Uma nova abordagem em modelagem e simulação de sensor de corrente indutivo aplicado a medição "on-line" de descargas parciais

Abstract

The present study describes the modeling and simulation of an inductive current sensor, called High Frequency Current Transformer (HFCT), for simulated partial discharges (PD) under noisy conditions, in order to replicate unconventional partial discharge measurements. By means of this dissertation the investigation of the variation of constructive and electrical parameters and the response of the inductive current sensor models were carried out in order to evaluate the deterioration or improvement of the response to the noisy signal, i.e. verify if it increase or descrease the mean-squared error and cross-correlation between the response signal and the applied signal to better understand how those parameters affect the sensor's Transfer Function. Therefore it includes: 1. Generation and Noise Addition: theoretical investigation of electromagnetic interference (EMI) in the measurement of partial discharges in the field and characterization as AWGN (Additive White Gaussian Noise), besides the creation of several signals (PD1 to PD10) with different levels of Signal-to-Noise Ratio (SNR) and partial discharge simulation using the discharge curve of the RC derivative circuit; 2. Sensor Modeling and Reference Model Response Analysis: transfer function modeling (TF) based on the Laplace Transform (s-domain), Electromagnetism Laws and by obtaining equivalent circuit of grouped parameters and constructive and electrical parameters of the sensor. In addition, the Control Group (analysis reference) was created, composed of the PAD model, and the validation of this model was carried out according to the time and frequency response and by the Root Locus Method; 3. Evaluation of Performance of Experimental Models and Validation of Hypothesis: creation of Experimental Group, formed by the N30, RL250, mur2300 and RLNmur models, through variation of the PAD model's constructive and electrical parameters and evaluation of the response to the applied signals (PD1 to PD10), through cross-correlation and mean squared error. In conclusion, it was possible to perform a hardware signal preprocessing through the variation of electrical or constructional parameters, resulting in a decrease of error and similarity increase between the response signal and the pure partial discharge signal applied even under conditions of low SNR.