Short-Term Residential Load Forecasting (STRLF) is a core task in smart grid dispatching and energy management, and its accuracy directly affects the economy and stability of power systems. Current mainstream methods still have limitations in addressing issues such as complex temporal patterns, strong stochasticity of load data, and insufficient model interpretability. To this end, this paper proposes an explainable and efficient forecasting framework named KAN+Transformer, which integrates Kolmogorov-Arnold Networks (KAN) with Transformers. The framework achieves performance breakthroughs through three innovative designs: constructing a Reversible Mixture of KAN Experts (RMoK) layer, which optimizes expert weight allocation using a load-balancing loss to enhance feature extraction capability while preserving model interpretability; designing an attention-guided cascading mechanism to dynamically fuse the local temporal patterns extracted by KAN with the global dependencies captured by the Transformer; and introducing a multi-objective loss function to explicitly model the periodicity and trend characteristics of load data. Experiments on four power benchmark datasets show that KAN+Transformer significantly outperforms advanced models such as Autoformer and Informer; ablation studies confirm that the KAN module and the specialized loss function bring accuracy improvements of 7.2% and 4.8%, respectively; visualization analysis further verifies the model's decision-making interpretability through weight-feature correlation, providing a new paradigm for high-precision and explainable load forecasting in smart grids. Collectively, the results demonstrate our model's superior capability in representing complex residential load dynamics and capturing both transient and stable consumption behaviors. By enabling more accurate, interpretable, and computationally efficient short-term load forecasting, the proposed KAN+Transformer framework provides effective support for demand-side management, renewable energy integration, and intelligent grid operation. As such, it contributes to improving energy utilization efficiency and enhancing the sustainability and resilience of modern power systems.

To address the dual challenges of data privacy and statistical heterogeneity in smart grid load forecasting, this paper proposes FedDATT, a novel federated learning method based on a dual-attention mechanism and hierarchical privacy protection. The method operates on a three-tier federated learning architecture and introduces a multi-level attention framework to tackle Non-Independent and Identically Distributed (Non-IID) data. Internally, a dual temporal and feature-wise attention mechanism enables client models to capture critical data patterns. Externally, a client-attention mechanism adaptively weights model updates during global aggregation, enhancing robustness against both client drift and model poisoning attacks. To mitigate privacy risks such as gradient leakage, a hierarchical privacy strategy is implemented, combining Local Differential Privacy (LDP) at the client-side for source data protection with Homomorphic Encryption (HE) at the server-side for secure aggregation. Comprehensive experiments on the public CKW Smart Meter Dataset demonstrate FedDATT's superiority. Compared to baselines like FedAvg and FedSTA, our method reduces the Mean Absolute Percentage Error (MAPE) by up to 31% and 22%, respectively. Under a simulated sign-flipping attack, FedDATT's MAPE increases by only 2.45%, demonstrating high robustness. The results validate FedDATT as an effective and secure solution for collaborative load forecasting.

Reservoir Computing, known for its simple yet efficient architecture, excels in time series prediction tasks. In this study, we propose an "Ordered Aggregation Multi-Reservoir Computing Model (OAMRC)" for nonlinear time series prediction and classification. Built on a multi-layered architecture, the OAMRC model systematically aggregates reservoir states across time steps, capturing both short-term and long-term temporal features. By adjusting the number of layers and aggregation step sizes, the model flexibly captures multi-scale temporal patterns. Compared to existing multi-reservoir models, OAMRC enhances flexibility through precise aggregation weight adjustments. The model was applied to ECG signal classification in a five-class task, achieving a maximum accuracy of 92.7% with proper tuning of reservoir size and layers. Results highlight its capability to discern complex physiological patterns. Additionally, experiments on time series prediction tasks, such as the Mackey-Glass and Lorenz systems, demonstrate the OAMRC model's superior predictive performance over other state-of-the-art RC models, including Deep Reservoir Computing. In real-world applications requiring extended memory, such as traffic flow prediction, the OAMRC model achieves a normalized root mean square error of 0.04, showcasing its ability to capture prolonged trends and intricate correlations. A comparative analysis reveals its advantage over alternative models in addressing diverse temporal challenges, making it a robust tool for complex time series tasks. The implementation code of the proposed OAMRC model is publicly available at GitHub (https://github.com/yangxuesong556/OAMRC).

