Energy Forecasting Research Articles

Optimizing deep neural network architectures for renewable energy forecasting

Optimizing deep neural network architectures for renewable energy forecasting

A Comprehensive Multi-Objective Energy Management Approach for Wearable Devices with Dynamic Energy Demands

Recent advancements in low-power electronics and machine-learning techniques have paved the way for innovative wearable Internet of Things (IoT) devices. However, these devices suffer from limited battery capacity and computational power. Hence, energy harvesting from ambient sources has emerged as a promising solution for powering low-energy wearables. Optimal management of the harvested energy is crucial for achieving energy-neutral operation and eliminating the need for frequent recharging. This task is challenging due to the dynamic nature of harvested energy and battery energy constraints. To tackle this challenge, we propose tinyMAN, a reinforcement learning-based energy management framework for resource-constrained wearable IoT devices. tinyMAN maximizes the target device utilization under battery energy constraints without relying on the harvested energy forecast, making it a prediction-free approach. It achieves up to 17% higher utility while reducing battery constraint violations by 80% compared to prior work. We also introduce tinyMAN-MO, a multi-objective extension of tinyMan for applications with time-varying energy demands. It learns the tradeoff between meeting the application’s energy demand and maintaining the battery energy level. We deployed our framework on a wearable device prototype using TensorFlow Lite for Micro, leveraging its small (less than 120 KB) memory footprint. Evaluations show that tinyMAN-MO operates within 10% of the Pareto-optimal solutions with only 1.98 ms execution time and 23.17 μJ energy consumption overhead.

Multi-Criteria Optimization of the Paper Production Process Using Numerical Taxonomy Methods: A Necessary Condition for Predicting Heat and Electricity Output in a Combined Heat and Power (CHP) System

The subject of this study is the optimization of the paper production process in one of Poland’s leading paper mills. In addition to its primary objective of paper production, the company generates heat and electricity for internal consumption and external clients, including the local municipality. Surplus energy may be sold on the power exchange; however, this requires forecasting the quantity of energy to be sold 24 h in advance, which introduces an element of uncertainty. Production stoppages, often caused by random events such as paper breakage, force a power decrease in the CHP system, further complicating energy forecasting. To minimize the occurrence of such events, numerical taxonomy methods were employed to determine the optimal screen speed (Vs) and winding speed (Vn) for two paper machines, based on the type and weight of the paper produced. This analysis utilized detailed daily data collected by the company over the period 2015–2020. The findings contribute to minimizing the occurrence of paper roll tearing, thereby reducing the risk of inaccurate forecasts of the energy and heat produced by the CHP system. Furthermore, the methodology employed in this study may be effectively applied to other optimization problems in industrial processes.

A Deep Learning-Based Approach for High-Dimensional Industrial Steam Consumption Prediction to Enhance Sustainability Management

The continuous increase in industrialized sustainable development and energy demand, particularly in the use of steam, highlights the critical importance of efficient energy forecasting for sustainability. While current deep learning models have proven effective, they often involve numerous hyperparameters that are challenging to control and optimize. To address these issues, this research presents an innovative deep learning model, automatically fine-tuned using an improved RIME optimization algorithm (IRIME), with the aim of enhancing accuracy in energy forecasting. Initially, the bidirectional gated recurrent unit (BiGRU) exhibited promising results in prediction tasks but encountered difficulties in handling the complexity of high-dimensional time-series data related to industrial steam. To overcome this limitation, a bidirectional temporal convolutional network (BiTCN) was introduced to more effectively capture long-term dependencies. Additionally, the integration of a multi-head self-attention (MSA) mechanism enabled the model to more accurately identify and predict key features within the data. The IRIME-BiTCN-BiGRU-MSA model achieved outstanding predictive performance, with an R2 of 0.87966, MAE of 0.25114, RMSE of 0.34127, and MAPE of 1.2178, outperforming several advanced forecasting methods. Although the model is computationally complex, its high precision and potential for automation offer a promising tool for high-precision forecasting of industrial steam emissions. This development supports broader objectives of enhancing energy efficiency and sustainability in industrial processes.

