Identifying Coexisting Attractors and Quantifying Dynamics Characteristics of DC–DC Converter Based on Improved Variational Mode Decomposition and Wavelet Transform

As a nonlinear system, the dc–dc converter exhibits nonlinear behaviors, such as bifurcation, chaos, and coexisting attractors, which greatly reduces its reliability. Therefore, predicting and identifying these nonlinear behaviors is of great significance to the application of dc–dc converter. And many effective methods have been proposed for identifying nonlinear behaviors. However, the traditional analytical methods have certain defects in identifying coexisting attractors and quantifying the dynamic behavior of transient states in the time domain. Therefore, this article adopts the improved variational mode decomposition and wavelet transform methods to extract the time-frequency features of the dc–dc converter with the influence of the coexisting attractors, and combines the quantification method of the wavelet time-frequency spectral border entropy to quantify the nonlinear behaviors. Moreover, a single-inductor dual-output boost converter is taken as an example to illustrate the extraction and quantization of time-frequency characteristics of transient states by the proposed method. The result demonstrates that the method can clearly identify the periodic state characteristics of the system with the influence of coexisting attractors in the time-frequency domain, and can quantify them accurately. Finally, an experiment verifies the theoretical analysis.