Lifetime prediction models of pulsed capacitor based on capacitance loss

Metallized film capacitor possesses characteristics of self-healing, high reliability, long lifetime and high energy density, which make it widely used in electromagnetic launch (EML) system. Lifetime and reliability of capacitor are the key factors that ensure the stability of the EML. In the capacitor lifetime test, it is a time-consuming work to get enough lifetime data to characterize the performance of the capacitor, because the capacitance loss (CL) caused by single pulse shot is small, and dispersion of experimental lifetime data is inevitable. So it is necessary to analyze the limited experimental data by effective and reasonable lifetime prediction model (LPM) to predict lifetime span and reliability of capacitor. This paper firstly introduces the characteristics of CL in lifetime test. In the early stage of the lifetime test, CL is uniform and steady. With the increasing shot numbers, CL is accelerated due to capacitor degradation and aging. Then it introduces traditional LPMs such as the least square method (LSM), the Weibull distribution model (WDM). LSM is only appropriate on the occasion that CL is uniform and steady, but it cannot analyze the lifetime reliability. WDM works best when large quantities of lifetime experiment data are acquired. New LPMs, also called performance degradation models, such as Birnbaum-Saunders distribution model (BSDM) and Poisson distribution model (PDM). New LPMs fully use CL in each shot, and it could faithfully reflect the work conditions of capacitor. In addition, PDM could even further consider the sudden CL phenomenon. The analysis methods and steps of each model are put forwards. A group of capacitor lifetime and its reliability are analyzed based on all the LPMs above.