Search Results (52)

Accurate short-term weather forecasting plays a vital role in disaster response, agriculture, and energy management, where timely and reliable predictions are essential for decision-making. Graph neural networks (GNNs), known for their ability to model complex spatial structures and relational data, have achieved remarkable success in meteorological forecasting by effectively capturing spatial dependencies among distributed weather stations. However, most existing GNN-based approaches rely on pairwise station connections, limiting their capacity to represent higher-order spatial interactions. Moreover, their dependence on supervised learning makes them vulnerable to spatial heterogeneity and temporal non-stationarity. This paper introduces a novel spatial–temporal pretraining framework, Hypergraph-enhanced Meteorological Pretraining (HyMePre), which combines hypergraph neural networks with self-supervised learning to model high-order spatial dependencies and improve generalization across diverse climate regimes. HyMePre employs a two-stage masking strategy, applying spatial and temporal masking separately, to learn disentangled representations from unlabeled meteorological time series. During forecasting, dynamic hypergraphs group stations based on meteorological similarity, explicitly capturing high-order dependencies. Extensive experiments on large-scale reanalysis datasets show that HyMePre outperforms conventional GNN models in predicting temperature, humidity, and wind speed. The integration of pretraining and hypergraph modeling enhances robustness to noisy data and improves generalization to unseen climate patterns, offering a scalable and effective solution for operational weather forecasting. Full article

The spread of COVID-19 is influenced by multiple factors, including control policies, virus characteristics, individual behaviors, and environmental conditions, exhibiting highly complex nonlinear dynamic features. The time series of new confirmed cases shows significant nonlinearity and non-stationarity. Traditional prediction methods that rely solely on one-dimensional case data struggle to capture the multi-dimensional features of the data and are limited in handling nonlinear and non-stationary characteristics. Their prediction accuracy and generalization capabilities remain insufficient, and most existing studies focus on single-step forecasting, with limited attention to multi-step prediction. To address these challenges, this paper proposes a multi-module fusion prediction model—TSMixer-BiKSA network—that integrates multi-feature inputs, Variational Mode Decomposition (VMD), and a dual-branch parallel architecture for 1- to 3-day-ahead multi-step forecasting of new COVID-19 cases. First, variables highly correlated with the target sequence are selected through correlation analysis to construct a feature matrix, which serves as one input branch. Simultaneously, the case sequence is decomposed using VMD to extract low-complexity, highly regular multi-scale modal components as the other input branch, enhancing the model’s ability to perceive and represent multi-source information. The two input branches are then processed in parallel by the TSMixer-BiKSA network model. Specifically, the TSMixer module employs a multilayer perceptron (MLP) structure to alternately model along the temporal and feature dimensions, capturing cross-time and cross-variable dependencies. The BiGRU module extracts bidirectional dynamic features of the sequence, improving long-term dependency modeling. The KAN module introduces hierarchical nonlinear transformations to enhance high-order feature interactions. Finally, the SA attention mechanism enables the adaptive weighted fusion of multi-source information, reinforcing inter-module synergy and enhancing the overall feature extraction and representation capability. Experimental results based on COVID-19 case data from Italy and the United States demonstrate that the proposed model significantly outperforms existing mainstream methods across various error metrics, achieving higher prediction accuracy and robustness. Full article

This paper examines the relationship between dependence and independence alternatives in general stochastic processes and explores the duality between the true (yet unknown) stochastic process and the functional representation that fits the observed data. We demonstrate that the solution depends on its historic realizations, challenging existing theoretical frameworks that assume independence between the solution and the history of the true process. Under orthogonality conditions, we investigate parameter spaces within data-generating processes and establish conditions under which data exhibit mean-reverting, random, cyclical, history-dependent, or explosive behaviors. We validate our theoretical framework through empirical analysis of an extensive dataset comprising daily prices from the S&P500, 10-year US Treasury bonds, the EUR/USD exchange rate, Brent oil, and Bitcoin from 1 January 2002 to 1 February 2024. Our out-of-sample predictions, covering the period from 17 February 2019 to 1 February 2024, demonstrate the model’s exceptional forecasting capability, yielding correct predictions with between 73% and 92% accuracy, significantly outperforming naïve and moving average models, which only achieved 47% to 54% accuracy. Full article

