[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
You seem to have javascript disabled. Please note that many of the page functionalities won't work as expected without javascript enabled.
 
 
Sign in to use this feature.

Years

Between: -

Subjects

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Journals

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Article Types

Countries / Regions

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Search Results (943)

Search Parameters:
Keywords = smart meter

Order results
Result details
Results per page
Select all
Export citation of selected articles as:
31 pages, 7160 KiB  
Article
Resource Optimization for Grid-Connected Smart Green Townhouses Using Deep Hybrid Machine Learning
by Seyed Morteza Moghimi, Thomas Aaron Gulliver, Ilamparithi Thirumarai Chelvan and Hossen Teimoorinia
Energies 2024, 17(23), 6201; https://doi.org/10.3390/en17236201 - 9 Dec 2024
Viewed by 530
Abstract
This paper examines Connected Smart Green Townhouses (CSGTs) as a modern residential building model in Burnaby, British Columbia (BC). This model incorporates a wide range of sustainable materials and smart components such as recycled insulation, Photovoltaic (PV) solar panels, smart meters, and high-efficiency [...] Read more.
This paper examines Connected Smart Green Townhouses (CSGTs) as a modern residential building model in Burnaby, British Columbia (BC). This model incorporates a wide range of sustainable materials and smart components such as recycled insulation, Photovoltaic (PV) solar panels, smart meters, and high-efficiency systems. The CSGTs operate in grid-connected mode to balance on-site renewables with grid resources to improve efficiency, cost-effectiveness, and sustainability. Real datasets are used to optimize resource consumption, including electricity, gas, and water. Renewable Energy Sources (RESs), such as PV systems, are integrated with smart grid technology. This creates an effective framework for managing energy consumption. The accuracy, efficiency, emissions, and cost are metrics used to evaluate CSGT performance. CSGTs with one to four bedrooms are investigated considering water systems and party walls. A deep Machine Learning (ML) model combining Long Short-Term Memory (LSTM) and a Convolutional Neural Network (CNN) is proposed to improve the performance. In particular, the Mean Absolute Percentage Error (MAPE) is below 5%, the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) are within acceptable levels, and R2 is consistently above 0.85. The proposed model outperforms other models such as Linear Regression (LR), CNN, LSTM, Random Forest (RF), and Gradient Boosting (GB) for all bedroom configurations. Full article
(This article belongs to the Section G: Energy and Buildings)
Show Figures

Figure 1

Figure 1
<p>Average daylight and sunshine in Burnaby [<a href="#B28-energies-17-06201" class="html-bibr">28</a>] during January to December 2023.</p>
Full article ">Figure 2
<p>BC climate zones based on Heating Degree Days (HDD) [<a href="#B13-energies-17-06201" class="html-bibr">13</a>].</p>
Full article ">Figure 3
<p>SGT components in grid-connected mode.</p>
Full article ">Figure 4
<p>One to four bedroom SGT floor plans.</p>
Full article ">Figure 5
<p>Block diagram of the data collection system.</p>
Full article ">Figure 6
<p>Flowchart of the proposed SGT algorithm.</p>
Full article ">Figure 7
<p>The proposed deep ML model architecture.</p>
Full article ">Figure 8
<p>The Peephole LSTM unit.</p>
Full article ">Figure 9
<p>The data processing flowchart.</p>
Full article ">Figure 10
<p>Monthly electricity consumption (2012–2014) for a new one-story townhouse (pink) and the base townhouse from [<a href="#B25-energies-17-06201" class="html-bibr">25</a>] (blue).</p>
Full article ">Figure 11
<p>Monthly electricity consumption (2012–2014) For one to four Bd SGTs and CSGTs in grid-connected mode.</p>
Full article ">Figure 12
<p>Monthly gas consumption (2012–2014) for one to four bedroom SGTs and CSGTs in grid-connected mode.</p>
Full article ">Figure 13
<p>Monthly total water consumption for one to four bedroom SGTs and CSGTs in grid-connected mode for January–December 2013.</p>
Full article ">Figure 14
<p>Actual versus predicted monthly electricity consumption with seven ML models for a one bedroom CSGT in grid-connected mode for 2012–2014.</p>
Full article ">Figure 15
<p>Actual versus predicted monthly electricity consumption with seven ML models for a two bedroom CSGT in grid-connected mode (2012–2014).</p>
Full article ">Figure 16
<p>Actual versus predicted monthly electricity consumption with seven ML models for a three bedroom CSGT in grid-connected mode (2012–2014).</p>
Full article ">Figure 17
<p>Actual versus predicted monthly electricity consumption with seven ML models for a four bedroom CSGT in grid-connected mode (2012–2014).</p>
Full article ">Figure 18
<p>Hourly one day ahead prediction MAPE and MAE for 3 January 2013.</p>
Full article ">
26 pages, 3798 KiB  
Review
An Overview About Second-Life Battery Utilization for Energy Storage: Key Challenges and Solutions
by Hua Song, Huaizhi Chen, Yanbo Wang and Xiang-E Sun
Energies 2024, 17(23), 6163; https://doi.org/10.3390/en17236163 - 6 Dec 2024
Viewed by 545
Abstract
This article provides a comprehensive overview of the potential challenges and solutions of second-life batteries. First, safety issues of second-life batteries are investigated, which is highly related to the thermal runaway of battery systems. The critical solutions for the thermal runaway problem are [...] Read more.
This article provides a comprehensive overview of the potential challenges and solutions of second-life batteries. First, safety issues of second-life batteries are investigated, which is highly related to the thermal runaway of battery systems. The critical solutions for the thermal runaway problem are discussed, including structural optimization, parameter identification, advanced BMS, and artificial intelligence (AI)-based control strategies. Furthermore, the cell inhomogeneity problem of second-life battery systems is analyzed, where the passive balancing strategy and active balancing strategy are reviewed, respectively. Then, the compatibility issue of second-life batteries is investigated to determine whether electrical dynamic characteristics of a second-life battery can meet the performance requirements for energy storage. In addition, date security and protection methods are reviewed, including digital passport, smart meters and Internet of Things (IoT). The future trends and solutions of key challenges for second-life battery utilization are discussed. Full article
Show Figures

