ON SOME ASPECTS OF NON-LINEAR FILTERING

Abstract

iwoortant problems in technology and applied science tn,oly, the estimation of the behaviour of physicSi processes# subject to random disturbances. and messirenant errors. Filter theory is one of the important tools to solve *zIh probleas. Fitter theory has been used to solve problems involved in nivtgstton, guidance, paper industry and an on In this dissertation the theory has beret applied to obtsin# forecast of veather utilising post data. The ct his covered the follovin g points s (1) Discussion of Filter theory and adoptive identification algorithms* (11) brief review of the model presented by Bbri Shares and ri Kahelsnabis. (iii) State space discrete time model, (iv) Identification of the system parameters and prediction rf daily maximum temperature and humidity. The model is tested for the rasa data and results are presented.