall pressure (ω) originating from a suddenly expanded flow field at subsonic Mach numbers. Second, the implementation of Garson approach to determine the critical operating parameters affecting the suddenly expanded subsonic flow process in the subsonic range. In a MATLAB environment, a network model was constructed based on a multilayer perceptron with an input, hidden, and output layer. The network input parameters were the Mach number (M), nozzle pressure ratio (η), area ratio (α), length to diameter ratio (γ), micro jet control (ϵ), and duct location to length ratio (δ). The network output included two variables; base pressure (β) and wall pressure (ω). The ANN was trained and tested using the experimental data. The experimental results found that micro-jet controls were successful in increasing the base pressure for low Mach numbers and high nozzle pressure ratios. It was also found that the wall pressure was same for with and without micro jet control. The ANN predicted values agreed well with the experimental values, with average relative errors of less than 5.02% for base pressure and 6.71% for wall pressure. Additionally, with a relative significance of 32% and 43%, the nozzle pressure ratio and duct location to length ratio had the highest influence on the base pressure and wall pressure, respectively. The results demonstrate that the ANN model is capable of accurately predicting the pressure results, enabling theoretical foundation for research into pressure distribution in aerodynamic systems.