The state estimation process is fundamental for the real-time operation of Electric Power Systems (EPSs), since the estimated values of the state variables are the basis for the execution of functions related to the real-time monitoring of EPSs. The traditional modeling of the state estimation process in EPSs assumes the hypothesis that the measurements available for estimation have independent random errors. However, considering that a metering system shares the signals from several sensors to compose the measurements to be used in the estimation process and several measurements may share signals from the same sensor, to admit the independence between measurement errors is not a always valid hypothesis. Also, attributing that the measurement errors are not correlated may imply in the omission of the identification of measurements with gross errors (GEs) affecting, cosequently, the accuracy of the state estimation process in EPSs. Few works address the propagation of sensor errors and their influence on the processing of GEs based on the dependence of measurement errors. Therefore, this dissertation approaches the state estimation process in EPSs from the acquisition of signals through the SCADA System, aiming to explore all the equipment present in a metering system. Moreover, is evaluated from the perspective of the accuracy of the estimates obtained and the handling of gross errors. Given the aforementioned, multiple scenarios will be analyzed, taking into account the presence of noise at different stages, the distinct characteristics of the network (symmetric and asymmetric) and load (balanced and unbalanced) of three-phase systems. The tests are conducted using the State Estimator (SE), considering both the traditional single-phase equivalent modeling and the three-phase version. Both approaches employ the Weighted Least Squares (WLS) technique associated with the Largest Normalized Residue Test, methodologies widely employed in practical applications and extensively studied in academic research. Furthermore, with the aim of assessing the contribution of measurement correlation information, the Dependent Weighted Least Squares (DWLS) estimator will be examined. This estimator, denominated in the literature for considering the dependence among measurements, applies the technique of weighted least squares. Based on the results of the case studies, it was observed that traditional single-phase modeling can exhibit shortcomings in various scenarios involving correlated errors. Conversely, both three-phase SE and DWLS SE showcased strong performance, albeit in different contexts. Consequently, there's a need to delve deeper into studies to craft a more resilient tool in the face of adverse error scenarios.