Towards Digital Energy Metering Systems by Means of Artificial Intelligence

This article presents a new and innovating method to upgrade the existing energy metering system by utilizing an intelligent optical character recognition (OCR) algorithm. The proposed approach assures small costs of the entire energy system's refurbishment, and it can be implemented on both old and new energy meters. The flow is relatively simple and straightforward: an artificial neural network is used to process and recognize the digits from a photo of the meter's analog display and populate a custom database, from which the data will be later retrieved and used to give different stakeholders consumption insights and patterns, graphs, energy efficiency solutions etc. For the OCR algorithm and transfer learning processes a mature Python library, namely Tesseract, was used. The developed software solution was tested on various gas and electricity meters and the results are presented.