Review of Digital Twins enabled applications for demand response

Digital Twins (DT), as one of the emerging trends in digitalisation, has attracted a large amount of research in the past few years due to its ability to provide more accurate models and better predictions for various energy-related applications. In the context of demand response (DR), the real-time bidirectional communication characteristic of DT allows for functions such as real-time energy forecast and decision making aid, which provided high potential in DR applications due to the nature of DR being highly time sensitive. The objective of this paper is to explore the concept and application of DT in the domain of DR, as there are currently limited numbers of review on this topic. It was identified that most work in this area shows promising results but is still exploratory. The main application of DT in the domain of DR is using DT as an enabling tool by leveraging the real-time and data processing function of DT for monitoring and decision-making, and providing a medium for better visualisation. Future studies can experiment with applying DT frameworks to larger case studies, and apply more elaborate DT with higher maturity levels.

Investigation of impact of community environment factors on the residents’ visiting rate to outdoor venues in South China

Understanding the activity patterns of urban residents is crucial for urban community design and urban energy forecasting. Predicting the visitation rates of different types of locations is a important issue because urban residents can only be in one place at a time. Achieving accurate predictions of outdoor public space visitation rates is equivalent to predicting the upper limit of the number of visitors to other buildings. Previous similar studies usually rely on survey methods such as questionnaires and GPS positioning. These methods have some drawbacks, such as small sample sizes and inaccurate data. In this paper, we used mobile signaling to investigate the park visit rates of residents in different communities in Guangzhou. In accordance with Chinese tradition, we define a community as a residential community, which consists of multiple residential units and a set of basic service facilities and management structures that are sufficient to meet the daily needs of residents. We found that the park visit rate of a community is closely related to the characteristics of the community itself. The park visit rate is positively correlated with the average LST (land surface temperature) of the communities, the average distance between the community and the large park, and the building coverage rate, while it is negatively correlated with the vegetation coverage rate, the average surface reflection rate, the convenience of transportation, and the per capita green space area. Given the different travel habits of urban residents on weekdays and weekends, we also used multiple linear regression to establish prediction models for community park visitation rates on weekdays and weekends.

Short‐term energy forecasting using deep neural networks: Prospects and challenges

AbstractThis study presents an in‐depth overview of deep neural networks (DNN) and their hybrid applications for short‐term energy forecasting (STEF). It examines DNN‐based STEF from three perspectives: basics, challenges, and prospects. The study compares recent literature using metrics like mean absolute error (MAE), mean average percentage error (MAPE), and root mean square error (RMSE). Findings indicate that combining automated data‐driven models with enhanced DNNs effectively addresses forecasting challenges. It also highlights the role of DNNs in integrating energy prosumers, renewable energy systems, microgrids, big data, and smart grids to improve STEF.

Particle swarm optimization tuned multi-headed long short-term memory networks approach for fuel prices forecasting

Increasing global energy demands and decreasing stocks of fossil fuels have led to a resurgence of research into energy forecasting. Artificial intelligence, explicitly time series forecasting holds great potential to improve predictions of cost and demand with many lucrative applications across several fields. Many factors influence prices on a global scale, from socio-economic factors to distribution, availability, and international policy. Also, various factors need to be considered in order to make an accurate forecast. By analyzing the current literature, a gap for improvements within this domain exists. Therefore, this work suggests and explores the potential of multi-headed long short-term memory models for gasoline price forecasting, since this issue was not tackled with multi-headed models before. Additionally, since the computational requirements for such models are relatively high, work focuses on lightweight approaches that consist of a relatively low number of neurons per layer, trained in a small number of epochs. However, as algorithm performance can be heavily dependent on appropriate hyper-parameter selections, a modified variant of the particle swarm optimization algorithm is also set forth to help in optimizing the model’s architecture and training parameters. A comparative analysis is conducted using energy data collected from multiple public sources between several contemporary optimizers. The outcomes are put through a meticulous statistical validation to ascertain the significance of the findings. The best-constructed models attained a mean square error of just 0.044025 with an R-squared of 0.911797, suggesting potential for real-world use.

