Lightweight Gramian Angular Field classification for edge internet of energy applications

Abstract

With adverse industrial effects on the global landscape, climate change is imploring the global economy to adopt sustainable solutions. The ongoing evolution of energy efficiency targets massive data collection and Artificial Intelligence (AI) for big data analytics. Besides, emerging on the Internet of Energy (IoE) paradigm, edge computing is playing a rising role in liberating private data from cloud centralization. In this direction, a creative visual approach to understanding energy data is introduced. Building upon micro-moments, which are timeseries of small contextual data points, the power of pictorial representations to encapsulate rich information in a small two-dimensional (2D) space is harnessed through a novel Gramian Angular Fields (GAF) classifier for energy micro-moments. Designed with edge computing efficiency in mind, current testing results on the ODROID-XU4 can classify up to 7 million GAF-converted datapoints with ~ 90% accuracy in less than 30 s, paving the path towards industrial adoption of edge IoE. Supplementary Information The online version contains supplementary material available at 10.1007/s10586-022-03704-1.

Analyzing domestic energy behavior with a multi-dimensional appliance-level dataset

“…Because of the 2D nature of the data, a plethora of DL methods can be employed, e.g., Convolutional Neural Networks (CNN), Long-Short Term Memory (LSTM), etc. to classify the data and gain useful insights with significantly lower computational time [20].…”

“…5, sample GAFs are classified based on a modified EfficientNet-B0 classifier (CNN with transfer learning). The algorithm, explained in detail in [20], classifies the data based on average normalized values of the GAF data, i.e., a GAF is normal is its average normal value within its duration window is less than a specified abnormality threshold, and vice versa. It is worth mentioning that the model is trained on a Linux instance while the testing has been carried out on an ODROID-XU4 board to signify the merits of 2D GAF classification in terms of classification computational performance, e.g.…”

Edge-cloud collaborative multi-level fault diagnosis based on stacked sparse autoencoder

“…Edge-cloud collaboration has been introduced for fault diagnosis since its conceptual proposal. Several methods have been developed, including multi-level fault diagnosis [1,2] and edge-inferencing after cloud-trained [3][4][5][6][7][8][9][10][11][12][13] . Multi-level fault diagnosis is to deploy simple models at the edge for rough classification to filter normal data and reduce the inference time.…”

“…When a fault occurs, a second-level fault diagnosis is performed in the cloud to infer the details of a fault [1,2] . Edgeinferencing after cloud-trained trains a diagnostic model in the cloud and then deploys it to edge for inference [3][4][5][6][7][8][9][10] . However, a lightweight model is usually adopted due to the limited computing resources on the edge side [11][12][13] .…”

Visualization of radiation intensity sequences for space infrared target recognition

“…A potential solution involves the reorganization of the signal and explicit modeling of the temporal dependencies to gain a deeper understanding of the dynamics in a time series. Certain studies [8][9][10][11][12][13] have demonstrated the ease of extracting of information from high-dimensional data. Silva et al [8,9] extended the recurrence plots (RP) paradigm for a time series with compressed distance and proposed to process the RP as grayscale texture images.…”