The dispatch optimization of Coal Mine Integrated Energy Systems (CMIES) is critical for energy efficiency and decarbonization in the mining industry, yet it poses challenges due to large-scale decision variables, strong multiple constraints, and a narrow feasible region. Existing constrained multi-objective evolutionary algorithms (CMOEAs) often fail to rapidly locate feasible solutions and suffer from slow convergence, making them impractical for real-time CMIES dispatch. To overcome these limitations, this paper proposes a Search Direction Learningbased Evolutionary Optimization (SDLEO) algorithm that prioritizes feasibility and convergence acceleration. A dual-direction learning mechanism via principal component analysis is proposed to guide the population toward high-quality Pareto regions, while tributary directions derived from constraint violation rankings steer the search directly into the union of feasible regions. An adaptive variable importance strategy selectively activates decision variables based on their sensitivity to objectives and constraints, further enhancing search efficiency. Three realistic dispatch optimization models for CMIES under typical scenarios are developed to faithfully characterize the complex constraints and trade-offs. Experimental results on benchmark problems and real-world CMIES cases demonstrate that SDLEO significantly outperforms 11 state-of-the-art CMOEAs, providing efficient solutions for large-scale constrained dispatch in CMIES.

Accurate short-term load forecasting is increasingly required across heterogeneous operating conditions, ranging from individual customers to higher aggregation levels (e.g., districts or regions). However, many existing approaches are developed for a specific setting and scale poorly across aggregation levels, while practical deployment is further complicated by limited historical data (e.g., cold-start users) and the need to adapt as demand patterns evolve. This article proposes Universal Online Functional Data Analysis (Universal-OFA), a unified framework for multiscale real-time daily load forecasting. The framework is designed to operate consistently across different user types and aggregation levels without full retraining. Universal-OFA represents daily load points as functional curves and organizes them into universal load profiles via a functional clustering module. It then performs real-time forecasting in an online forecasting module with a functional deep neural network that supports lightweight online updates. Using real-world smart meter data, we evaluate Universal-OFA at individual level with existing and new participants, and at higher aggregation levels with varying shares of new participants. Across both levels, Universal-OFA achieves strong forecasting performance, with particularly large improvements in scenarios with more new users. Beyond accuracy, Universal-OFA provides operational value in two ways. First, it supports the monitoring of load usage behavior shifts. Second, cost analysis under asymmetric penalties shows that Universal-OFA significantly decreases the forecast error cost (68.08% at the individual level and 80.36% at higher aggregation levels), indicating clear economic benefits in grid management.

The probabilities linguistic term set (PLTS) is an efficient tool to represent sentimental intensities hidden in unstructured text reviews that are useful for multicriteria online product ranking. Traditional machine learning-based sentiment analysis methods adopted in existing studies to obtain PLTSs often result in unsatisfying prediction accuracy and, thus, inevitably affect product ranking results. To overcome this limitation, in this study, we propose a deep learning-based sentiment analysis approach to produce PLTSs from online product reviews to rank online products. A natural language processing-based method is first applied to extract product features and corresponding feature texts from online reviews. Then, state-of-the-art deep learning-based models are implemented to conduct the sentiment classification for online product/feature review texts. To ensure classification accuracy, we propose an experimental matching mechanism to identify the level of sentiment tendency for all rating labels of a review dataset and then match each label with the most appropriate linguistic term. The experimental results reveal that our matching mechanism can benefit the training of a text classification model to identify sentiment tendencies from review texts with high prediction accuracy and with the help of the trained classification model, our approach can predict sentimental intensities of the extracted features' texts in the form of PLTSs with competitive accuracy. A case study of applying PLTSs output from our approach to an online product decision-making problem is also provided to validate the applicability of our approach.

Smart grids are essential for sustainable urban energy systems, improving efficiency and integrating renewable sources. Accurately forecasting load demand is key for effective management, but is challenging due to unpredictable behaviors and dynamic consumption patterns. This paper introduces a new data-driven approach using smart meter data from various buildings in Cardiff, UK to better understand electricity consumption behaviors across seasons. Our methodology combines machine learning techniques with an in-depth analysis of physical building characteristics to conduct dynamic load usage behavior simulation. We employ consensus-based clustering to identify buildings with similar consumption behaviors and track dynamic changes in load usage over time. Furthermore, we identify key load-related features that influence consumption patterns, enhancing the precision of load demand forecasting. Empirical validation of our approach underscores its effectiveness in enhancing forecast accuracy and providing robust, sustainable strategies for energy management within the smart grid paradigm.