Figure 1

Figure 1
<p>Diagram of the lifetime of an EV battery.</p>
Full article ">Figure 2
<p>The potential application of second-life batteries in future power grids.</p>
Full article ">Figure 3
<p>The phenomena, mechanisms, and control methods for battery safety [<a href="#B9-energies-17-06163" class="html-bibr">9</a>,<a href="#B10-energies-17-06163" class="html-bibr">10</a>,<a href="#B11-energies-17-06163" class="html-bibr">11</a>,<a href="#B12-energies-17-06163" class="html-bibr">12</a>,<a href="#B13-energies-17-06163" class="html-bibr">13</a>,<a href="#B14-energies-17-06163" class="html-bibr">14</a>,<a href="#B15-energies-17-06163" class="html-bibr">15</a>,<a href="#B16-energies-17-06163" class="html-bibr">16</a>].</p>
Full article ">Figure 4
<p>SOC inconsistency of different cells in a battery module.</p>
Full article ">Figure 5
<p>A diagram of battery balancing methods. (<b>a</b>) Active balancing method. (<b>b</b>) Passive balancing method.</p>
Full article ">Figure 6
<p>The potential compatibility problem for second-life battery utilization in energy storage systems.</p>
Full article ">Figure 7
<p>The concept, impact and stakeholders of DP technology.</p>
Full article ">
30 pages, 3547 KiB  
Article
New Forecasting Metrics Evaluated in Prophet, Random Forest, and Long Short-Term Memory Models for Load Forecasting
by Prajowal Manandhar, Hasan Rafiq, Edwin Rodriguez-Ubinas and Themis Palpanas
Energies 2024, 17(23), 6131; https://doi.org/10.3390/en17236131 - 5 Dec 2024
Viewed by 325
Abstract
Data mining is vital for smart grids because it enhances overall grid efficiency, enabling the analysis of large volumes of data, the optimization of energy distribution, the identification of patterns, and demand forecasting. Several performance metrics, such as the MAPE and RMSE, have [...] Read more.
Data mining is vital for smart grids because it enhances overall grid efficiency, enabling the analysis of large volumes of data, the optimization of energy distribution, the identification of patterns, and demand forecasting. Several performance metrics, such as the MAPE and RMSE, have been created to assess these forecasts. This paper presents new performance metrics called Evaluation Metrics for Performance Quantification (EMPQ), designed to evaluate forecasting models in a more comprehensive and detailed manner. These metrics fill the gap left by established metrics by assessing the likelihood of over- and under-forecasting. The proposed metrics quantify forecast bias through maximum and minimum deviation percentages, assessing the proximity of predicted values to actual consumption and differentiating between over- and under-forecasts. The effectiveness of these metrics is demonstrated through a comparative analysis of short-term load forecasting for residential customers in Dubai. This study was based on high-resolution smart meter data, weather data, and voluntary survey data of household characteristics, which permitted the subdivision of the customers into several groups. The new metrics were demonstrated on the Prophet, Random Forest (RF), and Long Short-term Memory (LSTM) models. EMPQ help to determine that the LSTM model exhibited a superior performance with a maximum deviation of approximately 10% for day-ahead and 20% for week-ahead forecasts in the “AC-included” category, outperforming the Prophet model, which had deviation rates of approximately 44% and 42%, respectively. EMPQ also help to determine that the RF excelled over LSTM for the ‘bedroom-number’ subcategory. The findings highlight the value of the proposed metrics in assessing model performance across diverse subcategories. This study demonstrates the value of tailored forecasting models for accurate load prediction and underscores the importance of enhanced performance metrics in informing model selection and supporting energy management strategies. Full article
(This article belongs to the Special Issue Data Mining Approaches for Smart Grids)
Show Figures

Figure 1

Figure 1
<p>Flowchart of the method utilized to process the data used in this study.</p>
Full article ">Figure 2
<p>Traditional evaluation metrics: summary of advantages and limitations.</p>
Full article ">Figure 3
<p>Illustration of the quantification of over-/under-forecasts.</p>
Full article ">Figure 4
<p>Trend and seasonal profiles of ‘AC-included’ and ‘AC-excluded’ category data (note that the vertical scales in ‘AC-included’ and ‘AC-excluded’ are different from each other).</p>
Full article ">Figure 5
<p>Comparison of Prophet, RF, and LSTM algorithms for 1-day-ahead forecasts for AC-included (<b>top</b>) and AC-excluded (<b>bottom</b>) categories.</p>
Full article ">Figure 6
<p>Week-ahead forecasts for the ‘AC-included’ (<b>left</b>) and ‘AC-excluded’ (<b>right</b>) categories using Prophet (<b>top</b>), RF (<b>middle</b>), and LSTM (<b>bottom</b>) algorithms.</p>
Full article ">
20 pages, 399 KiB  
Article
School Start Times for Solar Alignment: Evaluating the Benefits of Schedule Optimisation for Peak and Cost Reduction
by Terhemba Michael-Ahile, Jason Avron Samuels and Marthinus Johannes Booysen
Energies 2024, 17(23), 6112; https://doi.org/10.3390/en17236112 - 4 Dec 2024
Viewed by 392
Abstract
The global push towards sustainable energy usage and the increasing adoption of renewable energy sources, such as solar power, requires innovative approaches to energy management, particularly in energy-intensive sectors such as education. This study proposes a change in school start time from 7 [...] Read more.
The global push towards sustainable energy usage and the increasing adoption of renewable energy sources, such as solar power, requires innovative approaches to energy management, particularly in energy-intensive sectors such as education. This study proposes a change in school start time from 7 a.m. to 9 a.m. to align operational hours with periods of off-peak electricity demand and maximum solar availability. Four scenarios are compared: baseline (current schedule without solar), shifted schedule without solar, baseline with solar, and shifted schedule with solar integration. The analysis reveals that shifting the school’s operational hours alone leads to a peak demand reduction of 40%, mitigating strain on the grid during high-demand periods. Solar integration without schedule has a less pronounced effect on peak demand (26%). The combination of schedule shifting and solar integration delivers the most significant benefits, with the highest cost reductions (28%) and peak demand reductions (60%). This study demonstrates that synchronised solar energy generation and optimised scheduling can enhance energy efficiency and long-term financial savings, offering a practical solution for reducing operational costs and improving sustainability in schools. This study demonstrates how public institutions can contribute to the energy transition by adapting their operational schedules to align with renewable energy availability, rather than relying on conventional fixed schedules. Full article
(This article belongs to the Section G: Energy and Buildings)
Show Figures

