Quantization And Recognition Of Nonlinear Dynamic Characteristics Of Distributed DC-DC Switching Converters

Distributed power systems are widely used in micro-grids,electric vehicles,commercial communication systems,computers,multi-electric aircraft and other fields due to their unique advantages of high reliability,easy standardization,easy expansion,easy maintenance,flexible configuration,and substantial cost reduction,and it is one of the mainstream architectures of current power supply modes.In the distributed power system,DC-DC switching converter is its main structural style.As a strong nonlinear system,the nonlinear behaviors such as bifurcation,chaos and coexistent attractors exhibited by the DC-DC switching converter will increase the unpredictability of its output,thereby spreading to other converters and further affecting the overall system stability and reliability.On the other hand,when distributed DC-DC switching converters work in distributed power systems,the definition of non-linear phenomena is difficult to carry out,and there is currently no rapid positioning method.Therefore,this paper will study the different dynamic behaviors of various DC-DC converters in the distributed power system,and combine a symbol arrangement method to analyze the border collision bifurcation and chaotic behavior of different types of converters;then the time-frequency border entropy and support vector machine are used to quantitatively analyze and identify the dynamic behavior of the converter,which can explore and study the nonlinear characteristic mechanism of the distributed DC-DC switching converter.The study in the dissertation can be summarized as follows:(1)Based on discrete modeling with sampling at switching moment,a permutation entropy quantization method for nonlinear behavior of distributed DC-DC switching converters is proposed.Quantitative analysis of nonlinear behavior of Buck converters with three different control methods or structures such as peak current mode control,voltage-current double-loop control,and interleaved parallel type.Furthermore,the quantization results are displayed graphically,and the bifurcation graph and the largest Lyapunov exponent graph are compared in turn to verify the effectiveness of permutation entropy in quantifying the unstable behavior of fast-time scale.(2)Considering the local contradiction between the time domain and the frequency domain,the dynamic behavior of the single-inductor dual-output Boost converter is quantified using time-frequency border entropy.The optimized Variational Mode Decomposition(VMD)and wavelet transform are used to extract the nonlinear characteristics of the system and calculate the time-spectral edge entropy.Combined with the least squares support vector machine,each period from the period doubling bifurcation to chaos is realized.Identification and classification of periodic states.(3)Buck converter loaded by boost converter both under Current mode controlled(CMC)are investigated extensively due to the different modulation.In order to verify the accuracy of the non-linear phenomena of the converter and its quantitative research in this paper,a scatter diagram of the parameter stability domain is drawn,combined with circuit simulation,and the correctness of the numerical simulation quantitative analysis is verified.