The traditional power load forecasting learning method has problems such as overfitting and incomplete learning of time series information when dealing with complex nonlinear data, which affects the accuracy of short–medium term power load forecasting. A joint learning method, LSVM-MKL, was proposed based on the bidirectional promotion of deep kernel learning (DKL) and multiple kernel learning (MKL). The multi-kernel method was combined with the input layer, the highest coding layer, and the highest encoding layer to model the network of the stack autoencoder (SAE) to obtain more comprehensive information. At the same time, the deep kernel was integrated into the optimization training of Gaussian multi-kernel by means of the nonlinear product to form the nonlinear composite kernel. Through a large number of reference datasets and actual industrial data experiments, it was shown that compared with the Elman and LSTM-Seq2Seq methods, the proposed method achieved a higher prediction accuracy of 4.32%, which verified its adaptability to complex time-varying power load forecasting processes and greatly improved the accuracy of power load forecasting.

The optimal dispatching of integrated energy systems can effectively reduce energy costs and decrease carbon emissions. The accuracy of the load forecasting method directly determines the dispatching outcomes, yet considering the stochastic and non-periodic characteristics of port electricity load, traditional load forecasting methods may not be suitable due to the weak historical regularity of the load data themselves. Therefore, this paper proposes a method for forecasting the electricity load of container ports based on ship arrival and departure schedules as well as port handling tasks. By finely modeling the electricity consumption behavior of port machinery, effective prediction of the main electricity load of ports is achieved. On this basis, the overall structure of an integrated port energy system (IPES) including renewable energy systems, electricity/thermal/cooling/hydrogen energy storage systems, integrated energy dispatching equipment, and integrated loads is studied. Furthermore, a dispatching model considering demand response for the optimal operation of the IPES is established, and the day-ahead optimal dispatching of the IPES is achieved based on the forecasted load. The experimental results indicate that the developed method can ensure the operational efficiency of IPES, reduce port energy costs, and decrease carbon emissions.

In the research titled Comprehensive AI-Driven Cost Dynamics Model (AICD-CDM) for Sustainable Green Building Projects, we delve into the burgeoning field of artificial intelligence to revolutionize cost estimation and control in green building construction. This study introduces AICD-CDM, a novel framework that integrates several advanced machine learning techniques, Light Gradient Boosting (LGBoost), and Natural Gradient Boosting (NGBoost), to address the multifaceted challenges of cost prediction and management in sustainable building projects. By leveraging the distinct strengths of these methods, the AICD-CDM model offers a multi-dimensional approach to cost estimation, providing not only point predictions but also probabilistic forecasts to better manage uncertainties inherent in green building projects. The model's capability to process complex, non-linear relationships between a multitude of cost-influencing factors makes it exceptionally adept at handling the intricate dynamics of sustainable construction. Furthermore, the integration of AI techniques ensures enhanced accuracy, adaptability, and computational efficiency, making the AICD-CDM an invaluable tool for decision-makers in the green building sector. This research not only contributes to the field of construction management by introducing a sophisticated cost control mechanism but also aligns with global sustainability goals by promoting efficient resource allocation and cost optimization in green buildings. The findings and methodologies of this study have the potential to set new benchmarks in the application of AI in sustainable construction management.

