Optimal Current Control Strategy for Reliable Power Electronics Converters: Frequency-Domain Approach

TL;DR

This paper tackles semiconductor fatigue in power converters caused by thermal cycling and introduces a frequency-domain damage framework integrated into current control. It combines a Foster RC electro-thermal model with a single-moment spectral damage method and fatigue relationships (Coffin–Manson, Basquin) within an H∞-synthesized reliability controller. Simulations on a 2-level, 3-phase inverter show that the reliability design preserves current tracking while reducing junction-temperature swings and extending estimated device lifetime, validated by Monte Carlo analyses. The results provide a practical pathway to reliability-guided, sustainable power electronics operation.

Abstract

Power electronics converters are key enablers in the global energy transition for power generation, industrial and mobility applications; they convert electrical power in a controlled, reliable and efficient manner. The semiconductor switching devices, at the core of power converters, are the most likely component to fail due to the damage caused by the current-induced temperature cycling. Damage models of semiconductors have been developed and employed to study their reliability, improve their design and to estimate the lifetime of the converter in various power applications. However, those models can offer more if employed in the design of strategies to actively operate the converter. Specifically, properly controlling the current, and hence the temperature cycling, can effectively contribute to reducing the accumulated damage in the semiconductor and increase its reliability and lifetime. In this paper we propose a novel current control approach that integrates reliability requirements into the design framework, based on a frequency-domain model of the semiconductor damage.

Figures (11)

• Figure 1: One-line diagram of a DC-AC inverter showing the control system implemented in the dq-coordinate frame.
• Figure 2: (a) A typical package of a SiC MOSFET power module. The Foster thermal model consisting of sequential RC systems represents the heat transfer through the semiconductor module. (b) The bondwire lift-off due to thermal cycling (courtesy of Fraunhofer IISB).
• Figure 3: Fatigue Analysis: (a) Stress versus number-of-cycles-to-failure ( the Wöhler-curve or S-N curve), (b) The concept of the single-moment method for damage estimation by decomposing the power spectral density (PSD) of the stress signal into contribution of a set of narrow-band infinitesimal spectrals.
• Figure 4: The $W(\omega)$ evaluated at different frequencies, reflecting the contribution of damage at each frequency.
• Figure 5: The sensitivity function upper bound for the two control modes: the performance (red) and reliability (blue) designs.