Figure 1

Figure 1
<p>An overview of the study’s methodology.</p>
Full article ">Figure 2
<p>Average Daily Plane of Array (POA) Irradiance for an entire year.</p>
Full article ">Figure 3
<p>A chart showing cost and peak demand reductions for different scenarios in the low-demand group.</p>
Full article ">Figure 4
<p>A histogram of the average usage and average daily load curve for the baseline and shifted usage with and without solar integration (low scenario). (<b>a</b>) A histogram of the average daily usage for all four scenarios in the low-demand group. (<b>b</b>) The average daily load curve for all four scenarios in the low-demand group.</p>
Full article ">Figure 5
<p>A chart showing cost and peak demand reductions for different scenarios in the high-demand group.</p>
Full article ">Figure 6
<p>A histogram of the average usage and average daily load curve for the baseline and shifted usage with and without solar integration (low scenario). (<b>a</b>) A histogram of the average daily usage for all four scenarios. (<b>b</b>) The average daily load curve for all four scenarios.</p>
Full article ">
22 pages, 2553 KiB  
Review
Advancements in Indoor Precision Positioning: A Comprehensive Survey of UWB and Wi-Fi RTT Positioning Technologies
by Jiageng Qiao, Fan Yang, Jingbin Liu, Gege Huang, Wei Zhang and Mengxiang Li
Network 2024, 4(4), 545-566; https://doi.org/10.3390/network4040027 - 29 Nov 2024
Viewed by 613
Abstract
High-precision indoor positioning is essential for various applications, such as the Internet of Things, robotics, and smart manufacturing, requiring accuracy better than 1 m. Conventional indoor positioning methods, like Wi-Fi or Bluetooth fingerprinting, typically provide low accuracy within a range of several meters, [...] Read more.
High-precision indoor positioning is essential for various applications, such as the Internet of Things, robotics, and smart manufacturing, requiring accuracy better than 1 m. Conventional indoor positioning methods, like Wi-Fi or Bluetooth fingerprinting, typically provide low accuracy within a range of several meters, while techniques such as laser or visual odometry often require fusion with absolute positioning methods. Ultra-wideband (UWB) and Wi-Fi Round-Trip Time (RTT) are emerging radio positioning technologies supported by industry leaders like Apple and Google, respectively, both capable of achieving high-precision indoor positioning. This paper offers a comprehensive survey of UWB and Wi-Fi positioning, beginning with an overview of UWB and Wi-Fi RTT ranging, followed by an explanation of the fundamental principles of UWB and Wi-Fi RTT-based geometric positioning. Additionally, it compares the strengths and limitations of UWB and Wi-Fi RTT technologies and reviews advanced studies that address practical challenges in UWB and Wi-Fi RTT positioning, such as accuracy, reliability, continuity, and base station coordinate calibration issues. These challenges are primarily addressed through a multi-sensor fusion approach that integrates relative and absolute positioning. Finally, this paper highlights future directions for the development of UWB- and Wi-Fi RTT-based indoor positioning technologies. Full article
Show Figures

Figure 1

Figure 1
<p>An illustration of multipath resolution of (<b>a</b>) Wi-Fi signal and (<b>b</b>) UWB signal [<a href="#B6-network-04-00027" class="html-bibr">6</a>].</p>
Full article ">Figure 2
<p>The main structure pyramid of this article.</p>
Full article ">Figure 3
<p>Illustration of TOF positioning principle. (<b>a</b>) TOF-based UWB ranging protocols and (<b>b</b>) TOF-based Wi-Fi RTT ranging protocols.</p>
Full article ">Figure 4
<p>LOS, NLOS, and multipath components in an indoor positioning context.</p>
Full article ">Figure 5
<p>Illustration of (<b>a</b>) TOA and (<b>b</b>) TDOA positioning principles.</p>
Full article ">Figure 6
<p>Illustration of simultaneous autonomous positioning and estimating the unknown coordinates of base stations [<a href="#B96-network-04-00027" class="html-bibr">96</a>].</p>
Full article ">
18 pages, 4970 KiB  
Article
Efficient Simulator for P2P Energy Trading: Customizable Bid Preferences for Trading Agents
by Yasuhiro Takeda, Yosuke Suzuki, Kota Fukamachi, Yuji Yamada and Kenji Tanaka
Energies 2024, 17(23), 5945; https://doi.org/10.3390/en17235945 - 26 Nov 2024
Viewed by 468
Abstract
Given the accelerating global movement towards decarbonization, the importance of promoting renewable energy (RE) adoption and ensuring efficient transactions in energy markets is increasing worldwide. However, renewable energy sources, including photovoltaic (PV) systems, are subject to output fluctuations due to weather conditions, requiring [...] Read more.
Given the accelerating global movement towards decarbonization, the importance of promoting renewable energy (RE) adoption and ensuring efficient transactions in energy markets is increasing worldwide. However, renewable energy sources, including photovoltaic (PV) systems, are subject to output fluctuations due to weather conditions, requiring large-scale backup power to balance supply and demand. This makes trading electricity from large-scale PV systems connected to the existing grid challenging. To address this, peer-to-peer (P2P) energy markets where individual prosumers can trade excess power within their local communities have been garnering attention. This study introduces a simulator for P2P energy trading, designed to account for the diverse behaviors and objectives of participants within a market mechanism. The simulator incorporates two risk aversion parameters: one related to transaction timing, expressed through order prices, and another related to forecast errors, managed by adjusting trade volumes. This allows participants to customize their trading strategies, resulting in more realistic analyses of trading outcomes. To explore the effects of these risk aversion settings, we conduct a case study with 120 participants, including both consumers and prosumers, using real data from household smart meters collected on sunny and cloudy days. Our analysis shows that participants with higher aversion to transaction timing tend to settle trades earlier, often resulting in unnecessary transactions due to forecast inaccuracies. Furthermore, trading outcomes are significantly influenced by weather conditions: sunny days typically benefit buyers through lower settlement prices, while cloudy days favor sellers who execute trades closer to their actual needs. These findings demonstrate the trade-off between early execution and forecast error losses, emphasizing the simulator’s ability to analyze trading outcomes while accounting for participant risk aversion preferences. Full article
(This article belongs to the Section C: Energy Economics and Policy)
Show Figures