Updates to NOAA’s Unified Forecast System’s Cumulus Convection Parameterization Scheme between GFSv16 and GFSv17

Abstract This study outlines the updates made to cumulus convection parameterizations between Global Forecast System, version 16 (GFSv16), and the forthcoming GFSv17 which will be the first global forecast application to become operational under the Unified Forecast System infrastructure. The updates, addressing known systematic errors and biases, incorporate innovations such as stochasticity, three-dimensional subgrid organizational effects, and a prognostic closure evolution. The changes are shown to improve tropical temperature/humidity biases, convective available potential energy (CAPE) forecasts, CONUS precipitation, and tropical variability. By examining individual updates’ impact on temperature, humidity profiles, and precipitation power spectra, we find that the new prognostic closure and stricter precipitation evaporation criteria alleviate a dry and cold bias in the tropical boundary layer and enhance precipitation variability in the tropics. Stricter triggering criteria also allow for more CAPE buildup, particularly over the tropics. The cumulus convection updates have a modest impact on precipitation skill scores over CONUS, but overall, there is an improvement when comparing GFSv16 and the latest GFSv17 prototype configurations, in particularly for larger thresholds. The study also highlights the challenges in developing convection parameterizations suitable for both coupled and uncoupled model configurations. Evaluation of the MJO shows varying responses to cumulus convection changes depending on whether the model is coupled with a dynamic ocean model.

Energy forecast for a cogeneration system using dynamic factor models

Cogeneration is used in different sectors of industry and it allows that two types of energy to be efficiently obtained from a single source. Accurate predictions are fundamental to optimize energy production, considering the variability that occurs in the daily market. This study adjusts and predicts cogeneration using real data from a Spanish energy technology center, using dynamic factor analysis methodology and incorporating covariates such as temperature and relative humidity. A comparative analysis is performed to evaluate the improvements achieved by implementing cluster-structured dynamic models versus other methods. Furthermore, a robust interpolation method has been implemented to handle missing data in both the main variable and the covariates.

GWO-ANFIS-RD3PG: A reinforcement learning approach with dynamic adjustment and dual replay mechanism for building energy forecasting

Reliable prediction of building energy consumption is essential for effective and sustainable energy management. Traditional forecasting methods, however, face challenges when dealing with sudden changes in energy consumption. To address the above-mentioned issue, we introduce the GWO-ANFIS-RD3PG prediction model in this paper. This model aims to ensure global forecasting accuracy while minimizing prediction errors at sudden change points. A Grey Wolf Optimization (GWO) algorithm with the Adaptive Neuro-Fuzzy Inference System (ANFIS) is proposed to predict the energy consumption type and its fluctuation magnitude for the next time step. Within the Recurrent Dual Experience Replay Buffer DDPG (RD3PG) prediction module, we employ Long Short-Term Memory (LSTM), as the actor network in the Deep Deterministic Policy Gradient (DDPG), to account for long-term dependencies in time series data. Notably, we introduce an adaptive action modification mechanism that allows the agent to optimize actions based on the output of the aforementioned ANFIS, enabling real-time adjustments during sudden change. To further enhance the model’s performance, we adopt a dual experience replay mechanism to store data samples from sudden change points, capturing insight information during these anomalies. Experimental results on two real-world datasets demonstrate that this method reduces MAE by 5.81% and 13.47%, and RMSE by 5.69% and 15.21% compared to the state-of-art models, showing the significance of the proposed method for energy efficiency improvement in real-world building operations.