Single cell RNA sequencing (scRNA-seq), a powerful tool for studying the tumor microenvironment (TME), does not preserve/provide spatial information on tissue morphology and cellular interactions. To understand the crosstalk between diverse cellular components in proximity in the TME, we performed scRNA-seq coupled with spatial transcriptomic (ST) assay to profile 41,700 cells from three colorectal cancer (CRC) tumor-normal-blood pairs. Standalone scRNA-seq analyses revealed eight major cell populations, including B cells, T cells, Monocytes, NK cells, Epithelial cells, Fibroblasts, Mast cells, Endothelial cells. After the identification of malignant cells from epithelial cells, we observed seven subtypes of malignant cells that reflect heterogeneous status in tumor, including tumor_CAV1, tumor_ATF3_JUN | FOS, tumor_ZEB2, tumor_VIM, tumor_WSB1, tumor_LXN, and tumor_PGM1. By transferring the cellular annotations obtained by scRNA-seq to ST spots, we annotated four regions in a cryosection from CRC patients, including tumor, stroma, immune infiltration, and colon epithelium regions. Furthermore, we observed intensive intercellular interactions between stroma and tumor regions which were extremely proximal in the cryosection. In particular, one pair of ligands and receptors (C5AR1 and RPS19) was inferred to play key roles in the crosstalk of stroma and tumor regions. For the tumor region, a typical feature of TMSB4X-high expression was identified, which could be a potential marker of CRC. The stroma region was found to be characterized by VIM-high expression, suggesting it fostered a stromal niche in the TME. Collectively, single cell and spatial analysis in our study reveal the tumor heterogeneity and molecular interactions in CRC TME, which provides insights into the mechanisms underlying CRC progression and may contribute to the development of anticancer therapies targeting on non-tumor components, such as the extracellular matrix (ECM) in CRC. The typical genes we identified may facilitate to new molecular subtypes of CRC.

This study develops a novel variant of particle swarm optimization (PSO), which improves its balance of exploration and exploitation by modifying neighborhood topology, self-adaptive parameter strategies and deep search, namely differential evolutionary evolution PSO with orthogonal learning (OL), i.e., DEEPSO-OL in short. Evolutionary computing can explore the solution space efficiently because of its self-evolving attribute as iteration continues. The OL enhances its exploitation by focusing on deeper search for promising solutions. It utilizes the concept of orthogonal experimental design (OED) which predicts the best combination of control variables without exhaustive evaluation of all possible combinations. In addition, to avoid premature convergence in a local optimum, a stochastic star topology for particles is proposed. Such topology ensures just enough communication among the best performing particles, while encouraging them to explore other spaces. The efficacy of the algorithm is evaluated through real-world scenarios such as optimal power flow (OPF) and wind integrated OPF, which are hard to solve with classical mathematical methods. The proposed algorithm is run on a modified IEEE 30-bus test system and compared to the state-of-the-art evolutionary computing algorithms for a variety of cost objective functions with high levels of non-linearity and non-convexity. The DEEPSO-OL demonstrates its performance to generate more accurate feasible solutions and construct promising and efficient search method for real-world complex optimization problems.

