A.G. Dmitriev – Ph.D. (Eng.), Associate Professor, Cherepovets Higher Military Engineering School of Radio Electronics  E-mail: dag334a@fxmail.ru

The article is about an algorithm for the piecewise – continuous approximation of the experimental signal with unknown the initially number of intervals of partitioning of signals by «one type» fragments based on the method of dynamic programming. A criterion of quality of approximation is used. It takes into account the number of data at intervals of partitioning and «complexity» of local  models of signals used.

1. Mottl' V.V., Muchnik I.B. Skrytye markovskie modeli v strukturnom analize signalov. M.: Fizmatlit. 1999.
2. Dmitriev A.G. Algoritm optimal'noj strukturnoj approksimacii jeksperimental'nyh mnogomernyh signalov // Naukoemkie tehnologii. 2010. № 9. S. 31–35.
3. Kostin A. A., Krasotkina O. V., M. V. Markov M. V., Mottl' V.V., Muchnik I.B. Algoritmy dinamicheskogo programmirovanija dlja analiza nestacionarnyh  signalov // Zhurnal vychislitel'noj matematiki i matematicheskoj fiziki. 2004. T. 44. № 1. S. 70–86.
4. Dmitriev A.G., Dorofejuk A.A. Metody kusochnoj approksimacii mnogomernyh krivyh // Avtomatika i telemehanika. 1984. № 12. S. 101–108.
5. Bellman R., Drejfus S. Prikladnye zadachi dinamicheskogo programmirovanija. M.: Nauka. 1969.