Research On Robust State Estimation And Parameter Estimation Methods Of Electric Power Systems

Power system state estimation is a very important part of energy managementsystem. Its result directly influences the intelligent analysis and decision-making ofpower dispatching. How to improve the precision of state estimation are the keyelements of state estimation study. In practical operation power system, measuring baddata and parameters mistakes, lead to state estimation results is not accurate. Existingstate estimation methods generally assume that the measurement error is subject tosome expected distribution when the actual measurements do not obey or do not strictlyobey the expected distribution; the existing methods are difficult to exclude the impactof bad measurement data, particularly difficult to remove bad leverage measurement orthe measurement which is strongly related to bad data. Parameter estimation is animportant method to eliminate the effects of parameter errors; existing parameterestimation methods based on the entire network of the single-time measurement data.As more equipment and measurement information involved, parameters and betweenparameters and measurement error have mutual influence. It is difficult to ensure theaccuracy and numerical stability of parameter estimates.To solve the above problem, this paper presents the robust state estimation andparameter estimation method:Firstly, based on phasor measurement unit (PMU) or supervisory control and dataacquisition (SCADA) multi-period measurement information of transmission line andtransformer at both ends of the element, a robust estimation method of branchparameters is proposed. Considering fully the Normal Distribution characteristic ofrandom measurement error, andbased on analytical expression of the branch parametersand both ends measurements, this method use multi-period measurement informationcalculated value of the variance coefficient as the convergence criterion. This methodeschews the numerical stability in traditional methods, so it is simple and more accurate.Secondly, based on exponential weight function, the method of robust least squaresstate estimation is put forward. This method takes exponential function of thenormalized residuals as weight function. Its iterative calculation process is similar toweighted least squares estimate, but each iteration need to update the weight based onexponential weight function. While retaining the traditional least squares efficiencybased the exponential weight function, the method can automatically exclude the impact of bad data, and improve the state estimation method robust.Thirdly, this paper based on the robust least squares state estimate method, regardsthe suspected parameter as an additional state variable. The method of parameterestimation based on state vector augmentation is formulated.Finally, simulation results on several IEEE standard examples show that theproposed methods are corrective.