Modelling and state estimation of power plant steam turbines

The development of a robust state estimator for ill-conditioned systems based upon the Levenberg-Marquardt algorithm in connection with Givens orthogonal row processing technique is described. The estimator works in two stages: the square-root-free Givens transformation is applied to the measurement Jacobian matrix to reduce it to the product of a unit upper triangular matrix and a diagonal matrix, then the square-root-free Givens transformation is again used to transform a diagonal matrix whose diagonal element is equal to the Levenberg-Marquardt parameter. The estimator is applied to a 250 MW, double-flow, single reheat steam turbine which has seven feedwater heaters. Tests results show that the estimator performs well under a wide range of measurement errors and measurement redundancy ratios. A sample of estimated results for rated loading condition are presented