Genetic screening of newborns for deafness plays an important role in elucidating the etiology of deafness, diagnosing it early, and intervening in it. Genetic screening of newborns has been conducted for 11 years in Beijing. It started with a chip to screen for 9 variants of 4 genes in 2012; the chip screened for 15 variants of those genes in 2018, and it now screens for 23 variants of those genes. In the current study, a comparative analysis of three screening protocols and follow-up for infants with pathogenic variants was performed. The rates of detection and hearing test results of infants with pathogenic variants were analyzed. Subjects were 493,821 infants born at 122 maternal and child care centers in Beijing from April 2012 to August 2023. Positivity increased from 4.599% for the chip to screen for 9 variants to 4.971% for the chip to screen for 15 variants, and further to 11.489% for the chip to screen for 23 variants. The carrier frequency of the GJB2 gene increased from 2.489% for the chip to screen for 9 variants and 2.422% for the chip to screen for 15 variants to 9.055% for the chip to screen for 23 variants. The carrier frequency of the SLC26A4 gene increased from 1.621% for the chip to screen for 9 variants to 2.015% for the chip to screen for 15 variants and then to 2.151% for the chip to screen for 23 variants. According to the chip to screen for 9 variants and the chip to screen for 15 variants, the most frequent mutant allele was c.235delC. According to the chip to screen for 23 variants, the most frequent mutant allele was c.109G>A. The chip to screen for 15 variants was used to screen 66.67% (14/21) of newborns with biallelic variants in the SLC26A4 gene for newly added mutations. The chip to screen for 23 variants was used to screen 92.98% (53/57) of newborns with biallelic variants in the GJB2 gene (52 cases were biallelic c.109G>A) and 25% (1/4) of newborns with biallelic variants in the SLC26A4 gene for newly added mutations. Among the infants with pathogenic variants (biallelic variants in GJB2 or SLC26A4), 20.66% (25/121) currently have normal hearing. In addition, 34.62% (9/26) of newborns who passed the hearing screening were diagnosed with hearing loss. Findings indicate that a growing number of newborns have benefited, and especially in the early identification of potential late-onset hearing loss, as the number of screening sites has increased. Conducting long-term audiological monitoring for biallelic variants in individuals with normal hearing is of paramount significance.

Abstract circRNADisease v2.0 is an enhanced and reliable database that offers experimentally verified relationships between circular RNAs (circRNAs) and various diseases. It is accessible at http://cgga.org.cn/circRNADisease/ or http://cgga.org.cn:9091/circRNADisease/. The database currently includes 6998 circRNA-disease entries across multiple species, representing a remarkable 19.77-fold increase compared to the previous version. This expansion consists of a substantial rise in the number of circRNAs (from 330 to 4246), types of diseases (from 48 to 330) and covered species (from human only to 12 species). Furthermore, a new section has been introduced in the database, which collects information on circRNA-associated factors (genes, proteins and microRNAs), molecular mechanisms (molecular pathways), biological functions (proliferation, migration, invasion, etc.), tumor and/or cell line and/or patient-derived xenograft (PDX) details, and prognostic evidence in diseases. In addition, we identified 7 159 865 relationships between mutations and circRNAs among 30 TCGA cancer types. Due to notable enhancements and extensive data expansions, the circRNADisease 2.0 database has become an invaluable asset for both clinical practice and fundamental research. It enables researchers to develop a more comprehensive understanding of how circRNAs impact complex diseases.

Short-term load forecasting is critical in energy management and grid operations. Traditional statistical learning methods lack fitting capability, and recent deep learning-based time series forecasting methods suffer from poor generalization. In this study, an ensemble learning algorithm is proposed based on the additive attention mechanism, seamlessly integrating the LSTM algorithm and the Prophet algorithm. Comparatively experimental results demonstrate that the proposed method reasonably allocates dynamic weights to the two models. Its predictive performance exhibits a substantial improvement over individual algorithms, and the ensemble algorithms with static weights. Furthermore, it also outperforms approaches employing Support Vector Regression, Ridge Regression, Extreme Gradient Boosting Regressor and Artificial Neural Network as ensemble learning algorithms.

Domain adaptation solves image classification problems in the target domain by taking advantage of the labelled source data and unlabelled target data.Usually, the source and target domains share the same set of classes.As a special case, Open-Set Domain Adaptation (OSDA) assumes there exist additional classes in the target domain but are not present in the source domain.To solve such a domain adaptation problem, our proposed method learns discriminative common subspaces for the source and target domains using a novel Open-Set Locality Preserving Projection (OSLPP) algorithm.The source and target domain data are aligned in the learned common spaces classwise.To handle the open-set classification problem, our method progressively selects target samples to be pseudo-labelled as known classes, rejects the outliers if they are detected as unknown classes, and leaves the remaining target samples as uncertain.The common subspace learning algorithm OSLPP simultaneously aligns the labelled source data and pseudo-labelled target data from known classes and pushes the rejected target data away from the known classes.The common subspace learning and the pseudo-labelled sample selection/rejection facilitate each other in an iterative learning framework and achieve state-of-the-art performance on four benchmark datasets Office-31, Office-Home, VisDA17 and Syn2Real-O with the average HOS of 87.6%, 67.0%, 76.1% and 65.6% respectively.