Accurate power load forecasting is crucial for ensuring grid stability, optimizing economic dispatch, and facilitating renewable energy integration in modern smart grids. However, real load forecasting is often disturbed by the inherent non-stationarity and multi-factor coupling effects. To address this problem, a novel hybrid forecasting framework based on adaptive mode decomposition (AMD) and improved least squares support vector machine (ILSSVM) is proposed for effective short-term power load forecasting. First, AMD is utilized to obtain multiple components of the power load signal. In AMD, the minimum energy loss is used to adjust the decomposition parameter adaptively, which can effectively decrease the risk of generating spurious modes and losing critical load components. Then, the ILSSVM is presented to predict different power load components, separately. Different frequency features are effectively extracted by using the proposed combination kernel structure, which can achieve the balance of learning capacity and generalization capacity for each unique load component. Further, an optimized genetic algorithm is deployed to optimize model parameters in ILSSVM by integrating the adaptive genetic algorithm and simulated annealing to improve load forecasting accuracy. The real short-term power load dataset is collected from Guangxi region in China to test the proposed forecasting framework. Extensive experiments are carried out and the results demonstrate that our framework achieves an MAPE of 1.78%, which outperforms some other advanced forecasting models. Full article

Wind energy has emerged as a cornerstone in global efforts to transition to renewable energy, driven by its low environmental impact and significant generation potential. However, the inherent intermittency of wind, influenced by complex and dynamic atmospheric patterns, poses significant challenges for accurate wind speed prediction. Existing approaches, including statistical methods, machine learning, and deep learning, often struggle with limitations such as non-linearity, non-stationarity, computational demands, and the requirement for extensive, high-quality datasets. In response to these challenges, we propose a novel neighborhood preserving cross-dataset data augmentation framework for high-horizon wind speed prediction. The proposed method addresses data variability and dynamic behaviors through three key components: (i) the uniform manifold approximation and projection (UMAP) is employed as a non-linear dimensionality reduction technique to encode local relationships in wind speed time-series data while preserving neighborhood structures, (ii) a localized cross-dataset data augmentation (DA) approach is introduced using UMAP-reduced spaces to enhance data diversity and mitigate variability across datasets, and (iii) recurrent neural networks (RNNs) are trained on the augmented datasets to model temporal dependencies and non-linear patterns effectively. Our framework was evaluated using datasets from diverse geographical locations, including the Argonne Weather Observatory (USA), Chengdu Airport (China), and Beijing Capital International Airport (China). Comparative tests using regression-based measures on RNN, GRU, and LSTM architectures showed that the proposed method was better at improving the accuracy and generalizability of predictions, leading to an average reduction in prediction error. Consequently, our study highlights the potential of integrating advanced dimensionality reduction, data augmentation, and deep learning techniques to address critical challenges in renewable energy forecasting. Full article

The sustainable management of state-owned forest regions is significant for improving the nationally determined contribution and achieving carbon neutrality. The administrative area of key state-owned forest regions in northeast China and Inner Mongolia, hereafter referred to as forest regions, spans a forested area of 27.16 million hectares and a forest coverage rate of 82.97%. This represents China’s largest state-owned forest resource base, with extensive and concentrated forest areas. However, despite this vast forest coverage, the region’s forest stand density remains below the national and global average, underscoring the need for improved carbon sequestration performance. This study used the Stochastic Frontier Analysis (SFA) method to measure the carbon sequestration efficiency of key state-owned forest regions in northeast China and Inner Mongolia. A spatiotemporal Geographically and Temporally Weighted Regression model (GTWR) was employed to reveal the spatiotemporal non-stationarity of the driving mechanism of carbon sequestration efficiency. Finally, the benchmarking management method was applied to predict the carbon sequestration potential. The results indicated that the carbon sequestration efficiency of forest regions exhibited an overall increasing trend over time, with significant spatial and temporal heterogeneity among forest industry enterprises (forest farms). Specifically, the carbon sequestration efficiency ranked from highest to lowest is as follows: Greater Khingan Forestry Group, Inner Mongolia Forestry Industry Group, Longjiang Forestry Industry Group, Changbai Mountain Forestry Industry Group, Jilin Forestry Industry Group, and Yichun Forestry Industry Group. Furthermore, carbon sequestration efficiency was driven by both natural and socioeconomic factors, but the effects of these factors were spatiotemporally non-stationary. Generally, enterprise output value, labor compensation, tending, and accumulated temperature had positive effects on carbon sequestration efficiency, while capital structure, altitude, and precipitation had negative effects. Finally, our findings revealed that the carbon sequestration potential of forest regions is substantial. If technical efficiency is improved, the carbon sequestration potential of forest regions could expand by 0.86 times the current basis, reaching 31.29 mtCO₂ by 2030. These results underscore the importance of respecting the differences and conditionality of forest development paths and promoting the sustainable management of key state-owned forest regions through scientific approaches, which is crucial for achieving carbon neutrality goals. Full article