Spatial and temporal analysis of decomposition models in China

This study presents a comprehensive evaluation of 15 decomposition models (12 empirical and 3 atmospheric transmittance models) for estimating diffuse horizontal irradiance at 17 radiation sites in China, using hourly radiation data from 2011 to 2020. Our results show distinct patterns in model performance across different geographical regions, seasons, and sky conditions. The Liu model demonstrates the best overall performance with an RMSE of 78.20 W/m2, while model accuracy shows significant geographical variation, performing best in South and Southeast China (RMSE<70 W/m2) and worst in the Qinghai-Tibet Plateau and Northwest China (RMSE>90 W/m2). Seasonal analysis reveals better performance in winter than in summer, with RMSE differences approaching 40 W/m2, mainly due to the higher proportion of solar elevation angles exceeding 30° in summer. Under different sky conditions (classified by clearness index: 0–0.35 for overcast, 0.35–0.65 for partly cloudy, 0.65–1 for clear skies), most models follow an RMSE pattern of partly cloudy > clear sky > overcast. However, the Reindl2, Boland, DIRINT, and DIRINDEX models deviate from this trend due to their formula structure and sensitivity to atmospheric parameters. To reduce these regional disparities, we propose a new region-specific model selection strategy: the DIRINDEX model for eastern regions, DIRINT for central areas, and Karatasou for western regions. This combined approach reduces the overall RMSE to 73.17 W/m2. This research deepens our understanding of the application of decomposition models in China's complex geographical and climatic conditions, offering valuable references for solar radiation modeling and renewable energy forecasting in diverse climatic regions.

A New Grey Prediction Model and Its Application in Renewable Energy Consumption

Renewable energy is an energy resource that can be used continuously. At present, international oil prices continue to rise, the problem of global climate change is becoming increasingly prominent, and renewable energy and clean energy have ushered in a new round of development opportunities. Based on the new gray prediction model, this paper forecasts the consumption of renewable energy and further analyzes the sustainable development of renewable energy. In this paper, the combinatorial optimization method of cumulative order, background value coefficient, and initial conditions, parameter optimization combination, parameter combinatorial optimization process of the gray prediction model, and parameter optimization mechanism based on the PSO algorithm are proposed, and the reduction error analysis is carried out. The consumption of wind power and photovoltaic renewable energy is forecasted, and three different forecasting methods as exponential smoothing method, time series analysis method, and new gray forecasting method are compared, and the wind speed, irradiation intensity, and load are forecasted by these three different forecasting methods. Compared with the time series analysis method and the exponential smoothing method, the RMSE of the new grey prediction method is reduced by 127.12% and 160.59%, and the error rate is reduced by 3.16% and 4%, respectively. Based on the consumption forecast of renewable energy, this paper analyzes its sustainability from three directions economy, resource supply, and environment, and finally gives energy policy recommendations.

Investigating Intelligent Forecasting and Optimization in Electrical Power Systems: A Comprehensive Review of Techniques and Applications

Electrical power systems are the lifeblood of modern civilization, providing the essential energy infrastructure that powers our homes, industries, and technologies. As our world increasingly relies on electricity, and modern power systems incorporate renewable energy sources, the challenges have become more complex, necessitating advanced forecasting and optimization to ensure effective operation and sustainability. This review paper provides a comprehensive overview of electrical power systems and delves into the crucial roles that forecasting and optimization play in ensuring future sustainability. The paper examines various forecasting methodologies from traditional statistical approaches to advanced machine learning techniques, and it explores the challenges and importance of renewable energy forecasting. Additionally, the paper offers an in-depth look at various optimization problems in power systems including economic dispatch, unit commitment, optimal power flow, and network reconfiguration. Classical optimization methods and newer approaches such as meta-heuristic algorithms and artificial intelligence-based techniques are discussed. Furthermore, the review paper examines the integration of forecasting and optimization, demonstrating how accurate forecasts can enhance the effectiveness of optimization algorithms. This review serves as a reference for electrical engineers developing sophisticated forecasting and optimization techniques, leading to changing consumer behaviors, addressing environmental concerns, and ensuring a reliable, efficient, and sustainable energy future.

