FPGA implementation of the automatic multiscale based peak detection for real-time signal analysis on renewable energy systems

This paper proposes an FPGA implementation of a simple and reliable method for real-time peak detection of time series data. Focusing on an existing off-line method called Automatic Multiscale Based Peak Detection (AMPD), we show, how the original AMPD algorithm is modified to be an on-line method and how it is applied on an FPGA. To validate the accuracy of our design, simulation results in a case study are demonstrated. The evaluation in terms of peak detection latencies and required FPGA resources reveal that our design can detect peak points with a latency of up to 160 nsec with small resource usage, suggesting that FPGAs are a good solution of high-speed data analysis on renewable energy systems.