A Tutorial Introduction to Kalman Filtering卡尔曼滤波介绍.pdfVIP

-- -_____ _ _ _ . A Tutorial Introduction to Kalman Filtering P.J. Hargrave Introduction In what follows we present an elementary introduction to Kalman Filtering starting from the simplest of all estimation problems, namely that of estimating a time independent scalar quantity from a number of noisy measurements. From this we move on to consider the case when the quantity to be estimated is a function of time, and then generalise the results to the estimation of a time dependent vector. Finally we indicatehow the resulting Kalman Filter equations can be applied to an elementary but nevertheless real problem in navigation. Estimatine a Time Invariant Scalar Let x be a scalar quantity that is constant in time. If n measurements of )I are available which are corrupted by noise drawn from uncorrelated zero mean Gaussian distributions of equal variance, the best estimate of x is the mean of all the measurements. Denoting the ith measurement by 91,and the best estimate after n measurements by k,we thus have that i-n ;i,- 1c 9 , . n i-1 If a further measurement, denoted by in+lbecomes available the new best estimate obviously follows as i-n+l %+1 - 1ii+l. n+l i-1 One can express this new estimate in terms of the previous one, to obtain the following update equation n+l This equation can be rearranged to yield We may interpret this equation as follows. If we have an estimate % of a quantity x, and a new measurement of x becomes available,we may use it to update the estimate % to an improved estimate, k+l,by adding to % the difference between the new measurement and the previous estimate scaled b