Accelerating power quality signal analysis in a field programmable gate array

Abstract

This thesis presents an analysis and implementation of accelerated power quality signal processing in a field programmable gate array (fpga). The purpose of this study is to show that a medium size FPGA has the necessary resources to perform realtime analysis. The research is part of an ongoing development project in digital monitoring and autonomous correction of faults. Current implementations are based on an external processor or digital signal processor (DSP) and an FPGA. Several alternative system on chip (soc) architectures are reviewed. A combination of a softcore microprocessor and hardware accelerators which natively work on data in a polar vector based format is proposed. Methods are discussed not only with performance in mind, but rapid development time and solutions which are familiar to developers are requirements as well. Implementations are simulated and verified in hardware with artificial stimuli. The result is a design combining a general purpose softcore microprocessor and several hardware accelerators. This yields realtime performance with a footprint to match a medium size FPGA. It is also shown that some signal enhancing functions applied to correct for sensor- and sampling peculiarities can be performed very efficiently in software when working with polar vectors. Further, it is shown that transferring data more efficiently between a hardware accelerator and a microprocessor increases throughput by up to 55%. A method producing accurate results over a larger frequency range is introduced. The method adopts the sample interval to the frequency of the signal by measuring vector velocity. This is both efficient and offers self calibration. Compared to a previously developed implementation it requires significantly less resources. Embedded software developed takes advantage of dedicated hardware and provides an efficient and flexible environment. This allows engineers not familiar with hardware details to take advantage of the platform.