Low Voltage Warning System for Stand-Alone Metering Station Using AI on the Edge

Abstract

Artificial intelligence is used in many different aspects of life these days. It is also increasingly used to create intelligent embedded devices. Their main task is to demonstrate intelligent features using limited hardware resources. This paper introduces the possibility of using AI on the Edge as a system to warn of low voltage on the battery in a stand-alone metering station powered by solar panels. The paper presents theoretical knowledge about artificial intelligence on the Edge, time series and algorithms in time series forecasting. In addition, a practical approach is presented in the Approach section. It also describes the stand-alone measurement station that was used, what type of data was collected and used, and what type of the time series forecasting algorithm was used. The main advantage of the approach is the practical use of Holt’s Linear Trend method in battery voltage forecasting. Work on the project is still in progress therefore the results presented in the paper are generated during simulation.