Real-Time Implementation of Stockwell Transform in FPGA Platform Using Soft-core Processor Applied to Novelty Detection in Power Quality Signals.

Considering the smart grids establishment, where the presence of nonlinear loads and new power generation sources become increasingly expressive, there is a potential for unknown disturbances occurrence. Besides, given the huge amount of data coming from smart meters, it is important to preserve the relevant information and promote efficient storage of the data. Novelty Detection (ND) techniques can be used to address these challenges. The Stockwell Transform (ST) is a time–frequency distribution that has shown a great ability to detect novelties related to stationarity changes in signals. It can be applied to voltage and current signals from the grid. Therefore, the main contribution of this paper is to provide an FPGA real-time implementation of an ST-based novelty detector. Novelty detection goes beyond classic electrical disturbances well known in the area of power quality, as it relates any novelty event to a stationarity change in the analyzed signal. To perform the implementation in FPGA platform, the use of a soft-core processor is proposed to optimize the hardware resources of the FPGA. In addition, a strategy of voice selection is proposed to reduce the complexity and execution time of the algorithm in real-time implementation, while maintaining the detection capability. The proposed ND technique was implemented in a low-cost FPGA-based functional prototype, where synthesized and field-recorded real signals were applied, attesting the ND performance. The reported tests show that the strategies adopted for real-time implementation provided an optimization of hardware resources and enabled the execution of the ST algorithm and determination of the novelties. [ABSTRACT FROM AUTHOR]