Energy Management in an Agile Workspace using AI-driven Forecasting and Anomaly Detection

Abstract

Smart building technologies transform buildings into agile, sustainable, and health-conscious ecosystems by leveraging IoT platforms. In this regard, we have developed a Persuasive Energy Conscious Network (PECN) at the University of Glasgow to understand the user-centric energy consumption patterns in an agile workspace. PECN consists of desk-level energy monitoring sensors that enable us to develop user-centric models that characterizes the normal energy usage behavior of an office occupant. In this study, we make use of staked long short-term memory (LSTM) to forecast future energy demands. Moreover, we employed statistical techniques to automate the detection of anomalous power consumption patterns. Our experimental results indicate that post-anomaly resolution leads to 6.37% improvement in forecasting accuracy.

The Impact of COVID-19 on the Energy Sector and the Role of AI: An Analytical Review on Pre- to Post-Pandemic Perspectives

“…Power utilities can maintain the power system's normal operational state and deliver a reliable and steady supply of electricity to customers by precisely forecasting the load demand and effectively managing the generation and distribution of electricity. However, one major challenge was the neglect of electrical energy wastages, which were often overlooked in traditional energy systems [61,62].…”

“…Additionally, a lot of energy companies have set up systems for two-way communication between power providers and end users [61,62,65]. This brings benefits for the energy sector, including the ability to use the grid, renewable energy sources, and power systems more efficiently.…”

“…The energy sector saw a sharp decline in oil and gas demand in 2020 as a result of the pandemic, with global oil demand declining by 8.5% and natural gas demand declining by 2.5% [79] In response to the decline in oil demand and prices, there were efforts to transform the oil and gas market from a seller's market to a buyer's market and to re-boost the economy through the use of renewable energy alternatives [39,61,68]. Throughout 2020, there were numerous initiatives undertaken by governments around the world that aimed at reducing emissions from burning fossil fuels through increasing investments in renewable energies such as solar and wind power generation technologies.…”

Enhanced Adversarial Attack Resilience in Energy Networks through Energy and Privacy Aware Federated Learning

“…The traditional centralized strategies have been successfully used in the past for STLF, however, they require transferring all data to a central hub, which inevitably leads to significant network traffic [6]. Additionally, centralized machine learning involves sharing local data with centralized systems, raising concerns about security and privacy [7]. Moreover, adhering to stringent data regulations, such as the EU General Data Protection Regulation (GDPR), introduces its own set of challenges, as noted by Truong et al (2021) [8].…”

FedBranched: Leveraging Federated Learning for Anomaly-Aware Load Forecasting in Energy Networks

“…The proposed method uses differential privacy for security, and performance is evaluated by computing the mean absolute percentage error (MAPE). An LSTM-based neural network was proposed for STLF in an agile workspace at the University of Glasgow, UK, with the help of the Persuasive Energy Conscious Network (PECN) [ 15 ]. The authors of [ 9 ] propose residential energy forecasting using FL and edge computing to ensure the privacy of energy consumption data.…”