Diagnostics of DC-DC Converters with a Quasi-Impedance Link

Abstract

The paper is devoted to the development of a method for DC-DC converters with a quasi-impedance link diagnostics. The simulation of the operation of converters was carried out, which allowed to receive diagnostic information in the form of time charts for the normal mode and for various type of failure. The ability to determine the presence of Walsh spectrum disturbances for the steady state is shown.The research was carried out using the model of a quasi-impedance DC-DC converter implemented in the branching package MATLAB R2014a mathematical system - Simulink R2014a. Simulation results showed that since in the steady state the voltage of the secondary winding of a linear transformer in the form is similar to the Walsh function wal(a, t), then it is advisable to apply this transformation to the analysis in the spectral region. In the case of the complete coincidence of the Walsh basis function and the diagrams, the spectrum will contain only 1 count. This made it possible to predict that determining the state of failure by comparing the known spectrum of the normal state and the current one would take less time than comparing the time charts. The Walsh spectrum was calculated at intervals of 2 periods of voltage change (0-0.04 s). The diagrams of the converter's operation are given for various possible emergency transitions in this converter, such as breakages, breakdowns, short circuits, and their Walsh spectra are given.It is proposed to perform methods for identifying the type of malfunction in the converter by calculating the mean square error and the Euclidean distance for the current and operating modes in the time and spectral domain of the Walsh transform and comparing these values with the pre-calculated ones. It is found that by the Walsh spectrum it is easy to determine the presence of malfunctions - the spectrum for a normal mode has a unique non-zero count of the Walsh spectrum. The appearance of other non-zero components of the spectrum characterizes a malfunction or change in the parameters of the circuit. It is established that the identification of a specific malfunction should be performed using the Euclidean distance criterion in the spectral region, since in this case a 9.6 fold greater relative difference is obtained even for similar problems, as compared to the case in the time domain analysis.Research results are of interest for the development and optimization of diagnostics of DC-DC converters with quasi-impedance link, devices based on them, systems with alternative energy sourcesRef. 11, fig. 3, tabl. 2.