Despite its interpretability and excellence in time series forecasting, the fuzzy time series forecasting model (FTSFM) faces significant challenges when handling non-stationary time series. This paper proposes a novel hybrid non-stationary FTSFM that integrates time-variant FTSFM, Bayesian network (BN), and non-stationary fuzzy sets. We first apply first-order differencing to extract the fluctuation information of the time series while reducing non-stationarity. A novel time-variant FTSFM updating method is proposed to effectively merge historical knowledge with new observations, enhancing model stability while maintaining sensitivity to time series changes. The updating of fuzzy sets is achieved by incorporating non-stationary fuzzy sets and prediction residuals. Based on updated fuzzy sets, the system reconstructs fuzzy logical relationship groups by combining historical and new data. This approach implements dynamic quantitative modeling of fuzzy relationships between historical and predicted moments, integrating valuable historical temporal fuzzy patterns with emerging temporal fuzzy characteristics. This paper further develops an adaptive BN structure learning method with an adaptive scoring function to update temporal dependence relationships between any two moments while building upon existing dependence relationships. Experimental results indicate that the proposed model significantly outperforms benchmark algorithms. Full article

Factors related to catering distribution are typically characterized by local changes, but few studies have quantitatively investigated the inherent spatial nonstationarity correlations. In this study, a multiscale geographically weighted regression (MGWR) model was adopted to locally examine the impact of various factors on catering distribution, which were obtained through a novel method incorporating GeoDetector analysis and exploratory factor analysis (EFA) using point of interest (POI) data. GeoDetector analysis was used to identify the effective variables that truly contribute to catering distribution, and EFA was adopted to extract interpretable latent factors based on the underlying structure of the effective variables and thus eliminate multicollinearity. In our case study in Nanjing, China, four primary factors, namely commuting activities, shopping activities, tourism activities, and gathering activities, were retained from eight categories of POIs with respect to catering distribution. The results suggested that GeoDetector working in tandem with EFA could improve the representativeness of factors and infer POI configuration patterns. The MGWR model explained the most variations (adj. R²: 0.903) with the lowest AICc compared to the OLS regression model and the geographically weighted regression (GWR) model. Mapping MGWR parameter estimates revealed the spatial variability of relationships between various factors and catering distribution. The findings provide useful insights for guiding catering development and optimizing urban functional spaces. Full article

The nonstationarity of extreme precipitation is now well established in the presence of climate change and low-frequency variability. Consequently, the implications for urban flooding, for which there are not long flooding records, need to be understood better. The vulnerability is especially high in coastal cities, where the flat terrain and impervious cover present an additional challenge. In this paper, we estimate the time-varying probability distributions for hourly and daily extreme precipitation using the Generalized Additive Model for Location Scale and Shape (GAMLSS), employing different climate indices, such as Atlantic Multi-Decadal Oscillation (AMO), the El Niño 3.4 SST Index (ENSO), Pacific Decadal Oscillation (PDO), the Western Hemisphere Warm Pool (WHWP) and other covariates. Applications to selected coastal cities in the USA are considered. Overall, the AMO, PDO and WHWP are the dominant factors influencing the extreme rainfall. The nonstationary model outperforms the stationary model in 92% of cases during the fitting period. However, in terms of its predictive performance over the next 5 years, the ST model achieves a higher log-likelihood in 86% of cases. The implications for the time-varying design rainfall in coastal areas are considered, whether this corresponds to a structural design or the duration of a contract for a financial instrument for risk securitization. The opportunity to use these time-varying probabilistic models for adaptive flood risk management in a coastal city context is discussed. Full article