In recent years, the usage of new technologies in distributed generation is growing due to the electricity industry movement towards privatization and decentralized production of electric energy as well as competition among the new technologies. Microgrids are practical examples of the concept of decentralized production. Also, the employment of renewable energies is rapidly promoted due to the construction barriers of conventional power plants. Therefore, selecting and applying the appropriate technologies in planning and careful design of the type and capacity of generation and energy storage units are important. This paper proposed a robust model for day-ahead production scheduling of electrical and thermal units in a large-scale virtual power plant comprising a large number of distributed generation units and consumers accompanied by energy storage. To consider real conditions, the uncertainty of electrical demands is modeled by employing scenario-based analysis and proba-bilistic programming. So, five scenarios are created and modeled. In addition, two operation mode of a microgrid is considered connected and islanded modes. The objective function consists of total profit resulting from the daily operation of all units. The results confirm the applicability and effectiveness of the proposed model. Moreover, the total profit of the connected mode of the microgrid is about 20% more than the islanded mode, in all scenarios, and the cumulative profit cash flow of the connected mode is about 31% more than islanded mode.

Under the background of the new power system, there have been significant changes in the forms of the grid side, power source side, and load side. Clarifying the construction process and investment structure changes of the new power system from a development perspective is the key to discovering investment change patterns, guiding efficient investment of funds, and discovering and improving the investment efficiency of power grid infrastructure. This article first constructs a new power system evaluation index system and a new power system construction evaluation method from four dimensions: macro policies, power structure, grid structure, and terminal demand. Secondly, based on the higher demand for system regulation capability in the development stage of the power grid, as well as the urgent need to further improve power supply reliability, transaction flexibility, and diversity of power supply forms, an analysis of the investment direction of the power grid is conducted. Then, considering the changes in the demand for power grid investment functions, a power grid investment demand system that adapts to the new power system is constructed, including building a power grid investment demand calculation model based on project requirements at the project level, coordinating execution and planning, and considering factors such as load growth to construct a power grid investment demand calculation model that takes into account multiple factors. The power grid investment demand is calculated by combining bottom-up and top-down approaches. Finally, taking X province as an example, empirical calculations are conducted to verify the usability of the model.

This article has been retracted: please see Elsevier Policy on Article Withdrawal (https://www.elsevier.com/about/policies/article-withdrawal).Post-publication, the editor discovered suspicious changes in authorship between the original submission and the revised version of this paper. In summary, the author names Yan Zhang (new First Author and Corresponding Author), Fangmin Yuan, Huipeng Zhai and Chuang Song were added to the revised paper without explanation and without exceptional approval by the journal editor, which is contrary to the journal policy on changes to authorship.The editor reached out to the authors for an explanation, but they failed to provide a satisfactory explanation to these changes. In addition, it appears that Roza Poursoleiman was claiming an affiliation with Sun-Life Company, Baku, Azerbaijan. When questioned, the author was unable to provide convincing evidence of the existence and nature of this company. Overall, the editor feels that the findings of the manuscript cannot be relied uon and that the article needs to be retracted.

Under the background of energy transformation and sustainable development, new energy (NE) power generation has become one of the focuses of global attention. Large-scale NE power generation has had an important impact on the regional power grid, because NE is unstable and intermittent. The large-scale integration of NE has a significant impact on regional power grids because NE sources are characterized by their instability and intermittency, which differ greatly from the traditional fossil fuel-based power generation. The design of traditional power grids has been focused on meeting the stable electricity demands based on fossil fuel generation. However, the integration of NE introduces uncertainty factors. Therefore, in order to ensure the stable operation of the power grid after the NE generation is connected, it is necessary to carry out impact prediction and simulation. The purpose of this paper is to build a prediction model of the impact of large-scale NE generation access on regional power grid based on particle swarm optimization(PSO), so as to realize the goal of replacing traditional energy with clean energy. In this paper, PSO is used to establish a forecasting model, and various factors are fully considered. The simulation results show that the forecasting model of the impact of large-scale NE generation access on regional power grid based on PSO can get more accurate forecasting results, which provides scientific basis for formulating reasonable energy policies and planning and promotes the large-scale application and popularization of NE.