Figure 1

Figure 1
<p>Relationship between P2P energy trading agents, a Grid Agent, and the P2P energy trading market.</p>
Full article ">Figure 2
<p>Trading products and example of multiple markets using tags. The diagram illustrates the structure of the P2P energy trading market in the simulator, where electricity is traded in 30 min intervals, with 48 products traded each day. Each product corresponds to a distinct time slot, starting at either 00 or 30 min (e.g., Product 1 at 00:00 and Product 2 at 00:30). In this example, two types of electricity, identified by Tag1 and Tag2, are traded separately within the same time slot, each managed through its own order book. The order books for Tag1 and Tag2 independently display buy and sell orders organized by price and quantity, with buy orders shown in red and sell orders in green, illustrating the separate handling of different electricity types. The matching of buy and sell orders is conducted using the CDA mechanism, where trades are matched based on price and time priority, promoting efficient and competitive trading. As each trading period advances, the available products shift forward to the next time slot, allowing participants to continuously bid on upcoming delivery periods. Each product may also have a suitable bidding window specific to its tag and time slot, reflecting variations in supply, demand, and market conditions.</p>
Full article ">Figure 3
<p>Changes in buy price for each parameter <math display="inline"><semantics> <msub> <mi>C</mi> <mrow> <mi>b</mi> <mi>u</mi> <mi>y</mi> </mrow> </msub> </semantics></math>.</p>
Full article ">Figure 4
<p>Changes in sell price for each parameter <math display="inline"><semantics> <msub> <mi>C</mi> <mrow> <mi>s</mi> <mi>e</mi> <mi>l</mi> <mi>l</mi> </mrow> </msub> </semantics></math>.</p>
Full article ">Figure 5
<p>Relationship between prediction error and settlement timing.</p>
Full article ">Figure 6
<p>P2P energy trading simulator process flow.</p>
Full article ">Figure 7
<p>Net demand of all participants.</p>
Full article ">Figure 8
<p>Selected consumer’s net demand.</p>
Full article ">Figure 9
<p>Selected prosumer’s net demand.</p>
Full article ">Figure 10
<p>Risk aversion settings matrix: mapping low to high levels of prediction error and settlement timing.</p>
Full article ">Figure 11
<p>Box plot of trading results.</p>
Full article ">Figure 12
<p>Selected prosumer’s energy traded amounts by time gap between actual and settled trading timings. (<b>a</b>) Selling Amount on the Selected Sunny Day. (<b>b</b>) Buying Amount on the Selected Sunny Day.</p>
Full article ">Figure 13
<p>Relationship between total cost and error power in energy trading: results for the selected prosumer.</p>
Full article ">Figure 14
<p>Relationship between total cost and error power in energy trading: results for the selected consumer.</p>
Full article ">
20 pages, 1957 KiB  
Article
Predictive Analytics for Energy Efficiency: Leveraging Machine Learning to Optimize Household Energy Consumption
by Piotr Powroźnik and Paweł Szcześniak
Energies 2024, 17(23), 5866; https://doi.org/10.3390/en17235866 - 22 Nov 2024
Viewed by 413
Abstract
This paper presents a novel machine learning framework useful for optimizing energy consumption in households. Home appliances have a great potential to optimize electricity consumption by mitigating peaks in the grid load or peaks in renewable energy generation. However, such functionality of home [...] Read more.
This paper presents a novel machine learning framework useful for optimizing energy consumption in households. Home appliances have a great potential to optimize electricity consumption by mitigating peaks in the grid load or peaks in renewable energy generation. However, such functionality of home appliances requires their users to change their behavior regarding energy consumption. One of the criteria that could encourage electricity users to change their behavior is the cost of energy. The introduction of dynamic energy prices can significantly increase energy costs for unsuspecting consumers. In order to be able to make the right decisions about the process of electricity use in households, an algorithm based on machine learning is proposed. The presented proposal for optimizing electricity consumption takes into account dynamic changes in energy prices, energy production from renewable energy sources, and home appliances that can participate in the energy optimization process. The proposed model uses data from smart meters and dynamic price information to generate personalized recommendations tailored to individual households. The algorithm, based on machine learning and historical household behavior data, calculates a metric to determine whether to send a notification (message) to the user. This notification may suggest increasing or decreasing energy consumption at a specific time, or may inform the user about potential cost fluctuations in the upcoming hours. This will allow energy users to use energy more consciously or to set priorities in home energy management systems (HEMS). This is a different approach than in previous publications, where the main goal of optimizing energy consumption was to optimize the operation of the power system while taking into account the profits of energy suppliers. The proposed algorithms can be implemented either in HEMS or smart energy meters. In this work, simulations of the application of machine learning with different characteristics were carried out in the MATLAB program. An analysis of machine learning algorithms for different input data and amounts of data and the characteristic features of models is presented. Full article
(This article belongs to the Special Issue Novel Energy Management Approaches in Microgrid Systems)
Show Figures

