Title :
New insights on SINTRACK, a real-time algorithm for aircraft structural-modes identification.
Author :
Jeanneau, Matthieu ; Bastide, Emilie ; Raimbault, Nathalie
Author_Institution :
Res., Methods & Tools Team, Airbus, Toulouse, France
Abstract :
SINTRACK is a real-time signal-processing algorithm providing estimates of low damped sinusoids. Its good balance between accuracy and low CPU cost makes it an attractive tool for on-board applications. Some first theoretical analysis on its performances were performed and presented in EUSIPCO 98, along with heuristics for an optimal tuning of some algorithm parameters. Further studies were lately performed in order to fulfil industrial requirements. Domain of use with regard to signal-to-noise ratio and number of modes in the signal was evaluated. Normalization of the dependencies to the last two tuning-parameters was performed. Then optimisation of their tuning using neural networks was conducted. The result consists in an optimal universal solution valid for any single-mode signal. Many aeronautical applications requiring real-time estimates deal with such signals. Coding of the algorithm was performed in a very basic aeronautical language, used for embedded code production, authorizing only elementary functions.
Keywords :
aircraft communication; encoding; neural nets; on-board communications; real-time systems; signal processing; EUSIPCO 98; SINTRACK; aeronautical applications; aeronautical language; aircraft structural-modes identification; dependencies normalization; embedded code production; industrial requirements; low damped sinusoids; neural networks; optimal universal solution; real-time estimates; real-time signal-processing algorithm; signal-to-noise ratio; single-mode signal; tuning-parameters; Abstracts; Algorithm design and analysis; Encoding; Equations; Mathematical model; Robustness; Time-frequency analysis; Real-time estimation; neural network; structural modes identification;
Conference_Titel :
Signal Processing Conference, 2004 12th European
Conference_Location :
Vienna
Print_ISBN :
978-320-0001-65-7