The traditional single-loop optimization planning method of source-network load storage analyzes the energy flow of distributed new energy, but it can't obtain the collaborative planning of different regions, and it has a great impact on the environment, which leads to a poor single-loop optimization planning effect. Based on this, a multi-scale single-loop optimization planning method of distributed new energy source-network load storage is proposed. Firstly, based on the concept of load storage of the source network, the output characteristics of distributed new energy are obtained, and the mathematical model of single-loop optimization planning of distributed new energy is constructed. Aiming at the weighting process of each sub-objective function in the mathematical model, three sub-objective functions are dynamically processed, and the weight results are calculated to solve the mathematical model results of multi-scale source-network load storage single ring optimization planning. On the basis of the results, a multi-scale single-loop optimization planning strategy of source-network load storage is designed, and the multi-scale single-loop optimization planning of source-network load storage is realized according to the current calculation of the voltage outer loop. The experimental results show that the frequency control deviation of this method is small and the single loop optimization planning effect is good.

The Internet of Things means that many of the daily devices used by humans will share their functions and information with each other or with humans by connecting to the Internet. The most important factor of the Internet of Things is the integration of several technologies and communication solutions. Identification and tracking technologies, wired and wireless sensors and active networks, protocols for increasing communication and intelligence of objects are the most important parts of the Internet of Things. In this article, an attempt has been made to determine the parts that can be used to make a house smart among the concepts and technologies related to web-based programs based on Internet of Things technology. Since it is very time-consuming to investigate the effect of all the Internet of Things technologies in smart homes, by studying and examining various types of research, the web-based program based on the Internet of Things is selected as an independent variable, and its effect on smart home management is investigated. For this purpose, a web-based program based on the Internet of Things for intelligent building energy management, intelligent equipment management, and intelligent security has been designed and implemented. As experimental results shown the proposed method the proposed method achieves better results compared to other existing methods in energy consumption by 33.8% reducing energy usage.

In the era of big data, various factors (particularly meteorological factors) have been considered in power load prediction, and the result shows a clear discrepancy in timescales. To capture the complicated multiscale relationship between power load and related factors, a novel multifactor and multiscale method is proposed for power load forecasting. Three primary steps are implemented: (1) multifactor analysis to select predictive factors via statistical tests; (2) multiscale analysis to extract scale-aligned components via multivariate empirical mode decomposition; and (3) power load prediction, including individual prediction at each timescale and ensemble prediction across different timescales. The empirical study focuses on the power load of Nanyang and indicates that the proposed multifactor and multiscale learning paradigms statistically outperform their corresponding original techniques (without multifactor and multiscale analysis) and semi-improved variants (with either multifactor or multiscale analysis) in terms of prediction accuracy.

This paper takes the annual data of total retail sales of consumer goods in China from 2001 to 2020 as the research object. Using methods such as principal component estimation, ridge regression and Lasso regression, a multivariate linear regression model is established, and the regression results of the three models are compared and analyzed, and the significant factors affecting the total retail sales of consumer goods in China are the income level of residents, the level of consumption of residents, and the development of the tertiary industry, and the less significant factors are demographic factors. In addition, this paper takes the annual data of china's total retail sales of social consumer goods from 1973 to 2020 as the research object, and uses the ALAMA (0,1,1) model to predict and analyze the total retail sales of social consumer goods in China in the next ten years, and concludes that the total retail sales of social consumer goods in China will gradually rise from 2021 to 2030, which is consistent with the actual situation. Thus, it puts forward rationalization suggestions for increasing the total retail sales of consumer goods in China and promoting economic development.

Although the accuracy of load forecasting has been studied by many works, the actual deployability of a model is rarely considered. In this work, we consider the actual deployability of a model from four aspects: 1) the prediction performance of the model; 2) the robustness of the model; 3) the dependence of the model on external data; and 4) the storage size of the model. From these four aspects, we propose a multiple wavelet convolutional neural network (MWCNN) for load forecasting. On two public data sets, we verified the performance and robustness of the MWCNN. The MWCNN only uses load data, and the storage size of the model is only 497 kB, which shows that MWCNN has good deployability. In addition, our MWCNN prediction results are interpretable. The experimental results show that the MWCNN can effectively capture the periodic characteristics of load data.