Figure 1

Figure 1
<p>A system for collecting real-time energy consumption data from ECs within an HEMS.</p>
Full article ">Figure 2
<p>Power consumption profile <math display="inline"><semantics> <msub> <mi>P</mi> <mi>SA</mi> </msub> </semantics></math> of thirteen ECs.</p>
Full article ">Figure 3
<p>Temporal distribution of <tt>cost</tt>.</p>
Full article ">Figure 4
<p>Comparison of <tt>message</tt> classification accuracy for one day: without a test set of data reserved for testing.</p>
Full article ">Figure 5
<p>Validation confusion matrix for <math display="inline"><semantics> <mrow> <mi>m</mi> <msub> <mi>l</mi> <mn>3</mn> </msub> </mrow> </semantics></math> for a single day without a reserved test set.</p>
Full article ">Figure 6
<p>Validation confusion matrix for <math display="inline"><semantics> <mrow> <mi>m</mi> <msub> <mi>l</mi> <mn>3</mn> </msub> </mrow> </semantics></math> using a single day of data with a 10% test set.</p>
Full article ">Figure 7
<p>Simulation results for <math display="inline"><semantics> <mrow> <mi>m</mi> <mi>l</mi> </mrow> </semantics></math> models, showing accuracy for different scenarios (<math display="inline"><semantics> <mrow> <mi>s</mi> <mi>c</mi> </mrow> </semantics></math>).</p>
Full article ">Figure 8
<p>Dependence of prediction speed on model architecture in scenario <math display="inline"><semantics> <mrow> <mi>s</mi> <msub> <mi>c</mi> <mn>6</mn> </msub> </mrow> </semantics></math>.</p>
Full article ">Figure 9
<p>Relationship between model size and computational performance.</p>
Full article ">Figure 10
<p>Time required to train individual machine learning models.</p>
Full article ">Figure 11
<p>Weighted precision of the classifier for <math display="inline"><semantics> <mrow> <mi>s</mi> <msub> <mi>c</mi> <mn>6</mn> </msub> </mrow> </semantics></math>.</p>
Full article ">Figure 12
<p>Weighted recall of the classifier for <math display="inline"><semantics> <mrow> <mi>s</mi> <msub> <mi>c</mi> <mn>6</mn> </msub> </mrow> </semantics></math>.</p>
Full article ">Figure 13
<p>Weighted F1-score of the classifier for <math display="inline"><semantics> <mrow> <mi>s</mi> <msub> <mi>c</mi> <mn>6</mn> </msub> </mrow> </semantics></math>.</p>
Full article ">Figure 14
<p>Household energy consumption schedule for a case study.</p>
Full article ">Figure 15
<p>Prediction of <tt>message</tt> indications for a user based on the <math display="inline"><semantics> <mrow> <mi>m</mi> <msub> <mi>l</mi> <mn>1</mn> </msub> </mrow> </semantics></math>.</p>
Full article ">
19 pages, 7807 KiB  
Article
Harnessing Risks with Data: A Leakage Assessment Framework for WDN Using Multi-Attention Mechanisms and Conditional GAN-Based Data Balancing
by Wenhong Wu, Jiahao Zhang, Yunkai Kang, Zhengju Tang, Xinyu Pan and Ning Liu
Water 2024, 16(22), 3329; https://doi.org/10.3390/w16223329 - 19 Nov 2024
Viewed by 488
Abstract
Assessing leakage risks in water distribution networks (WDNs) and implementing preventive monitoring for high-risk pipelines has become a widely accepted approach for leakage control. However, existing methods face significant data barriers between Geographic Information System (GIS) and leakage prediction systems. These barriers hinder [...] Read more.
Assessing leakage risks in water distribution networks (WDNs) and implementing preventive monitoring for high-risk pipelines has become a widely accepted approach for leakage control. However, existing methods face significant data barriers between Geographic Information System (GIS) and leakage prediction systems. These barriers hinder traditional pipeline risk assessment methods, particularly when addressing challenges such as data imbalance, poor model interpretability, and lack of intuitive prediction results. To overcome these limitations, this study proposes a leakage assessment framework for water distribution networks based on multiple attention mechanisms and a generative model-based data balancing method. Extensive comparative experiments were conducted using water distribution network data from B2 and B3 District Metered Areas in Zhengzhou. The results show that the proposed model, optimized with a balanced data method, achieved a 40.76% improvement in the recall rate for leakage segment assessments, outperforming the second-best model using the same strategy by 1.7%. Furthermore, the strategy effectively enhanced the performance of all models, further proving that incorporating more valid data contributes to improved assessment results. This study comprehensively demonstrates the application of data-driven models in the field of “smart water management”, providing practical guidance and reference cases for advancing the development of intelligent water infrastructure. Full article
Show Figures

Figure 1

Figure 1
<p>Framework.</p>
Full article ">Figure 2
<p>Data Enhancement Program.</p>
Full article ">Figure 3
<p>Pipeline Risk Prediction Modeling Framework.</p>
Full article ">Figure 4
<p>Enhanced sample balance results for leakage sample data: (<b>a</b>) Original Data; (<b>b</b>) SMOTE Data Enhancement Method; (<b>c</b>) Conditional GAN Data Augmentation Method. (The horizontal and vertical axes represent the two-dimensional vector values obtained by dimensionality reduction of the high-dimensional representation of the samples, serving only as markers).</p>
Full article ">Figure 5
<p>SHAP analysis results: Overall Ranking Analysis of SHAP Risk Factors.</p>
Full article ">Figure 6
<p>SHAP analysis results: SHAP Single Case Analysis-Case 10850.</p>
Full article ">Figure 7
<p>SHAP analysis results: SHAP Single Example Analysis-Case 486.</p>
Full article ">Figure 8
<p>Pipeline age leakage rate analysis: (<b>a</b>) Age distribution of pipelines assessed as a level of risk, (<b>b</b>) Leakage ratios for pipelines of different ages.</p>
Full article ">Figure 9
<p>Visualization Platform: (<b>a</b>) Leakage Point; (<b>b</b>) Pipeline Location; (<b>c</b>) B2, B3 DMA location; (<b>d</b>) Total Pipeline leakage risk status statistics; (<b>e</b>) Classified pipeline leakage risk status statistics and positioning; (<b>f</b>) Individual pipeline leakage risk status information; (<b>g</b>) Region-specific leakage risk status statistics.</p>
Full article ">
13 pages, 6430 KiB  
Proceeding Paper
Detection of Non-Technical Losses in Special Customers with Telemetering, Based on Artificial Intelligence
by José Luis Llagua Arévalo and Patricio Antonio Pesántez Sarmiento
Eng. Proc. 2024, 77(1), 29; https://doi.org/10.3390/engproc2024077029 - 18 Nov 2024
Viewed by 240
Abstract
The Ecuadorian electricity sector, until April 2024, presented losses of 15.64% (6.6% technical and 9.04% non-technical), so it is important to detect the areas that potentially sub-register energy in order to reduce Non-Technical Losses (NTLs). The “Empresa Eléctrica de Ambato Sociedad Anónima” (EEASA), [...] Read more.
The Ecuadorian electricity sector, until April 2024, presented losses of 15.64% (6.6% technical and 9.04% non-technical), so it is important to detect the areas that potentially sub-register energy in order to reduce Non-Technical Losses (NTLs). The “Empresa Eléctrica de Ambato Sociedad Anónima” (EEASA), as a distribution company, has, to reduce NTLs, incorporated many smart meters in special clients, generating a large amount of data that are stored. This historical information is analyzed to detect anomalous consumption that is not easily recognized and is a significant part of the NTLs. The use of machine learning with appropriate clustering techniques and deep learning neural networks work together to detect abnormal curves that record lower readings than the real energy consumption. The developed methodology uses three k-means validation indices to classify daily energy curves based on the days of the week and holidays that present similar behaviors in terms of energy consumption. The developed algorithm groups similar consumption patterns as input data sets for learning, testing, and validating the densely connected classification neural network, allowing for the identification of daily curves described by customers. The results obtained from the system detected customers who sub-register energy. It is worth mentioning that this methodology is replicable for distribution companies that store historical consumption data with Advanced Measurement Infrastructure (AMI) systems. Full article
(This article belongs to the Proceedings of The XXXII Conference on Electrical and Electronic Engineering)
Show Figures