As the retail industry increasingly adopts an omnichannel strategy amidst the growth of e-commerce, this study explores the interplay between inventory and return policies in such a setting. Specifically, we focus on “return losses,” defined as the losses incurred by retailers due to customer product returns. We target how these return losses impact retailers’ profits and physical inventories. While previous research has mainly concentrated on cross-channel returns, little attention has been paid to how omnichannel return policies affect store inventory and profit. We posit that profit-maximizing retailers allow consumers to select their purchase channels based on utility but require returns to follow specific policies. In this paper, we model four return policies based on the channel of return: original purchasing channel return, offline return, online return, and cross-channel return. We use a newsvendor model to construct an optimal profit function that accounts for the additional profit from offline return, uncertainty demand, and inventory cost. Our analysis identifies the conditions under which certain return policies are beneficial or detrimental to omnichannel retailers. We discover that the retailer's product pricing, return losses, and consumer return hassle costs are the main factors influencing the best return policy and inventory policy decisions. Moreover, whether to increase or decrease store inventory and the relationship between physical inventory and return loss depend on the return policy, price, and return hassle cost. Numerical simulations support our findings, which offer practical guidance for omnichannel retailers aiming to optimize their inventory and return policies.

Non-intrusive load monitoring (NILM), which usually utilizes machine learning methods and is effective in disaggregating smart meter readings from the household level into appliance-level consumption, can help analyze the electricity consumption behaviors of users and enable practical smart energy and smart grid applications. Recent studies have proposed many novel non-intrusive load monitoring frameworks based on federated deep learning. However, there is a lack of comprehensive research exploring the utility optimization schemes and the privacy-preserving schemes in different federated learning-based NILM application scenarios. In this study, a distributed and privacy-preserving non-intrusive load monitoring (DP2-NILM) framework was developed to make the first attempt to conduct federated learning-based NILM focusing on both utility optimization and privacy-preserving. Specifically, two alternative federated learning strategies are examined in the utility optimization schemes, i.e., the FedAvg and the FedProx. Moreover, different levels of privacy guarantees, i.e., the local differential privacy federated learning and the global differential privacy federated learning are provided in the DP2-NILM. Extensive comparison experiments are conducted on three real-world datasets to evaluate the proposed framework.

This paper proposes a customized energy pricing scheme for energy retailers in multi-energy (i.e., electricity and natural gas) retail markets. Microgrids with distributed energy resources (DERs) and demand response (DR) programs are considered on the demand side. We adopt a bilevel single-leader multi-follower model to analyze the customized multi-energy pricing decisions where the retailer's profit maximization problem is formulated at the upper level, and the microgrids' operation costs minimization problems are considered at the lower level. A particle swarm optimization (PSO) based evolutionary solution approach is developed to solve the proposed bilevel decision-making problem. Through a numerical case study, we demonstrate the feasibility and effectiveness of the proposed bilevel model and the solution algorithm. We reveal that the proposed customized pricing scheme could offer differentiated optimal pricing decisions to various microgrids characterized by their energy conversion efficiencies.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

In the context of increasing demands for long-term multi-energy load forecasting in real-world applications, this paper introduces Patchformer, a novel model that integrates patch embedding with encoder-decoder Transformer-based architectures. To address the limitation in existing Transformer-based models, which struggle with intricate temporal patterns in long-term forecasting, Patchformer employs patch embedding, which predicts multivariate time-series data by separating it into multiple univariate data and segmenting each of them into multiple patches. This method effectively enhances the model's ability to capture local and global semantic dependencies. The numerical analysis shows that the Patchformer obtains overall better prediction accuracy in both multivariate and univariate long-term forecasting on the novel Multi-Energy dataset and other benchmark datasets. In addition, the positive effect of the interdependence among energy-related products on the performance of long-term time-series forecasting across Patchformer and other compared models is discovered, and the superiority of the Patchformer against other models is also demonstrated, which presents a significant advancement in handling the interdependence and complexities of long-term multi-energy forecasting. Lastly, Patchformer is illustrated as the only model that follows the positive correlation between model performance and the length of the past sequence, which states its ability to capture long-range past local semantic information.

