The colloquium of the Center for Intelligent Power and Energy Systems (CiPES) was held on May 20, 2022. This is a monthly event where we invite students and experts in the area of power and energy to give presentations about their research.
Today the presenter is Mr.Xinguo Zhang from PSPAL. He gave a talk on his research: "Online Capacitance Estimation Method in DC-DC Converters with Characteristic Frequency Injection for Optimal Sensitivity". After the presentation, the audience discussed with the presenter about the details of his research.
Mr. Xinguo Zhang joined PSPAL in July 2020. He is currently a master student in PSPAL (starting from September 2020). He received the B.S. degree of Electrical Engineering and Automation from Fuzhou University, Fuzhou, Fujian, China, in Summer 2019. His research interests include fault diagnosis and condition monitoring of power electronic circuits.
The abstract of this talk is as follows:
DC-DC converters are widely adopted in renewable energy, transportation electrification, data communication, and industrial electrification. The reliability and safety of the DC-DC converters are of great importance. Compared to “hard faults” such as component failures or open/short circuit faults, “soft faults” such as parametric faults (parameter drifts due to aging/degradations) are usually more challenging to be detected. Nevertheless, detection of “soft faults” is essential since “soft faults” may evolve into hard faults without proper interference. Therefore, parametric fault detection (also known as fault prognosis, condition monitoring) methods need to be adopted to monitor the degradation of components. Specifically, the aluminum electrolytic capacitor aging process is a common parametric fault due to the loss of electrolyte. Generally, the 20% capacitance reduction or 2 to 3 times increase of the ESR are indicators of end of life of aluminum electrolytic capacitors. Therefore, accurate capacitance estimation is important to ensure reliable operation of converters.
The existing parametric fault detection methods can be mainly classified into offline and online methods. Offline methods usually interrupt normal operation of converters; afterwards electric signals with certain characteristics can be applied to the component of interest for parameter identification. To avoid interruptions of normal operation of converters, researchers also proposed online methods. Online methods generally inject signals within a wide frequency range to estimate parameters. However, the key frequency that enables accurate estimation of capacitance still requires further studies.
Therefore, an online capacitance estimation method based on characteristic frequency injection is proposed. First, the key challenges of capacitance estimation methods using the switching frequency information are discussed: the parameter sensitivity is insufficient. Next, the injection frequency is derived based on the optimal sensitivity, to provide guidance on the selection of the injection frequency. Finally, the capacitance is estimated based on model at the injection frequency.