New methods and models for some data processing systems


Based on artificial neural networks, cellular automaton, deformable templates methods, new mathematical methods and algorithms for fast and reliable experimental data processing have been proposed [25]. The developed codes allow to solve such problems as event identification, track recognition , noise removal. The created programs have been included in the software for data analysis of the experiments DISTO, FOBOS [26], SINDRUM [27], EXCHARM [28], STAR, CERES/NA-45, ATLAS. So, programs implementing the robust algorithm for the fast circle fit after extensive tests on real data of the CERES experiment were included in the basic software of this experiment [29]. A model of second-level trigger, to be applied for track recognition, in searching for secondary vertex, and for identifying the secondary particles is realized now on the basis of 4-RISC processors, which are used in the spectrometer DISTO for data acquisition and on-line analysis [30]. New programs for track recognition were developed. The developed software is successfully used for JINR experiments at U-70 (Protvino, Russia) and LHC (Geneva, Switzerland) in real time monitoring, experimental data processing and simulation of detector performance.