Figure 1

Figure 1
<p>Methodology flowchart.</p>
Full article ">Figure 2
<p>Variabilit.</p>
Full article ">Figure 3
<p>Demand.</p>
Full article ">Figure 4
<p>Grouping using the Soft-DTW k-means index for a k = 5, represented the centroid curves in red color.</p>
Full article ">Figure 5
<p>Grouping assigned values.</p>
Full article ">Figure 6
<p>Normal and fraudulent consumption curves with percentage decrease. (<b>a</b>) Type 1 with 36% of customer 6 in zone 2. (<b>b</b>) Type 2 with 56% of customer 4 in zone 1 and (<b>c</b>) Type 3 with 82% of customer 6 in zone 7.</p>
Full article ">Figure 7
<p>Model network design for the holiday group.</p>
Full article ">Figure 8
<p>KNIME—Python link and deep learning libraries.</p>
Full article ">Figure 9
<p>Completed neural network in the working environment.</p>
Full article ">Figure 10
<p>Accuracy curves of the neural network.</p>
Full article ">Figure 11
<p>Losses curves of the neural network.</p>
Full article ">Figure 12
<p>Weekend neural network results.</p>
Full article ">Figure 13
<p>Results of the neural network from Monday to Friday.</p>
Full article ">
31 pages, 11682 KiB  
Article
Comparative Study of Time Series Analysis Algorithms Suitable for Short-Term Forecasting in Implementing Demand Response Based on AMI
by Myung-Joo Park and Hyo-Sik Yang
Sensors 2024, 24(22), 7205; https://doi.org/10.3390/s24227205 - 11 Nov 2024
Viewed by 592
Abstract
This paper compares four time series forecasting algorithms—ARIMA, SARIMA, LSTM, and SVM—suitable for short-term load forecasting using Advanced Metering Infrastructure (AMI) data. The primary focus is on evaluating the applicability and performance of these forecasting models in predicting electricity consumption patterns, which is [...] Read more.
This paper compares four time series forecasting algorithms—ARIMA, SARIMA, LSTM, and SVM—suitable for short-term load forecasting using Advanced Metering Infrastructure (AMI) data. The primary focus is on evaluating the applicability and performance of these forecasting models in predicting electricity consumption patterns, which is a critical component for implementing effective demand response (DR) strategies. The study provides a comprehensive analysis of the predictive accuracy, computational efficiency, and scalability of each algorithm using a dataset of real-time electricity consumption collected from AMI systems over a designated period. Through extensive experiments, we demonstrate that each algorithm has distinct strengths and weaknesses depending on the characteristics of the dataset. Specifically, SVM exhibited superior performance in handling nonlinear patterns and high volatility, while SARIMA effectively captured seasonal trends. LSTM showed potential in modeling complex temporal dependencies but was sensitive to hyperparameter settings and required a substantial amount of training data. This research offers practical guidelines for selecting the optimal forecasting model based on data characteristics and application needs, contributing to the development of more efficient and dynamic energy management strategies. The findings highlight the importance of integrating advanced forecasting techniques into smart grid systems to enhance the reliability and responsiveness of DR programs. This study lays a solid foundation for future research on integrating these forecasting models into real-world AMI applications to support effective demand response and grid stability. Full article
(This article belongs to the Special Issue IoT and Big Data Analytics for Smart Cities)
Show Figures

Figure 1

Figure 1
<p>Possibility of overfitting due to amount of learning data.</p>
Full article ">Figure 2
<p>Neural network without dropout.</p>
Full article ">Figure 3
<p>Neural network with dropout.</p>
Full article ">Figure 4
<p>Distribution and trend line graphs of sampled data by consumer number (CNSMR_NO).</p>
Full article ">Figure 5
<p>Differences in SGPowerUsage values between predictions and actual data in DJ0200309001501.</p>
Full article ">Figure 6
<p>Differences in SGPowerUsage values between predictions and actual data in DJ0800133001204.</p>
Full article ">Figure 7
<p>Differences in SGPowerUsage values between predictions and actual data in DJ1200215000404.</p>
Full article ">Figure 8
<p>Differences in SGPowerUsage values between predictions and actual data in CB0100106000505.</p>
Full article ">Figure 9
<p>Differences in SGPowerUsage values between predictions and actual data in CN0100107001801.</p>
Full article ">Figure 10
<p>Differences in SGPowerUsage values between predictions and actual data in CN0200311001801.</p>
Full article ">Figure 11
<p>Differences in SGPowerUsage values between predictions and actual data in CN1100106000103.</p>
Full article ">Figure 12
<p>Differences in SGPowerUsage values between predictions and actual data in CN1600102001004.</p>
Full article ">Figure 13
<p>Differences in SGPowerUsage values between predictions and actual data in CN0500311000403.</p>
Full article ">Figure 14
<p>Differences in SGPowerUsage values between predictions and actual data in CN0700109000102.</p>
Full article ">Figure 15
<p>Differences in SGPowerUsage values between predictions and actual data.</p>
Full article ">
18 pages, 3183 KiB  
Article
Determine the Profiles of Power Consumption in Commercial Buildings in a Very Hot Humid Climate Using a Temporary Series
by E. Catalina Vallejo-Coral, Ricardo Garzón, Miguel Darío Ortega López, Javier Martínez-Gómez and Marcelo Moya
Sustainability 2024, 16(22), 9770; https://doi.org/10.3390/su16229770 - 8 Nov 2024
Viewed by 738
Abstract
With the growth of the nations, the commercial and public services sectors have recently seen an increase in their electricity usage. This demonstrates how crucial it is to understand a building’s behavior in order to lower its usage. This requires on-site data collection [...] Read more.
With the growth of the nations, the commercial and public services sectors have recently seen an increase in their electricity usage. This demonstrates how crucial it is to understand a building’s behavior in order to lower its usage. This requires on-site data collection by qualified professionals and specialized equipment, which represents high costs. However, multiple studies have demonstrated that it is possible to find electricity-saving strategies from the study of electricity usage, recorded in an hourly period or less, captured by smart meters. In this context, the present study applies a methodology to determine useful information on the operation and characteristics of public buildings on the Ecuadorian coast based on the data gathered over a period of five consecutive months from smart meters. The methodology consists of four steps: (1) data cleaning and filling, (2) time-series decomposition, (3) the generation of consumption profile and (4) the identification of the temperature influence. According to the results, the pre-cooling of spaces accounts for 5% of all electricity used in the commercial buildings, while prolonged shutdown uses 10%. Approximately USD 1100 per month would be spent on the main building and USD 78 on the agency as a result. Full article
Show Figures

