Health Estimation of Individual Capacitors in a Bank With Reduced Sensor Requirements

• Published in: IEEE transactions on industrial electronics (1982) Vol. 66; no. 9; pp. 7250 - 7259
• Main Authors: Gupta, Yugal, Ahmad, Md. Waseem, Narale, Sagar, Anand, Sandeep
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.09.2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Aluminium electrolytic capacitor (AEC); Aluminum; Capacitance; Capacitor banks; Capacitors; Converters; Degradation; Electric potential; Electrolytic capacitors; equivalent series resistance (ESR); Estimation; Extended Kalman filter; extended Kalman filter (EKF); Health; Inverters; Mathematical models; Monitoring; physics of failure (PoF); Sensors; Temperature measurement; Thermal analysis
• ISSN: 0278-0046; 1557-9948
• DOI: 10.1109/TIE.2018.2880725

Aluminum electrolytic capacitors (AECs) are connected in a bank to meet the requirement of capacitance, voltage, and current rating for dc-link of power electronic converters. The performance of capacitor bank deteriorates with time due to the degradation of AECs in the bank. Further, each AEC in the bank degrades differently based on its core temperature, which in turn depends on the current and its equivalent series resistance (ESR). Therefore, health monitoring of a bank would require monitoring of individual AECs in the bank. However, extending the available literature on health monitoring of an AEC to the capacitor bank would require voltage/current sensors for individual AECs. To address the aforementioned issue, this paper suggests a method to assess the health of individual AECs online in the bank without using sensors for each AECs. The proposed method involves estimation of current through each capacitor, which is used to determine the core temperature by using a thermal model of AEC. The core temperature is used for estimation of ESR and capacitance of each capacitor using physics-of-failure based degradation model. For the consistent matching of the degradation model to the actual degradation of the capacitor, the coefficients of the model are sequentially updated. For updating, an algorithm based on extended Kalman filter is used. The proposed scheme is experimentally validated on laboratory prototype of single-phase, grid-connected solar PV inverter under variable load conditions.