Knowledge-Based Aerodynamic Estimation of Airships

Abstract

High fidelity flight modeling and simulation generally involves development of mathematical models based on aerodynamics and flight characteristics derived from experimental or numerical data. The data is generally recorded in look-up tables and called in during simulation. This results in very high computational cost (time & hardware requirements). An alternative, discussed in this paper, is to use a computational and knowledge-based paradigm, called neural networks. The network is presented with the experimental data and learns the relationships between forces and moments in six degrees of freedom. This modeling strategy has important implications for modeling the behavior of novel and complex flying configurations, such as airships that are considered in this paper. The pipeline includes, digitization of wind tunnel data, compatibility of digitized data with neural network (feedforward) followed by development of six degree of freedom aerodynamic model. The preliminary results of using neural networks to model aerodynamic forces & moments look promising. 

Aerodynamic characteristic analysis of blast wave lateral sweeping aircraft based on CFD transient numerical simulation method

“…For example, length, position or angle were considered as variables, and numerical research methods were employed to assess the precision and feasibility of aerodynamic characteristics [6]. The relationship between forces and moments in the six degrees of freedom of the experimental data was learned by Khurram using a neural network, leading to the development of a sophisticated six-degree-of-freedom aerodynamic model [7]. Based on Archimedes' principle, the Bernoulli equation, and the continuity equation, J.A.…”