A scalar Kalman filter is applied for a Least-Square estimated value H.sub.s at s. The filter has an input for receiving H.sub.s, a filter equation and an out for the corrected estimated value H.sub.s.sup.k for the k.sup.th variable. The filter equation is H.sub.s.sup.k=K.sub.gainS.sub.n[k] wherein: correction S.sub.n[k]=S+K.sub.n(H.sub.s-S); prediction of the correction S=K.sub.aS.sub.n[k]; Kalman filter gain K.sub.n=P/(1+P); minimum predication MSE P=K.sub.a.sup.2P.sub.n[k]+K.sub.b; minimum MSE P.sub.n[k]=P (1-K.sub.n); and K.sub.a, K.sub.gain and K.sub.b are constants.