Remaining useful life (RUL) is crucial to the condition and health monitoring. This paper proposes an adaptive RUL prediction method for DC-link film capacitors for power electronic applications. By using the proportional hazards model framework, this method integrates the degradation data and the time-to-failure data, to quantify the component failure behavior in a probabilistic way. It employs a mixed-effects model to characterize the degradation behavior of capacitors. The hazard rate is applied to characterize the likelihood of capacitor failure. The operational conditions are incorporated into the hazard model to capture the influence on the RUL. The Bayesian updating mechanism is developed to calibrate and tailor the offline model for the in-situ component, enabling the adaptive RUL prediction along with sequential monitoring data in real time. The method is experimentally verified with DC film capacitors subjected to accelerating humidity conditions. It is accompanied by an online tool at https://rul-capacitor.streamlit.app that can interactively investigate the method details.

Accurate temperature estimation of permanent magnet synchronous motors serves as the fundamental basis for designing effective thermal management strategies. Model-based estimation methods exhibit superior real-time performance, but the intricate modeling process requires substantial expert knowledge and lacks versatility. Conversely, data-driven estimation methods, while offering flexibility, often lack physical implications in terms of system dynamics. This paper proposed a structured linear neural dynamics model for motor temperature estimation. This model is data-driven, with prior knowledge integrated into its structure, which preserves flexibility while guaranteeing system stability through the Perron-Frobenius theorem. Additionally, this paper achieves the decoupling of control input from state transitions and the embedded deployment of this model. The method is validated with a real dataset. The lightweight feature is demonstrated by the implementation of an STM32 Microcontroller with 1.808 KB and 27 mW. The paper is accompanied by the open source data and code at GitHub: https://github.com/ms140429/Explainable-Neural-Dynamics-Model

In this paper we propose a new evaluation method for revealing the inner structure information from the measurements of the thermal transient response of semiconductor devices. In contrast with the conventional methods based on the frequency domain deconvolution, the proposed method casts the problem as a regularized least squares problem for an integral equation directly in the time domain. The regularization is selected based on the knowledge of certain physical quantities involved in the measurement. We observe that the thermal time constant spectrum produced by the proposed method tends to have a better capability of identifying two adjacent parameters. With the benchmark of the true values of the thermal networks, the structure function of the proposed method reduces the relative error by an order of magnitude. A series of simulations based on different parameters and structures verify the viability and performance of the proposed method. A proof-of-concept experiment based on a commercial power semiconductor device also validates the effectiveness of the proposed method.

This is a preprint version of the manuscript submitted to IEEE on June 4, 2020. This paper gives an overview of the Artificial Intelligence (AI) applications for power electronic systems. The three distinctive life-cycle phases, design, control, and maintenance are correlated with one or more tasks to be addressed by AI, including optimization, classification, regression, and data structure exploration. The applications of four categories of AI are discussed, which are expert system, fuzzy logic, metaheuristic method, and machine learning. More than 500 publications have been reviewed to identify the common understandings, practical implementation challenges, and research opportunities in the application of AI for power electronics.