Abstract
A real-time state estimator has been developed for multivariable continuous systems. It is based on least-squares theory, therefore no a priori noise statistics are needed and the stability of the algorithm is guaranteed even for unstable systems. Successive integrations are used to filter the noise and to accumulate the necessary data in a parallel fashion with much flexibility in combining these data. This estimator requires a square matrix to be inverted in an off-line fashion; hence the real-time computations are reduced to only matrix-vector multiplications. For noise free systems the developed estimator can be used as an observer.