With the increasing research of deep learning in the EEG field, it becomes more and more important to fully extract the characteristics of EEG signals. Traditional EEG signal classification prediction neither considers the topological structure between the electrodes of the signal collection device nor the data structure of the Euclidean space to accurately reflect the interaction between signals. Graph neural networks can effectively extract features of non-Euclidean spatial data. Therefore, this paper proposes a feature selection method for epilepsy EEG classification based on graph convolutional neural networks (GCNs) and long short-term memory (LSTM) cells. While enriching the input of LSTM, it also makes full use of the information hidden in the EEG signals. In the automatic detection of epileptic seizures based on neural networks, due to the strong non-stationarity and large background noise of the EEG signal, the analysis and processing of the EEG signal has always been a challenging research. Therefore, experiments were conducted using the preprocessed Boston Children’s Hospital epilepsy EEG dataset, and input it into the GCN-LSTM model for deep feature extraction. The GCN network built by the graph convolution layer learns spatial features, then LSTM extracts sequence information, and the final prediction is performed by fully connected and softmax layers. The introduced method has been experimentally proven to be effective in improving the accuracy of epileptic EEG seizure detection. Experimental results show that the average accuracy of binary classification on the CHB-MIT dataset is 99.39%, and the average accuracy of ternary classification is 98.69%. Full article

The spatial variability of geotechnical parameters, such as soil shear wave velocity (V_s), exhibits significant nonlinearity and non-stationarity with respect to depth (h) due to the influence of overlying stress. Existing stochastic field models for describing the variability of geotechnical parameters are insufficient for simultaneously capturing both the nonlinearity and non-stationarity of these parameters. In this study, a power function V_s = V_s0[f(h)]ⁿ is proposed to describe the nonlinear trend in soil shear wave velocity (V_s) as a function of depth-related variable f(h). Considering the physical significance and correlation of the power function parameters V_s0 and n, the variability of these parameters is modeled using a random variable model and a stationary stochastic field model, respectively. This leads to the development of a non-stationary stochastic field model that describes the spatial variability of V_s. The proposed method is applied to simulate the random V_s-structure of a seabed site in China, and the obtained V_s results are used to assess the liquefaction probability of the seabed. The results indicate that ignoring the correlation between geotechnical parameters significantly increases the variability of the final simulation results. However, the proposed method accurately captures the nonlinear trend and non-stationary characteristics of soil V_s with depth, and the liquefaction probability predictions are consistent with those derived from in situ V_s measurements in the study area. This approach provides valuable guidance for simulating the spatial variability of depth-dependent geotechnical parameters, particularly those significantly influenced by overlying pressure. Full article

Accurate load forecasting is crucial to the stable operation of integrated energy systems (IES), which plays a significant role in advancing sustainable development. Addressing the challenge of insufficient prediction accuracy caused by the inherent uncertainty and volatility of load data, this study proposes a multi-energy load forecasting method for IES using an improved VMD-TCN-BiLSTM model. The proposed model consists of optimizing the Variational Mode Decomposition (VMD) parameters through a mathematical model based on minimizing the average permutation entropy (PE). Moreover, load sequences are decomposed into different Intrinsic Mode Functions (IMFs) using VMD, with the optimal number of models determined by the average PE to reduce the non-stationarity of the original sequences. Considering the coupling relationship among electrical, thermal, and cooling loads, the input features of the forecasting model are constructed by combining the IMF set of multi-energy loads with meteorological data and related load information. As a result, a hybrid neural network structure, integrating a Temporal Convolutional Network (TCN) with a Bidirectional Long Short-Term Memory (BiLSTM) network for load prediction is developed. The Sand Cat Swarm Optimization (SCSO) algorithm is employed to obtain the optimal hyper-parameters of the TCN-BiLSTM model. A case analysis is performed using the Arizona State University Tempe campus dataset. The findings demonstrate that the proposed method can outperform six other existing models in terms of Mean Absolute Percentage Error (MAPE) and Coefficient of Determination (R²), verifying its effectiveness and superiority in load forecasting. Full article