Figure 1

Figure 1
<p>Building images of E1 (<b>a</b>), E2 (<b>b</b>) and E3 (<b>c</b>).</p>
Full article ">Figure 2
<p>Time-series decomposition of ambient temperature.</p>
Full article ">Figure 3
<p>E1 time-series decomposition. Measurements are taken every 15 min.</p>
Full article ">Figure 4
<p>E2 time-series decomposition. Measurements are taken every 15 min.</p>
Full article ">Figure 5
<p>E3 time-series decomposition. Measurements are taken every 15 min.</p>
Full article ">Figure 6
<p>Daily energy consumption profiles in E1 (<b>a</b>), E2 (<b>b</b>) and E3 (<b>c</b>) for days of the week. All data from the analyzed period have been used.</p>
Full article ">Figure 7
<p>Pearson correlation between ambient temperature and power consumption for E1 (<b>a</b>), E2 (<b>b</b>) and E3 (<b>c</b>).</p>
Full article ">
27 pages, 15476 KiB  
Article
Explainable AI-Based Ensemble Clustering for Load Profiling and Demand Response
by Elissaios Sarmas, Afroditi Fragkiadaki and Vangelis Marinakis
Energies 2024, 17(22), 5559; https://doi.org/10.3390/en17225559 - 7 Nov 2024
Viewed by 567
Abstract
Smart meter data provide an in-depth perspective on household energy usage. This research leverages on such data to enhance demand response (DR) programs through a novel application of ensemble clustering. Despite its promising capabilities, our literature review identified a notable under-utilization of ensemble [...] Read more.
Smart meter data provide an in-depth perspective on household energy usage. This research leverages on such data to enhance demand response (DR) programs through a novel application of ensemble clustering. Despite its promising capabilities, our literature review identified a notable under-utilization of ensemble clustering in this domain. To address this shortcoming, we applied an advanced ensemble clustering method and compared its performance with traditional algorithms, namely, K-Means++, fuzzy K-Means, Hierarchical Agglomerative Clustering, Spectral Clustering, Gaussian Mixture Models (GMMs), BIRCH, and Self-Organizing Maps (SOMs), across a dataset of 5567 households for a range of cluster counts from three to nine. The performance of these algorithms was assessed using an extensive set of evaluation metrics, including the Silhouette Score, the Davies–Bouldin Score, the Calinski–Harabasz Score, and the Dunn Index. Notably, while ensemble clustering often ranked among the top performers, it did not consistently surpass all individual algorithms, indicating its potential for further optimization. Unlike approaches that seek the algorithmically optimal number of clusters, our method proposes a practical six-cluster solution designed to meet the operational needs of utility providers. For this case, the best performing algorithm according to the evaluation metrics was ensemble clustering. This study is further enhanced by integrating Explainable AI (xAI) techniques, which improve the interpretability and transparency of our clustering results. Full article
(This article belongs to the Special Issue Advances in Energy Market and Distributed Generation)
Show Figures

Figure 1

Figure 1
<p>Detailed overview of methodology.</p>
Full article ">Figure 2
<p>Snapshot of the dataset.</p>
Full article ">Figure 3
<p>Silhouette scores for cluster ranges from 3 to 9.</p>
Full article ">Figure 4
<p>Davies–Bouldin scores for cluster ranges from 3 to 9.</p>
Full article ">Figure 5
<p>Calinski–Harabasz scores for cluster ranges from 3 to 9.</p>
Full article ">Figure 6
<p>Dunn index for cluster ranges from 3 to 9.</p>
Full article ">Figure 7
<p>Non-normalized load profiles of all households in each cluster.</p>
Full article ">Figure 8
<p>Normalized load profiles of all households in each cluster.</p>
Full article ">Figure 9
<p>Weekend vs. weekday mean cluster profiles.</p>
Full article ">Figure 10
<p>SHAP values analysis for cluster 1.</p>
Full article ">Figure 11
<p>SHAP values analysis for cluster 2.</p>
Full article ">
24 pages, 9406 KiB  
Article
Lightweight Digit Recognition in Smart Metering System Using Narrowband Internet of Things and Federated Learning
by Vladimir Nikić, Dušan Bortnik, Milan Lukić, Dejan Vukobratović and Ivan Mezei
Future Internet 2024, 16(11), 402; https://doi.org/10.3390/fi16110402 - 31 Oct 2024
Viewed by 2223
Abstract
Replacing mechanical utility meters with digital ones is crucial due to the numerous benefits they offer, including increased time resolution in measuring consumption, remote monitoring capabilities for operational efficiency, real-time data for informed decision-making, support for time-of-use billing, and integration with smart grids, [...] Read more.
Replacing mechanical utility meters with digital ones is crucial due to the numerous benefits they offer, including increased time resolution in measuring consumption, remote monitoring capabilities for operational efficiency, real-time data for informed decision-making, support for time-of-use billing, and integration with smart grids, leading to enhanced customer service, reduced energy waste, and progress towards environmental sustainability goals. However, the cost associated with replacing mechanical meters with their digital counterparts is a key factor contributing to the relatively slow roll-out of such devices. In this paper, we present a low-cost and power-efficient solution for retrofitting the existing metering infrastructure, based on state-of-the-art communication and artificial intelligence technologies. The edge device we developed contains a camera for capturing images of a dial meter, a 32-bit microcontroller capable of running the digit recognition algorithm, and an NB-IoT module with (E)GPRS fallback, which enables nearly ubiquitous connectivity even in difficult radio conditions. Our digit recognition methodology, based on the on-device training and inference, augmented with federated learning, achieves a high level of accuracy (97.01%) while minimizing the energy consumption and associated communication overhead (87 μWh per day on average). Full article
Show Figures

