I-algorithm yokuhlanganisa izinzwa

Isihlungi se-Kalman

Isihlungi se-Kalman sijwayelekile.

Umongo we-algorithm uwukusetha isethi yezinto "zenkolelo" yenzwa ngayinye. Isikhathi ngasinye, idatha yenzwa kusukela esikhathini sokugcina izosetshenziselwa izibalo ukuthuthukisa ukuqagela (ukuzengeza ngokwakho), futhi ikhwalithi yenzwa nayo izokwahlulelwa. Uma kuqhathaniswa phakathi kwevelu ebikezelwe kanye nenani elilinganiselwe lenzwa, inani elihle kakhulu lizolinganiswa futhi likhishwe.

Lokhu kusho ukuthi uma inzwa ihlale ikunikeza inani elihle nelingaguquguquki futhi iqala ukukutshela okuthile okungalindelekile, ileveli yokwethembeka kwenzwa izokwehla ngama-millisecond ambalwa ize iqale ukwenza umqondo futhi.

Lokhu kungcono kunokulinganisa okulula noma ukuvota, ngoba isihlungi se-Kalman singabhekana nesimo lapho izinzwa eziningi zingasebenzi okwesikhashana. Uma nje umuntu engakwazi ukugcina isizathu esihle, angenza irobhothi ngesikhathi esimnyama.

Isihlungi se-Kalman siwukusetshenziswa kwemiqondo evamile kakhulu ye-Markov chain nokucabanga kwe-Bayesian, okuwuhlelo lwezibalo oluthuthukisa ngokuphindaphindiwe ukuqagela kwabo ngokusebenzisa ubufakazi. Lawa mathuluzi angamathuluzi asetshenziselwa ukusiza isayensi ngokwayo ukuhlola imibono (ephinde ibe yisisekelo salokho esikubiza ngokuthi "ukubaluleka kwezibalo").

Ngakho-ke, kungashiwo ngokusankondlo ukuthi ezinye izinhlelo zokuhlanganisa izinzwa ziveza ingqikithi yesayensi ngesivinini sezikhathi eziyi-1000 ngomzuzwana.

Izihlungi ze-Kalman zisetshenziswe eziteshini ze-orbital zamasathelayithi asemkhathini amashumi eminyaka. Ngoba izilawuli ezincane zesimanje zingasebenzisa i-algorithm ngesikhathi sangempela, ziya ngokuya ziduma kumarobhothi.