Capacitor Lifetime Extension in Power Converter Systems Using Neural Networks

Alemdar O. S., Şahin İ., Oner M. U., Altun O., Keysan O.

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, vol.72, no.11, pp.11250-11260, 2025 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 72 Issue: 11
  • Publication Date: 2025
  • Doi Number: 10.1109/tie.2025.3553174
  • Journal Name: IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Business Source Elite, Business Source Premier, Communication Abstracts, Compendex, Computer & Applied Sciences, INSPEC, Metadex, Civil Engineering Abstracts
  • Page Numbers: pp.11250-11260
  • Keywords: Artificial neural networks, capacitor, interleaving, lifetime, reliability
  • Middle East Technical University Affiliated: Yes

Abstract

Capacitors are one of the main components that limit the overall reliability of a power electronics converter. The major parameter limiting the lifetime of a capacitor is the hotspot temperature, which is directly related to the amount of ripple current flowing through the capacitor. For converters employing paralleled power stages, PWM interleaving has been traditionally applied to minimize capacitor current ripple. In multiinput, single-output power converters, PWM interleaving can also be employed to minimize the common capacitors' current ripple. In these systems, finding the optimal phase shift is not trivial though. In this article, neural networks yield optimal phase shift values to operate the common capacitor at the minimum ripple state, enabling capacitor lifetime extension. The proposed technique was validated on a three-level Boost converter with 2 and 3 cells. The effectiveness of the proposed technique is shown through experimental results.