Ground settlement poses a serious threat to railway operation safety

and current settlement measurement methods have problems such as poor robustness

short service life

and susceptibility to interference.To solve this problem

a settlement monitor based on fiberoptic sensing is proposed.Furthermore

a radial basis function(RBF)unscented Kalman filter(UKF)algorithm based on the butterfly northern goshawk optimization(BTNGO)search algorithm is developed for fiber-optic parameter identification and settlement state estimation.Firstly

the fiber-optic sensing measurement values are input into UKF

and the obtained error

gain

and innovation are input into with the RBF network improved by BTNGO for secondary estimation.Verified with actual sampling data

and compared with the extended filtering and baseline models

the estimation error of BTNGO-RBF combined with UKF is reduced to 0.17

and the precision is improved by over 30%

fully meeting practical needs.It also significantly improves the convergence speed.BTNGO search advances convergence of the model by about 10 rounds and reduces training time by about 20%.This method improves the precision and convergence speed of railway settlement monitoring

promoting the development of settlement monitoring towards precision and intelligence.