Various real-world problems can be attributed to constrained multiobjective optimization problems (CMOPs). Although there are various solution methods, it is still very challenging to automatically select efficient solving strategies for CMOPs. Given this, a process knowledge-guided constrained multiobjective autonomous evolutionary optimization method is proposed. First, the effects of different solving strategies on population states are evaluated in the early evolutionary stage. Then, the mapping model of population states and solving strategies is established. Finally, the model recommends subsequent solving strategies based on the current population state. This method can be embedded into existing evolutionary algorithms, which can improve their performances to different degrees. The proposed method is applied to 41 benchmarks and 30 dispatch optimization problems of the integrated coal mine energy system. Experimental results verify the effectiveness and superiority of the proposed method in solving CMOPs.

We address the Unsupervised Domain Adaptation (UDA) problem in image classification from a new perspective. In contrast to most existing works which either align the data distributions or learn domain-invariant features, we directly learn a unified classifier for both the source and target domains in the high-dimensional homogeneous feature space without explicit domain alignment. To this end, we employ the effective Selective Pseudo-Labelling (SPL) technique to take advantage of the unlabelled samples in the target domain. Surprisingly, data distribution discrepancy across the source and target domains can be well handled by a computationally simple classifier (e.g., a shallow Multi-Layer Perceptron) trained in the original feature space. Besides, we propose a novel generative model norm-AE to generate synthetic features for the target domain as a data augmentation strategy to enhance the classifier training. Experimental results on several benchmark datasets demonstrate the pseudo-labelling strategy itself can lead to comparable performance to many state-of-the-art methods whilst the use of norm-AE for feature augmentation can further improve the performance in most cases. As a result, our proposed methods (i.e. naive-SPL and norm-AE-SPL) can achieve comparable performance with state-of-the-art methods with the average accuracy of 93.4% and 90.4% on Office-Caltech and ImageCLEF-DA datasets, and achieve competitive performance on Digits, Office31 and Office-Home datasets with the average accuracy of 97.2%, 87.6% and 68.6% respectively.

As the distributed energy resources (DERs) increasingly penetrate the unbalanced distribution network, it becomes challenging to accommodate such penetration technically and economically. Therefore, this paper tackles an optimal allocation of PV systems (locations and sizes) to maximize the penetration while minimizing voltage violation. It is challenging because the problem is a mixed integer nonlinear programming (MINLP) problem with non-linear and non-convex properties. In addition, the network is unbalanced which brings burdens on solving load flows. Computational intelligent methods, particularly evolutionary algorithms (EAs) have proven its efficiency and robustness in large optimization problems and thus, this paper explores two EAs on the problem with the help of a robust unbalanced load flow algorithm. A comparative study is conducted on particle swarm optimization (PSO) and artificial bee colony (ABC) based on IEEE 13 and 37 bus systems. Optimal allocation based on peak hour and day-ahead scenarios are considered. After 30 times run, the test cases have shown that both EAs are successful and yet ABC generally converges to better solution and yet with larger statistical deviations on solutions.

A customized and cost-effective building thermal control system is critical for accommodating thermal performance differences within the building, as well as satisfying the individual thermal comfort needs of occupants. Moreover, incorporating a building indoor thermal simulation procedure into the thermal control system can reduce the necessity of installing various expensive sensors (e.g., wearable sensors for personal thermal comfort management) in individual offices, as well as the requirement of extensive computing facilities without rendering the control performance, resulting into more sustainable building operations. An important step in achieving the above-mentioned goal is understanding how different offices/rooms behave differently given the same outdoor weather conditions. This study proposes a smart building indoor thermal profiling system to identify underlying physical factors that affect thermal performance in different seasons and to track dynamic cluster trajectories of considered offices to suggest indoor thermal optimization strategies. A consensus-based clustering approach is adopted to robustly cluster offices into different groups based on their hourly indoor temperature profiles for different seasons. Experimental results showed that our proposed approach could effectively discover more indoor thermal patterns in the buildings and is able to identify distinct dynamic cluster trajectories across four seasons (i.e., eight distinct dynamic trajectories in our case study). The data-driven analysis conducted in this study also indicated promising applications of the proposed smart building indoor thermal profiling system in effectively guiding the design of customized thermal control strategies for buildings. It also suggested that the proposed approach could be applied to a wide range of other applications, such as customized building energy management, energy pricing, as well as the economic benefit analysis of building retrofits and design.