Thursday 23 January 2025
The quest for precise measurement has led researchers to develop innovative solutions, and a recent breakthrough in the field of power electronics is no exception. A team of engineers has created a machine learning-based probe skew correction method that can accurately identify and compensate for phase discrepancies in high-frequency BH loop measurements.
BH loops are used to measure the core loss of magnetic components, which is crucial for designing efficient power converters. However, the process is susceptible to probe skew errors, caused by differences in propagation delays between voltage and current probes. These errors can result in significant deviations in measured core losses, leading to inaccurate modeling and design of power electronic systems.
The researchers employed a convolutional neural network (CNN) to develop their correction method. They converted BH loop measurements into image data and trained the CNN to recognize patterns associated with probe skew distortions. The model was tested on two common core materials, 3C90 and N87, and demonstrated remarkable accuracy in identifying probe skew errors.
The proposed method is particularly noteworthy for its ability to correct skew errors at a resolution of 3.5×10-4 degrees, equivalent to approximately 0.02 nanoseconds at 50 kHz. This level of precision is essential for accurate modeling and design of power electronic systems, which are increasingly critical in applications such as electric vehicles and renewable energy.
The researchers utilized an open-source dataset, MagNet, to train their model. This dataset provides experimentally measured waveforms from various materials and operating conditions, allowing the CNN to learn patterns and correlations specific to each material. The team’s approach is transferable across different materials and operating points, making it a robust solution for probe skew correction.
The machine learning-based method offers several advantages over traditional hardware-based de-skew tools. It eliminates the need for custom-designed testing circuits and enables accurate correction of probe skew errors without requiring physical modification of the measurement setup. Additionally, the CNN model can be trained on a variety of datasets, allowing it to adapt to new materials and operating conditions.
The implications of this breakthrough are significant. The proposed method has the potential to revolutionize the field of power electronics by enabling more accurate measurements and modeling of magnetic components. This, in turn, will lead to the development of more efficient and reliable power converters, which are critical for a wide range of applications, from electric vehicles to renewable energy systems.
In summary, the researchers have developed a machine learning-based probe skew correction method that can accurately identify and compensate for phase discrepancies in high-frequency BH loop measurements.
Cite this article: “Accurate Probe Skew Correction Method for High-Frequency BH Loop Measurements”, The Science Archive, 2025.
Power Electronics, Machine Learning, Probe Skew Correction, Bh Loops, Core Loss, Magnetic Components, Convolutional Neural Network, Precision Measurement, Power Converters, Renewable Energy.