AI for Climate Action: Leveraging Artificial Intelligence to Address Climate Change Challenges

This comprehensive article explores the transformative role of Artificial Intelligence (AI) in addressing the global climate crisis, focusing on its applications in climate modeling, renewable energy optimization, and disaster response. Through an extensive literature review, case studies, and expert interviews, we examine how AI technologies are revolutionizing climate prediction accuracy, enhancing data processing capabilities, and enabling sophisticated climate scenario simulations. The article delves into AI's contributions to renewable energy forecasting, smart grid management, and energy storage optimization, highlighting its potential to accelerate the transition to sustainable energy systems. We also investigate AI-driven approaches to disaster response and resilience, including early warning systems, resource allocation during climate-related disasters, and post-disaster recovery planning. While acknowledging the significant advancements AI brings to climate action, this study also addresses the challenges and limitations, such as data quality issues, ethical considerations, and technical barriers. Looking ahead, we discuss emerging AI technologies, their integration with other solutions, and the policy implications for effective climate change mitigation and adaptation. This research underscores the critical role of AI in combating climate change, while emphasizing the need for responsible development and deployment of these technologies to ensure a sustainable and resilient future.

Machine Learning for Optimising Renewable Energy and Grid Efficiency

This research investigates the application of machine learning models to optimise renewable energy systems and contribute to achieving Net Zero emissions targets. The primary objective is to evaluate how machine learning can improve energy forecasting, grid management, and storage optimisation, thereby enhancing the reliability and efficiency of renewable energy sources. The methodology involved the application of various machine learning models, including Long Short-Term Memory (LSTM), Random Forest, Support Vector Machines (SVMs), and ARIMA, to predict energy generation and demand patterns. These models were evaluated using metrics such as Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Key findings include a 15% improvement in grid efficiency after optimisation and a 10–20% increase in battery storage efficiency. Random Forest achieved the lowest MAE, reducing prediction error by approximately 8.5%. The study quantified CO2 emission reductions by energy source, with wind power accounting for a 15,000-ton annual reduction, followed by hydropower and solar reducing emissions by 10,000 and 7500 tons, respectively. The research concludes that machine learning can significantly enhance renewable energy system performance, with measurable reductions in errors and emissions. These improvements could help close the “ambition gap” by 20%, supporting global efforts to meet the 1.5 °C Paris Agreement targets.

Centralised vs. decentralised federated load forecasting in smart buildings: Who holds the key to adversarial attack robustness?

The integration of AI and ML into energy forecasting is crucial for modern energy management. Federated Learning (FL) is particularly noteworthy because it enhances data privacy and facilitates collaboration among distributed energy resources. It enables model training across multiple locations while minimizing reliance on centralized servers and data transfers. However, FL faces significant security challenges, particularly from adversarial attacks that can undermine the models' integrity and reliability. This paper addresses these security concerns by evaluating the effectiveness of Centralized Federated Learning (CFL) and Decentralized Federated Learning (DFL) in distributed load forecasting. We conducted a comparative analysis using two publicly available datasets, household data and smart grid data, for short-term load forecasting. Our study finds that DFL is more robust against adversarial attacks and artificial noise compared to CFL. Specifically, adversarial model poisoning attacks impact only the targeted client in DFL, while CFL is more broadly affected. For instance, in the presence of attacks, CFL's average client Mean Absolute Error (MAE) increased from 0.084 to 0.78 kWh for Dataset 1 and from 0.475 to 11.159 MW for Dataset 2, whereas DFL maintained relatively low error rates. Conversely, artificial noise had a less pronounced effect on CFL compared to DFL. Additionally, we introduce Decentralized Random Layer Aggregation (DRLA) to enhance DFL's robustness, offering further insights into improving FL methodologies in the energy sector. Our findings also indicate that only a single layer of the neural network is required for local updates during the aggregation process in DFL.

