Autoencoder Application for Anomaly Detection in Power Consumption of Lighting Systems

Detecting energy consumption anomalies is a popular topic of industrial research, but there is a noticeable lack of research reported in the literature on energy consumption anomalies for road lighting systems. However, there is a need for such research because the lighting system, a key element of the Smart City concept, creates new monitoring opportunities and challenges. This paper examines algorithms based on the deep learning method using the Autoencoder model with LSTM and 1D Convolutional networks for various configurations and training periods. The evaluation of the algorithms was carried out based on real data from an extensive lighting control system. A practical approach was proposed using real-time, unsupervised algorithms employing limited computing resources that can be implemented in industrial devices designed to control intelligent city lighting. An anomaly detection algorithm based on classic LSTM networks, single-layer and multi-layer, was used for comparison purposes. Error matrix calculus was used to assess the quality of the models. It was shown that based on the Autoencoder method, it is possible to construct an algorithm that correctly detects anomalies in power measurements of lighting systems, and it is possible to build a model so that the algorithm works correctly regardless of the season of the year.