A4 Article in conference proceedings

Implementation of K-nearest Neighbor on Field Programmable Gate Arrays for Appliance Classification




List of Authors: Amleset Kelati, Hossam Gaber, Juha Plosila, Hannu Tenhunen

Conference name: International Conference on Smart Energy Grid Engineering

Publication year: 2020

Journal: IEEE international conference on smart energy grid engineering

Book title *: 2020 IEEE 8th International Conference on Smart Energy Grid Engineering (SEGE)

ISBN: 978-1-7281-9913-9

eISBN: 978-1-7281-9912-2

ISSN: 2575-2677

DOI: http://dx.doi.org/10.1109/SEGE49949.2020.9181975.

URL: https://ieeexplore.ieee.org/document/9181975


Abstract

Accurate appliance energy consumption information can perform with the Non-Intrusive Appliances Load Monitoring (NIALM) system. However, faster and advanced appliance classification accuracy can be enhanced by the implementation of the k-nearest neighbor (k-NN) classifier in hardware. A field-programmable gate array (FPGA) hardware implementation can speed up the processing time with a high level of performance accuracy. The result proved that the HLS-based solution has reduced design complexity and time for cost-effectiveness. The Plug Load Appliance Identification Dataset (PLAID) is used as a benchmark for the implementation. The selected appliance identification is implemented using Xilinx Zynq-7000 and the HLS-based solution has used an area of 37.1% for LUT and 21% for FF from the available chip. Thus, the implementation improved the cost and classification accuracy with a processing time of 5.9 ms and the consumed power was 1.94 W.


Last updated on 2021-24-06 at 10:21