V.N. Zhurakovsky – Ph.D.(Eng.), Associate Professor,
Bauman Moscow State Technical University
One of the main tasks of radar is to assess the position of the direction finding object with the least error. This problem cannot always be solved at the stage of primary signal processing, as a result of which the parameter estimates have significant errors. The goal is to consider a machine learning method based on the method of k-nearest neighbors and helping to reduce the resulting error in estimating the angle by dropping out all the «unreliable» estimates.
The question of reducing the errors in estimating the coordinates of radar objects due to machine learning for filtering highly noisy estimates by the example of estimating the direction angles to a location object is studied.
As a result of applying the method, it was possible to reduce the standard deviation of the error in estimating the angle by more than four times.
- Zagoruyko N.G. Prikladnye metody analiza dannykh i znaniy. Novosibirsk: IM SO RAN. 1999. 266 s. (In Russian).
- Zhuravlev Yu.I., Ryazanov V.V., Sen’ko O.V. «Raspoznavanie». Matematicheskie metody. Programmnaya sistema. Prakticheskie primeneniya. M.: Fazis. 2006. 159 s. (In Russian).
- Zhurakovskiy V.N., Matyunin D.V., Silin S.I. Opredelenie napravleniy na otrazhayushchie ob’‘ekty metodom maksimal’nogo pravdopodobiya pri ogranichennom chisle luchey, formiruemykh antennoy reshetkoy. Antenny. 2016. № 7. S. 35−39. (In Russian).
- Ermolaev V.T., Flaksman A.G. Metody otsenivaniya parametrov istochnikov signalov i pomekh, prinimaemykh antennoy reshetkoy. 2007. 100 s. (In Russian).