Preprocessing for Unintended Conducted Emissions Classification with ResNet

Characterization of Unintended Conducted Emissions (UCE) from electronic devices is important when diagnosing electromagnetic interference, performing nonintrusive load monitoring (NILM) of power systems, and monitoring electronic device health, among other applications. Prior work has demonstrated that UCE analysis can serve as a diagnostic tool for energy efficiency investigations and detailed load analysis. While explaining the feature selection of deep networks with certainty is often not fully comprehensive, or in other applications, quite lacking, additional tools/methods for further corroboration and confirmation can help further the understanding of the researcher. This is true especially in the subject application of the study in this paper. Often the focus of such efforts is the selected features themselves, and there is not as much understanding gained about the noise in the collected data. If selected feature and noise characteristics are known, it can be used to further shape the design of the deep network or associated preprocessing. This is additionally difficult when the available data are limited, as in the case which the authors investigated in this study. Here, the authors present a novel work (which is a proposed complementary portion of the overall solution to the deep network classification explainability problem for this application) by applying a systematic progression of preprocessing and a deep neural network (ResNet architecture) to classify UCE data obtained via current transformers. By using a methodical application of preprocessing techniques prior to a deep classifier, hypotheses can be produced concerning what features the deep network deems important relative to what it perceives as noise. For instance, it is hypothesized in this particular study as a result of execution of the proposed method and periodic inspection of the classifier output that the UCE spectral features are relatively close to each other or to the interferers, as systematically reducing the beta parameter of the Kaiser window produced progressively better classification performance, but only to a point, as going below the Beta of eight produced decreased classifier performance, as well as the hypothesis that further spectral feature resolution was not as important to the classifier as rejection of the leakage from a spectrally distant interference. This can be very important in unpredictable low-FNR applications, where knowing the difference between features and noise is difficult. As a side-benefit, much was learned regarding the best preprocessing to use with the selected deep network for the UCE collected from these low power consumer devices obtained via current transformers. Baseline rectangular windowed FFT preprocessing provided a 62% classification increase versus using raw samples. After performing a more optimal preprocessing, more than 90% classification accuracy was achieved across 18 low-power consumer devices for scenarios in which the in-band features-to-noise ratio (FNR) was very poor.

A Brief Introduction in the Mitigation of Conducted Electromagnetic Interference lssues

This paper deals with the mitigation of the influence of electromagnetic conducted emissions in low voltage grids, which can be performed using different filtering methods. Due to the relatively young age of the electromagnetic compatibility domain, the specific terminology is not yet fully consecrated. That is why the specific literature abounds in a bunch of definitions and notions, incomplete, redundant, or worse, even contradictory. Therefore, all over this paper, the terminology from the successive issues of the standard IEC 60050-161 International Electrotechnical Vocabulary, is used. The introductory section presents generalities regarding the broader context of electromagnetic compatibility in which the paper fit. Section II is devoted to measurement techniques and measuring equipment used in conducted electromagnetic interference tests, specifically for electromagnetic emissions that flow in/from the equipment under test through power lines in the standardized frequency range from 100 kHz to 30 MHz. These measurement techniques and equipment are further used in the next section which presents electromagnetic interference experiments, performed on an induction motor driven by a frequency converter. To mitigate the conducted electromagnetic emissions to fit into the standard limits, a cascade of two EMI filtering cells has been designed and implemented. This demonstrated the usefulness and effectiveness of mains EMI filters in low voltage power applications. The experiment also demonstrated that in some cases it would be necessary to retrofit more than one filtering cell

Experimental Measurement of Conducted Emissions Generated by Static Converters in Common and Differential Modes

The commutation of semiconductors and their interactions with parasitic elements linked to the environment is known to be the main source of interferences conducted in power electronic converters. Thus, identifying these sources of interferences in commutation cells and determining the level of these conducted emissions generated in energy conversion systems presents a major challenge for designers. In this work, a study of the electromagnetic interferences (EMI) generated by the association of serial rectifier-chopper connected to a Line Impedance Stabilization Network (LISN) is proposed, in which a determination of these interferences is presented in both common and differential modes. All simulations are carried out using the LT-spice software and the results obtained are validated by experimental measurements realized at APELEC laboratory (University of Sidi Bel-Abbes, Algeria).

Reducing Conducted Emissions at the Output of Full-Bridge DCDC Converters with High Voltage Steps

In this work, we analyze the impact of output filter design techniques aimed to reduce conducted emissions at the output of a DCDC power converter. A thorough analysis, based on high-frequency circuit models of the converter, is performed to assess expected improvements offered by different design strategies. This analysis is then confronted with measurements of conducted emissions at the output of a 300 W 48 V to 12 V Phase Shift Full Bridge (PSFB) prototype. Those experimental results demonstrate that a symmetric arrangement of the output LC filter and a direct bonding of the return output terminal of the converter to chassis are effective to reduce common mode conducted emissions at the output. Those results also demonstrate that the symmetry of the output LC filter can reduce conducted emissions in differential mode at high frequencies, where common mode to differential mode conversion is the predominant contribution to differential mode noise. However, direct bonding to chassis of the return output terminal may be ineffective at high frequencies due to the parasitic inductance associated with this connection. Main conclusions drawn for this analysis are applicable in general for isolated converters with a high voltage step between high and low voltage sides. Since the techniques of reduction of conducted emissions studied here do not increase the number of filter components, they are especially suitable for applications where high power density is an important requirement, e.g., aerospace or automotive applications.