Efficient Design of Energy Disaggregation Model with BERT-NILM Trained by AdaX Optimization Method for Smart Grid

Abstract

One of the basic conditions for the successful implementation of energy demand-side management (EDM) in smart grids is the monitoring of different loads with an electrical load monitoring system. Energy and sustainability concerns present a multitude of issues that can be addressed using approaches of data mining and machine learning. However, resolving such problems due to the lack of publicly available datasets is cumbersome. In this study, we first designed an efficient energy disaggregation (ED) model and evaluated it on the basis of publicly available benchmark data from the Residential Energy Disaggregation Dataset (REDD), and then we aimed to advance ED research in smart grids using the Turkey Electrical Appliances Dataset (TEAD) containing household electricity usage data. In addition, the TEAD was evaluated using the proposed ED model tested with benchmark REDD data. The Internet of things (IoT) architecture with sensors and Node-Red software installations were established to collect data in the research. In the context of smart metering, a nonintrusive load monitoring (NILM) model was designed to classify household appliances according to TEAD data. A highly accurate supervised ED is introduced, which was designed to raise awareness to customers and generate feedback by demand without the need for smart sensors. It is also cost-effective, maintainable, and easy to install, it does not require much space, and it can be trained to monitor multiple devices. We propose an efficient BERT-NILM tuned by new adaptive gradient descent with exponential long-term memory (Adax), using a deep learning (DL) architecture based on bidirectional encoder representations from transformers (BERT). In this paper, an improved training function was designed specifically for tuning of NILM neural networks. We adapted the Adax optimization technique to the ED field and learned the sequence-to-sequence patterns. With the updated training function, BERT-NILM outperformed state-of-the-art adaptive moment estimation (Adam) optimization across various metrics on REDD datasets; lastly, we evaluated the TEAD dataset using BERT-NILM training.

Sustainability concepts in global high-rise residential buildings: a scientometric and systematic review

“…Campana et al (2017) also developed an optimisation model for the planning of residential urban districts with special consideration of renewables and water harvesting integration. Meanwhile, Çavdar and Feryad (2021) designed and tested an efficient energy disaggregation (ED) model. In order to achieve optimisation goals, Bingham et al (2019) considered the impacts of building envelope upgrades as well as a renewable energy system in the form of photovoltaic (PV) and battery electricity storage.…”

Parameter monitoring of electric energy metering system based on Internet of Things technology

“…This sampling method is tested as follows: two groups of signals are compared by using a dual-channel AC/DC comparator. In the test, the proportion of the front-end current transformer is switched so that the secondary output of the current transformer can always ensure the output of 80mA signal when the current transformer inputs 1A, 0.5A, 0.2A, 0.1A.0.05A.0.02A signals respectively, and then the IU converter converts the 80mA signal into 4V signal and sends it to the channel 1 of the AC/DC comparator [12]. Channel 2 uses a 1A/80mA bipolar current transformer with a fixed ratio in the test process, and converts the current into a voltage signal through an IU converter, and then the small signal is amplified by an amplifier for a programmable gain meter and sent to channel 2 of an AC/DC comparator; After the primary terminals of the current transformers of the two channels are connected in series, the same 1A, 0.5A, 0.2A, 0.1A, 0.05A.0.02A signals are input respectively, and the calculation results are recorded by AC/DC comparators, as shown in Tables 1 and 2 [13].…”

Efficient Localness Transformer for Smart Sensor-Based Energy Disaggregation

“…In particular, self-attention matches every input element with the entire input sequence to compute attention scores. The output sequence is then generated by a decovolutional layer and linear transformation [10]. Nevertheless, for sequence with length l, self-attention operation yields O(l 2 ) complexity in both space and time.…”

“…More recently, transformer models with self-attention mechanism have achieved state-of-the-art performance in multiple natural language processing (NLP) tasks [8]. Inspired by selfattention, transformer-based models are designed to improve NILM performance over long input sequences [9], [10].…”

“…In this sense, they often fail to fully exploit shortterm and local signal patterns, which can result in performance deterioration on multi-status appliances [11], [12]. As a result, current transformer models for NILM only achieve marginal performance gains compared to previous methods [9], [10].…”