Algoritmo para Estimação de Estado pelo Critério de Máxima Correntropia com Novo Ajuste das Janelas de Parzen

Abstract

State estimation in power systems is a crucial tool in the context of real-time operation, responsible for providing estimates regarding the current operational state of the system. In the search for robust and accurate estimators in the face of gross errors in measurements and non-Gaussian noise, estimators based on the maximum correntropy criterion emerge as an alternative to the traditional least squares method. Grounded in information theory and using non-parametric models for estimating the probability density of measurement errors, maximum correntropy estimators are resilient in the face of outlier data. In this context, this paper proposes a new algorithm for estimation using the maximum correntropy criterion, employing a new technique for adjusting Parzen windows, used as an approximation of the distribution of measurement residuals. Results obtained from tests conducted on the IEEE 30-bus system show that the proposed estimator demonstrated the ability to find accurate estimates even in the presence of spurious measurements, across various scenarios and measurement noises, without measurement removal.