CASPIAN JOURNAL

MANAGEMENT AND HIGH TECHNOLOGIES

Neural network approach to determination of parameters of destruction of technically difficult objects

Read Savochkin Alexandr E. Neural network approach to determination of parameters of destruction of technically difficult objects  // Caspian journal : management and high technologies. — 2013. — №2. — pp. 151-160.

Savochkin Alexandr E. - post-graduate student, Penza State Technological Academy, 1a/11 Baydukov proezd/Gagarin St., Penza, 440039, Russian Federation, aebrat@mail.ru

By use of neural network programming’s technologies of the analytical Deductor platform and Matlab Simulink on a material of base of vibrosignals were approved in this article. Possibility’s research the device’s use of artificial neural networks for the solution identification’s problems of technically difficult objects (TDO) condition was the work’s purpose. For the decision of this task the KDD technology – Knowledge Discovery in Database – extraction of knowledge from databases was used. Within the conducted researches architecture of multilayered neural networks were tested, testing and an assessment of results’ accuracy of identification was held. After that the optimum architecture of the artificial neural network (ANN) was chose. By means of modeling of damage degree TDO (S) damage rate for several test signals was revealed and tests for accuracy and the adequacy testifying to high identification abilities of used ANN are carried out.

Key words: vibration signal,identification,technically difficult object,data mining,neural network,neuro computing