Recursive relationships are developed for an optimal #x3b1;#x2014;#x3b2;#x2014;#x3b3; filter having an imperfect predictor. The statistics of the measurement and prediction errors are incorporated in the determination of oprimal#x3b1;, #x3b2; and #x3b3; and parameters. When the prediction uncertainties and the measurement statistics are white Gaussian, the optimal #x3b1;#x2014;#x3b2;#x2014;#x3b3; filter is shown to be equivalent to the Kalman filter.
展开▼