A Transfer Learning Framework for Predictive Energy-Related Scenarios in Smart Buildings

Abstract

Human activities and city routines follow patterns. Transfer learning can help achieve scalable solutions towards the realisation of smart cities accounting for similarities between regions, domains, and activities. In this study, we propose a Transfer Learning-based framework for smart buildings to test this hypothesis in energy-related problems. Our framework has two major components: the network creation and the transferable predictive model. In order to create the network that groups buildings sharing characteristics, we evaluated two strategies: a novel clustering algorithm for mixed data, k-prod, and clustering the image-based representation of time series. Then, a combination of Long Short Term Memory and Convolutional Neural Network was trained on the centroids of the clusters for energy consumption prediction. The Coefficient of Variation of the Root Mean Squared Error (CVRMSE) of the predictions in such clusters vary between 3.85 and 58.85 %. The obtained parameters were transferred to the rest of the buildings for predictive purposes, finding accurate results in buildings with little data. Our framework deals with insufficient training data since parameters from scenarios with more sensors can be received. It also carries out state-of-the-art performance on 3 datasets from different sources having in total 533 rooms/buildings and two energy efficiency domains: consumption prediction reducing the CVRMSE in a 21.6 %, and air conditioning usage prediction moving from a 4.18 % to a 0.28% CVRMSE. Our framework extracts more knowledge from available IoT deployments, so that smartness could be spread between environments at a fewer cost given that less individual effort will be needed.

Security Threats and Promising Solutions Arising from the Intersection of AI and IoT: A Study of IoMT and IoET Applications

“…The rise of IoET and the spread of smart meters, together with the advances in AI, render access to and mining of EC data possible at the right time for early ETD. Deep learning methods using convolutional neural networks (CNNs) have been proven to successfully extract latent features in time series EC data for accurate representation and classification [10]. In the context of ETD, theft occurrences are less than 10% of EC data, a major challenge to data-driven CNN methods, which tend to replicate and even reinforce the bias in a skewed dataset.…”

PLEIAData: consumption, HVAC, temperature, weather and motion sensor data for smart buildings applications

“…In this line of work, where the focus is energy consumption and the development of control strategies, there are lots of available public datasets. For example, in 5 a transfer learning approach is taken, using data from buildings in different regions via various open datasets. First, from The Building non-residential Data Genome Project 6 where data is collected in 1238 buildings mainly from the United States, but also from Europe, Asia and Oceania.…”

“… Developing a demand flexibility strategy coupling consumption and temperature forecasts 17 . Applying transfer learning to not sensorized or partially sensorized buildings with similar characteristics 5 . Implementing reinforcement learning techniques adding, for example, electricity prices to this dataset 18 .…”

“…Applying transfer learning to not sensorized or partially sensorized buildings with similar characteristics 5 .…”

Design and Implementation of an Interoperable Architecture for Integrating Building Legacy Systems into Scalable Energy Management Systems

“…AI techniques, such as Fuzzy Logic [15] or neural networks, offer forecasting of energy production and consumption and allow the optimization of energy management schedules in order to enhance buildings' energy efficiency in terms of cost savings and environmental impact [15]. Additionally, AI techniques can transfer the extracted knowledge on energy consumption between buildings with different levels of maturity with regards to their IoT deployments [16]. While the essence and definitions of resilience in power systems [17,18] and interdependent infrastructure systems [19] are still under development, the resilience benefits of such AI-based BMS architectures can be foreseen.…”