Signal-devices management and data-driven evidential constraints based robust dispatch strategy of virtual power plant

Ensuring the safety and reliability of energy systems is very important in power grid operation. However, inherent intricacies and uncertainties of modern-day power systems, such as the mismatch between signal frequency and equipment response speed and the stochasticity associated with renewable energy and load forecasts, create significant challenges for generation dispatch planning. This study proposes a robust optimization approach with constraints based on signal-devices management and data-driven evidential distribution constraints. The first stage of this approach is to decompose and recombine the net power demand according to the Hilbert-Huang transform and device capability. The signal-device management constraints are then established based on this. The second stage formulates an evidential constraint based on data-driven evidential distribution and chance constraints in a distributionally robust optimization framework that caters to the stochasticity associated with renewable energy and load forecasts. The proposed model is validated on a virtual power plant to assess the approach’s efficacy for different dispatching scenarios. Penalty-based sensitivity analysis provides further insights into the proposed method’s performance for varying levels of CO2 emissions. Simulation results demonstrate that system flexibility becomes increasingly crucial for maintaining system stability and security as the penetration of renewable energy grows. Compared with chance constraint, the proposed data-driven evidential constraint effectively enables the optimization framework to handle stochasticity after sacrificing 0.62% more economic loss and 0.7% more environmental loss. Excessively high penalty parameters for CO2 do not promote economic development, resulting in 21.08% and 52.51% more economic and environmental losses without obvious environmental protection.

Solar energy prediction with synergistic adversarial energy forecasting system (Solar-SAFS): Harnessing advanced hybrid techniques

An effective solar forecasting is an important part of managing and enhancing the effectiveness of solar power systems. The purpose of this paper is to develop a distinct and smart energy prediction system for solar PV systems. An efficient Solar Synergistic Adversarial Energy Forecasting System (Solar-SAFS) has been developed, which greatly enhances the precision and reliability of solar power predictions. It combines various deep learning methods including Graph Convolutional Networks (GCNs), Variational Autoencoders (VAEs), Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks to effectively understand spatial and time-related patterns in solar data. The strength and accuracy of this model are increased by using adversarial training, which makes sure it gives reliable predictions in different situations. For making the hyper parameters better and to increase model's effectiveness and efficiency, we are using Grey-Oystercatcher Hybrid Optimization (GOHO) technique. The Solar-SAFS is validated and tested with different data sets including Fingrid and DSK solar. It shows that this method gives better results compared to current deep learning models. Our findings suggest that Solar-SAFS improves the accuracy and stability of solar energy prediction by a large margin, offering an advanced solution for managing renewable energy resources. Moreover, it significantly helps to obtain better predictions for solar energy production, which is a crucial need in this sector. It also encourages more efficient and dependable forecasting methods for renewable energy sources like sunlight. Compared to existing deep learning models, the Solar-SAFS demonstrates significantly lower error rates across multiple performance metrics, including Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Bias Error (MBE).

OI url: https://doi.org/10.30574/wjarr.2024.23.3.2934

Artificial intelligence (AI) has emerged as a key enabler in optimizing renewable energy systems, significantly contributing to global efforts toward environmental sustainability. This review explores the application of AI technologies in enhancing the efficiency, reliability, and integration of renewable energy sources such as solar, wind, and hydropower. It focuses on how machine learning (ML), deep learning (DL), and other AI-driven algorithms improve energy forecasting, grid management, and storage optimization. Survey data and case studies demonstrate the potential of AI to minimize energy waste, reduce costs, and lower greenhouse gas emissions, reinforcing its role in transitioning to a sustainable energy future. The review concludes with a discussion of challenges and future research directions.