Estudo dos pesos na estimação de estado de sistemas elétricos de potência usando modelagem de programação não linear no AMPL

Abstract

State estimation is the computational method used in control centers to minimize errors in the measurement system. Generally, the readings of five quantities are received in the database: active power injection, reactive power injection, active power flow, reactive power flow and voltage. In this work, the problem of state estimation of electric power systems is represented through a mathematical programming approach instead of Newton's interactive method. In this approach, a non­linear mathematical model based on the classical weighted least squares method is used to solve the state estimation problem. The weight values are classically given in the literature as constant values depending on the type of measurement and the standard deviation of the measurements. The objective of this work is to improve the state estimation result by proposing new weights. For this, twelve tests with varying weights are performed for each of the five measurement types in the IEEE14 bus test­system. This test aims to analyze the sensitivity and influence of the weights of each type of measure on the estimation results. The results of these tests were used to guide the attempt to conjecture new weights. The estimation results, when containing only gaussian errors in the database, with new weights, achieved a performance forty times better for the stress modulus and five times better for the stress angles, when compared with the estimation using classical weights. The state estimation with the new weights proposed by this work made the state estimator immune to gross and gaussian errors for the voltage measurements of the IEEE 14bus test­system. The non­linear programming model is implemented in AMPL that uses the Knitro solver, both in student version. To validate the results of the proposed methodologies, the electrical system operating points are compared with the results obtained using the PowerWorld simulation package, as well as with the results obtained when inserting measurements without errors into the database. The results allowed us to conclude that the brute force methodology used in this work to conjecture new weights can be successfully applied to estimate the state of electrical systems more efficiently.