Orthogonal Transformation And Trust Region Based Power Systems Variance Estimation Algorithm

State estimation plays a very important role in the power grid monitoring. It’s basic algorithm is weighed least squares method. In practical state estimation applications, measurement error variances insufficient and weights wrongly configured bring difficulties to the application of state estimation. The absence of priority information of the measurement variances make calculating measurement error variance become particularly important. But the existing measurement variance estimation algorithm has the shortages of numerical instability and calculation difficult.In order to improve the stability of state estimation, and then establish a stable and reliable power systems measurement error variances estimation and weights configuration algorithm.Specific studies are as follows.Firstly, based on the weighted least squares method, the reasons of power system state estimation numerical problems are analyzed. And the orthogonal transformation with numerical stability is introduced. Householder and Givens methods are researched. Then, according to the characteristics of Givens method, the efficiency strategies are studied. The method based on orthogonal transformation is applied to deal with bad data.Secondly, the mathematical model of the iterative re-weighted measurement error variances estimation algorithm is established. The algorithm, which uses the history measurement data samples, repeatedly estimates the weights to achieve gradual convergence. As results, the measurement error variances and weights are obtained.On the basis of parameters combination method, the two-step estimation method, the posteriori estimate of weight functions, and trust regions method are researched. The properties of the residual sensitivity matrix are analyzed. Two models of calculating error variances are given.Finally, an algorithm based on orthogonal transformation and trust regions to estimate measurement variances and weights is proposed. Through the state trustregions as constraints, a global convergence model of state estimation is established to improve the accuracy of each iteration. The intermediate results of orthogonal transforming is used to simplify the original residual sensitivity matrix. As results, a new measurement variances calculation formula is deduced. With the enhancement of numerical stability and the improvement of computational efficiency, a new algorithm to initialize and update measurement error variances is established.