Figure 1

Figure 1
<p>Proposed SM architecture used for old TM retrofitting.</p>
Full article ">Figure 2
<p>The system architecture consists of a collection of deployed SMs that communicate via MNO with the cloud.</p>
Full article ">Figure 3
<p>Distribution of ML models on edge devices, which provides the basis for FL.</p>
Full article ">Figure 4
<p>Design and components of edge device is depicted on the left, whereas right images display fabricated devices.</p>
Full article ">Figure 5
<p>View of metering device through camera lens of the edge device used for creation of datasets.</p>
Full article ">Figure 6
<p>Conversion to B/W image format using a fixed threshold: TH = 128 (<b>left</b>), TH = 192 (<b>right</b>).</p>
Full article ">Figure 7
<p>Initial image taken using camera on edge device, intermediate image after B/W conversion and the final image without artifacts.</p>
Full article ">Figure 8
<p>Scheme used for detecting differences between two digit images.</p>
Full article ">Figure 9
<p>Proposed CNN architecture used for digit recognition.</p>
Full article ">Figure 10
<p>The first test case comprises two distinct scenarios representing different training methodologies, one incorporating federated learning (FL) and the other without its use. (<b>a</b>) Scenario 1, which does not use FL. (<b>b</b>) Scenario 2, which utilizes averaging methodology for FL.</p>
Full article ">Figure 11
<p>Second test case displaying training scheme where second batch of devices is trained based on results of training on first batch.</p>
Full article ">Figure 12
<p>Power consumption profile of image capture + preprocessing + inference.</p>
Full article ">Figure 13
<p>Power consumption profile of data packet transmission via NB-IoT.</p>
Full article ">
21 pages, 3037 KiB  
Article
Bi-Directional Charging with V2L Integration for Optimal Energy Management in Electric Vehicles
by Balakumar Muniandi, Siyi Wan and Mohammad El-Yabroudi
Electronics 2024, 13(21), 4221; https://doi.org/10.3390/electronics13214221 - 28 Oct 2024
Viewed by 907
Abstract
Electric vehicles (EVs) are becoming increasingly popular as an efficient transportation solution but they also present unique challenges for energy management. Bi-directional charging (BDC) is a solution that allows EVs to not only consume energy from the grid but also supply energy back [...] Read more.
Electric vehicles (EVs) are becoming increasingly popular as an efficient transportation solution but they also present unique challenges for energy management. Bi-directional charging (BDC) is a solution that allows EVs to not only consume energy from the grid but also supply energy back to the grid. This facilitates vehicle-to-load (V2L) integration, where EVs can act as mobile power sources for homes, buildings, and the grid. V2L enables better energy management by utilizing EVs as a flexible resource to balance grid demand and supply in the proposed system. This is achieved through intelligent coordination between the EVs, charging stations, and the grid, using smart meters and communication networks. Integration of BDC and V2L also enables EVs to provide backup power during grid outages, reduce the need for costly grid infrastructure, and support renewable energy integration. BDC with V2L integration is a promising approach for optimal energy management in EVs and can play a significant role in the future of sustainable transportation and energy systems. The proposed model reached 95.13% charging efficiency, 95.03% energy management, 95.69% power rating, 96.28% voltage support and 87.99% temperature management. Full article
(This article belongs to the Special Issue Power Electronics and Renewable Energy System)
Show Figures

Figure 1

Figure 1
<p>Construction of proposed model.</p>
Full article ">Figure 2
<p>Bi-directional charging and grid control.</p>
Full article ">Figure 3
<p>Load integration.</p>
Full article ">Figure 4
<p>Comparison of charging efficiency.</p>
Full article ">Figure 5
<p>Comparison of energy management.</p>
Full article ">Figure 6
<p>Comparison of power rating.</p>
Full article ">Figure 7
<p>Comparison of voltage support.</p>
Full article ">Figure 8
<p>Comparison of temperature management.</p>
Full article ">
17 pages, 24201 KiB  
Article
An Echo State Network-Based Light Framework for Online Anomaly Detection: An Approach to Using AI at the Edge
by Andrea Bonci, Renat Kermenov, Lorenzo Longarini, Sauro Longhi, Geremia Pompei, Mariorosario Prist and Carlo Verdini
Machines 2024, 12(10), 743; https://doi.org/10.3390/machines12100743 - 21 Oct 2024
Viewed by 729
Abstract
Production efficiency is used to determine the best conditions for manufacturing goods at the lowest possible unit cost. When achieved, production efficiency leads to increased revenues for the manufacturer, enhanced employee safety, and a satisfied customer base. Production efficiency not only measures the [...] Read more.
Production efficiency is used to determine the best conditions for manufacturing goods at the lowest possible unit cost. When achieved, production efficiency leads to increased revenues for the manufacturer, enhanced employee safety, and a satisfied customer base. Production efficiency not only measures the amount of resources that are needed for production but also considers the productivity levels and the state of the production lines. In this context, online anomaly detection (AD) is an important tool for maintaining the reliability of the production ecosystem. With advancements in artificial intelligence and the growing significance of identifying and mitigating anomalies across different fields, approaches based on artificial neural networks facilitate the recognition of intricate types of anomalies by taking into account both temporal and contextual attributes. In this paper, a lightweight framework based on the Echo State Network (ESN) model running at the edge is introduced for online AD. Compared to other AD methods, such as Long Short-Term Memory (LSTM), it achieves superior precision, accuracy, and recall metrics while reducing training time, CO2 emissions, and the need for high computational resources. The preliminary evaluation of the proposed solution was conducted using a low-resource computing device at the edge of the real production machine through an Industrial Internet of Things (IIoT) smart meter module. The machine used to test the proposed solution was provided by the Italian company SIFIM Srl, which manufactures filter mats for industrial kitchens. Experimental results demonstrate the feasibility of developing an AD method that achieves high accuracy, with the ESN-based framework reaching 85% compared to 80.88% for the LSTM-based model. Furthermore, this method requires minimal hardware resources, with a training time of 9.5 s compared to 2.100 s for the other model. Full article
Show Figures

Figure 1

Figure 1
<p>ESN diagram.</p>
Full article ">Figure 2
<p>LSTM diagram.</p>
Full article ">Figure 3
<p>Architecture for the computation of the standard deviation of the error in the training set.</p>
Full article ">Figure 4
<p>Framework architecture for the inference phase.</p>
Full article ">Figure 5
<p>Architecture diagram.</p>
Full article ">Figure 6
<p>Production machine and production layout.</p>
Full article ">Figure 7
<p>Cloud dashboard—CO<sub>2</sub> production.</p>
Full article ">Figure 8
<p>Standard deviation of the error of both the ESN and LSTM model-based approaches.</p>
Full article ">Figure 9
<p>Comparison between real time series and time-series prediction (ESN-based model on the left and LSTM-based model on the right).</p>
Full article ">Figure 10
<p>Accuracy for each epoch (LSTM-based and ESN-based methods).</p>
Full article ">
Back to TopTop