데이터 기반 방식과 모델 기반 방식을 이용한 DC/AC 컨버터의 입력 커패시터 ESR 추정 비교 연구

Abstract

In this paper, the performance of the input capacitor state estimation algorithm of the DC/AC single-phase converter implemented using the model-based method and the data-based method is compared. The state of the capacitor is estimated through two existing algorithms implemented as a model-based method and the data based algorithm implemented as a data-based method. By applying the capacitor voltage and capacitor current data collected through the experiment to three methods, the equivalent series resistor (ESR) component was derived. The two model-based methods used in this paper are the method of estimating the ESR component using a bandpass filter and the method of using the average power loss. The data-based method estimates ESR using a component twice the fundamental frequency and a component twice the switching frequency in the frequency domain as input to the neural network. Comparing the results, it was confirmed that in the case of the model-based method, the deviation of the error according to the ESR was very large. On the other hand, in the case of the data-based method, it was confirmed that the estimation was made more accurately using data of a large amount of information. Therefore, it was confirmed that the data-based method showed excellent estimation results in all fields