Impedance Estimation Using Randomized Pulse Width Modulation and Power Converters

With the adoption of technologies such as alternative energy production, DC power grids, and electric vehicles, the use of high power switching converters has seen a dramatic increase. These power converters serve many rolls such as grid-tied inverters in solar farms, high power charging for electric vehicles, motor drives for industrial applications, and DC links in transmission systems. With the increased prevalence of such devices, it is only natural to attempt to optimize their operation. As with any level of converter, it is desirable to have accurate control over the generated voltages and currents. Often, these controllers implement some form of predictive control which requires knowledge of system parameter values to operate properly. Due to several factors, including temperature and component non-linearity, these component values can vary during normal operation. This can lead to degradation of closed loop control and system instabilities. If one is able to measure system parameters while the converter is operating, control parameters can be updated in real time to optimize the system performance. A significant percentage of the size and cost of switching converters are filter elements meant to reduce the amount of noise injected into other attached circuits, or in the case of grid-tied converters, noise injected into the grid. As power levels increase, the size, cost, and power lost in the filter becomes greater. To minimize these negative effects, methods have been developed that reduce harmonic injections, thus allowing for smaller filter elements. One such technique is Randomized Pulse Width Modulation which removes the large harmonic spikes present in standard switching systems, and replaces them with a wide frequency energy spectrum. The objective of this research is to examine the feasibility of online impedance identification by combining and modifying existing technologies. Specifically, Randomized Pulse Width Modulation and Wideband System Identification techniques are used to simultaneously reduce system noise and create an estimation of system filter element impedances. This allows for the reduction of the filter size while simultaneously providing a real-time estimate of the filter impedance with the goal of better feedback control performance