Developing quick aerodynamic design method of small unmanned aerial vehicle based on artificial neural webs technology requires a rich database used for creating, learning and testing algorythms. This database must be significantly larger than the number of existing vehicles, that's why creating a layout generator is important. On the first stage the database was increased by varying the parameters of the layout mathematical model in the accepted range. The layout generator is an artificial neural web trained on the rich database of unmanned aerial vehicles layouts in the frames of the simplified mathematical model. The important element in creating a new layout is the selection algorithm applied to the neural network output.The initial number of layouts was equal to 25. The simplified mathematical model describes the unmanned aerial vehicle layouts with 50 parameters. The layout generator forms an input file for the CFD code, which dimension is of the order of several thousands. The CFD codes BLWF and VISTRAN were chosen for this task. They add to each layout theinformation about its aerodynamic characteristics. Selection of the database was done into two levels. At first it was im-plemented on the layout generator output. Then on output vector after the CFD calculation. The selected databaser was used for creating an artificial neural networks number of which were equal to the number of aerodynamic coefficients. The algorithm was built in MATLAB and has convenient interface, which can be used for design process.
LAYOUT GENERATOR OF SMALL SIZED UNMANNED AERIAL VEHICLE
2017
Article (Journal)
Electronic Resource
Unknown
Metadata by DOAJ is licensed under CC BY-SA 1.0
SIMPLIFIED MATHEMATICAL MODEL OF SMALL SIZED UNMANNED AIRCRAFT VEHICLE LAYOUT
DOAJ | 2017
|