The precise forecasting of groundwater levels significantly influences plant growth and the sustainable management of ecosystems. Nonetheless, the non-stationary characteristics of groundwater level data often hinder the current deep learning algorithms from precisely capturing variations in groundwater levels. We used Variational Mode Decomposition (VMD) and an enhanced Transformer model to address this issue. Our objective was to develop a deep learning model called VMD-iTransformer, which aims to forecast variations in the groundwater level. This research used nine groundwater level monitoring stations located in Hangjinqi Ecological Reserve in Kubuqi Desert, China, as case studies to forecast the groundwater level over four months. To enhance the predictive performance of VMD-iTransformer, we introduced a novel approach to model the fluctuations in groundwater levels in the Kubuqi Desert region. This technique aims to achieve precise predictions of the non-stationary groundwater level conditions. Compared with the classic Transformer model, our deep learning model more effectively captured the non-stationarity of groundwater level variations and enhanced the prediction accuracy by 70% in the test set. The novelty of this deep learning model lies in its initial decomposition of multimodal signals using an adaptive approach, followed by the reconfiguration of the conventional Transformer model’s structure (via self-attention and inversion of a feed-forward neural network (FNN)) to effectively address the challenge of multivariate time prediction. Through the evaluation of the prediction results, we determined that the method had a mean absolute error (MAE) of 0.0251, a root mean square error (RMSE) of 0.0262, a mean absolute percentage error (MAPE) of 1.2811%, and a coefficient of determination (R²) of 0.9287. This study validated VMD and the iTransformer deep learning model, offering a novel modeling approach for precisely predicting fluctuations in groundwater levels in a non-stationary context, thereby aiding sustainable water resource management in ecological reserves. The VMD-iTransformer model enhances projections of the water level, facilitating the reasonable distribution of water resources and the long-term preservation of ecosystems, providing technical assistance for ecosystems’ vitality and sustainable regional development. Full article

Displacement prediction models based on measured data have been widely applied in structural health monitoring. However, most models neglect the particularity of displacement monitoring for arch dams with cracks, nor do they thoroughly analyze the non-stationarity and uncertainty of displacement. To address this issue, the influencing factors of displacement were first considered, with crack opening displacement being incorporated into them, leading to the construction of the HSCT model that accounts for the effects of cracks. Feature selection was performed on the factors of the HSCT model utilizing the max-relevance and min-redundancy (mRMR) algorithm, resulting in the screened subset of displacement influence factors. Next, displacement was decomposed into trend, seasonal, and remainder components applying the seasonal-trend decomposition using loess (STL) algorithm. The multifractal characteristics of these displacement components were then analyzed by multifractal detrended fluctuation analysis (MF-DFA). Subsequently, displacement components were predicted employing the convolutional neural network-long short-term memory (CNN-LSTM) model. Finally, the impact of uncertainty factors was quantified using prediction intervals based on the bootstrap method. The results indicate that the proposed methods and models are effective, yielding satisfactory prediction accuracy and providing scientific basis and technical support for the health diagnosis of hydraulic structures. Full article

Many techniques have been proposed to model space-varying observation processes with a nonstationary spatial covariance structure and/or anisotropy, usually on a geostatistical framework. Nevertheless, there is an increasing interest in point process applications, and methodologies that take nonstationarity into account are welcomed. In this sense, this work proposes an extension of a class of spatial Cox process using spatial deformation. The proposed method enables the deformation behavior to be data-driven, through a multivariate latent Gaussian process. Inference leads to intractable posterior distributions that are approximated via MCMC. The convergence of algorithms based on the Metropolis–Hastings steps proved to be slow, and the computational efficiency of the Bayesian updating scheme was improved by adopting Hamiltonian Monte Carlo (HMC) methods. Our proposal was also compared against an alternative anisotropic formulation. Studies based on synthetic data provided empirical evidence of the benefit brought by the adoption of nonstationarity through our anisotropic structure. A real data application was conducted on the spatial spread of the Spodoptera frugiperda pest in a corn-producing agricultural area in southern Brazil. Once again, the proposed method demonstrated its benefit over alternatives. Full article