Methods for injection-error analysis and their comparison

Statistical techniques for the analysis of missile injection errors are studied in detail. The commonly used direct and adjoint methods are reviewed and extended. It is shown that the determination of the covariance matrix is equivalent to the determination of two transformation matrices for both methods. In general, the adjoint method is more efficient. But for a special case the direct method could be preferable depending upon the relative dimension of the system state and the error source. Two examples are given to verify the results. The techniques presented can be applied equally well to a wide variety of control system problems.