Estimação de harmônicos e inter-harmônicos baseada na técnica de invariância rotacional aplicada a sub-bandas do espectro

Abstract

The Estimation of Signal Parameters via Rotational Invariance Technique (ESPRIT), which is a high resolution parametric estimation method, has gained attention of the researchers due to its high performance for harmonic and inter-harmonic parameters’ estimation of power system signals. Nevertheless, this method presents high computational complexity and is very sensitive to colored noise. With the aim of reducing computational complexity and improving the performance for colored noise scenarios, this work presents a new harmonic and inter-harmonic parameter estimation method based on rotational invariance technique and multirate signal processing through the use of filter banks, called ESPRIT-FB. Adopting the classical multirate signal processing approach, the proposed method decimate the signal in each sub-band, which reduces the number of samples of the signal to be processed in each sub-band and, as a result, reduces the size of the data matrix built by the rotational invariance technique. Moreover, the proposed method can be used in its dynamic form, Dynamic ESPRIT-BF (D-ESPRIT-BF), where the filter bank is designed according to the signal characteristics, or the method can be used with a fixed pre-designed filter bank. It was show that the computational complexity of the proposed method, for the fixed filter bank case, can be up to L2 times smaller than the conventional ESPRIT, where L is the decimator factor used in the sub-bands. The proposed method was evaluated with synthetic and real power system signals and compared with with the conventional ESPRIT. The results showed that the proposed method improves the accuracy and the precision and reduces the computational complexity with reference to the conventional one. The proposed method was also compared to Wavelet Sliding Window ESPRIT Hybrid method, called here W-SWESPRIT, showing better performance for the studied cases. It is also worth to mention, that colored noise scenarios were simulated, and the results showed that the ESPRIT-BF highly improves the estimation performance in comparison with the conventional one. Therefore, this work is an important contribution to the area, improving the performance, reducing the computational complexity and enlarging the range of applications of the ESPRIT method.