New technique of experimental data analysis for early detection of PWM converter faults owing to nonlinear phenomena

Faults owing to nonlinear phenomena concerned with bifurcations in PWM energy converter (PEC) dynamics lead to one of the hardest negative consequences since it suppose the loss of operating process stability and crucial changes in the dynamics. From the practical viewpoint the preferable way out of this problem seems to be in early fault detection based on real-time data analysis while PEC operation. However the analyzed data contain nonlinear and noise components which are “mixed” over time by intricate manner. So, it is necessary at least integrate possibilities of the bifurcation, spectrum and statistical methods. In the paper such new technique allowing a visible reduction of uncertainties on bifurcation diagrams is proposed and discussed.