KALSY FILTER DESIGN USING THE LEVIKSOK ALGORITHX A?jD OLTPLT STATISTICS
Eglin Afb, Thomas E. Bullock · 1984
In this *per we present a new method to identifjthe rrinimn parameter autoregressive mving-average (ARK&) meel of a systen. The mdel identified, when used as a one-step ahead predictor, produces a minimuc; error variance estimate. The parameters are found from outp:ut statistics by solving a set of linear equations. The ANNA. model found is equivalent to the Kalman filter inr.ovati0r.s model but we avoid solving a Riccati-type equation. The equivalence is demonstrated through a nunerical example.