Network Units Research Articles

A MILP-based power system parallel restoration model with the support of mobile energy storage systems

During bulk power system restoration, black-start (BS) units are used to crank non-BS units in such a way that the over system generation capability is maximized. However, traditional power system restoration methods rely on stationary BS units, whose number is limited due to high investment and geographic conditions, impeding a fast system restoration. To enhance restoration efficiency, this paper proposes an integrated power system parallel restoration method considering the support of mobile energy storage systems (MESSs), which can provide cranking power to non-BS units and transform them into MESS-assisted BS units. MESSs are routed and scheduled via transportation network, along with restoration planning in power network by integrating them into a mixed integrated linear programming model. Specifically, to describe the interactive behavior between MESSs and the power system, a mobility model for MESSs is proposed, which indicates the travel path, travel time and connection nodes. Then, an integrated optimization model involving BS unit location, zone partitioning, network reconfiguration and generator start-up is formulated. The proposed method was simulated in the modified IEEE 30-bus transmission system coupled with 11-node transportation system. The results show that, in contrast with a decoupled restoration method, the proposed method increases system generation capability by 6.3%.

Machine learning Ethereum cryptocurrency prediction and knowledge-based investment strategies

This work proposes a novel methodology to help in decision making in the cryptocurrency market. Two investment strategies have been designed for Ethereum (ETH), based on predictions of the price and trend of this cryptocurrency using real data. The two Ethereum cryptocurrency prediction systems rely solely on past values of other contextual stock indices, market indicators and online trends, and ignore any technical indicators of price evolution. Real data from cryptocurrency market has been collected and processed with different feature selection methods. Applying a regression approach with Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) networks, prediction models for the ETH price for 1, 7 and 15 days are obtained and compared. Also, support vector machine (SVM) is applied to predict the ETH price trend by applying a classification approach. In both approaches, sentiment analysis has been included to check its effect on the prediction results. The reliability of these prediction models in the current market has been evaluated by designing two original knowledge-based investment strategies. They are tested over two different time periods with real cryptocurrency market data. The results show that it is possible to generate up to 5.16 profit factor with few operations using these models. Furthermore, adding sentiment analysis has shown to have little influence. In this way, we contribute to the advancement of our knowledge of this volatile and still young cryptocurrency market, and specifically of the evolution of Ethereum and the factors that can influence its behavior.

Performance enhancement in hydroponic and soil compound prediction by deep learning techniques

The soil quality plays a crucial role in providing essential nutrients for crop growth and ensuring a bountiful yield. Identifying the soil composition, which includes sand, silt particles, and the mixture of clay in specific proportions, is vital for making informed decisions about crop selection and managing weed growth. Furthermore, soil pollution from emerging contaminants presents a substantial risk to water resource management and food production. Developing numerical models to comprehensively describe the transport and reactions of chemicals within both the plants and soil is of utmost importance in crafting effective mitigation strategies. To address the limitations of traditional models, this paper devises an innovative approach that leverages deep learning to predict hydroponic and soil compound dynamics during plant growth. This method not only enhances the understanding of how plants interact with their environment but also aids in making more informed decisions about agriculture, ultimately contributing to more sustainable and efficient crop production. The data needed to perform the developed hydroponic and soil compound prediction model is acquired from online resources. After that, this data is forwarded to the feature extraction phase. The weighted features, deep belief network (DBN) features, and the original features are achieved in the feature extraction stage. To get the weighted features, the weights are optimally obtained using the Iteration-assisted Enhanced Mother Optimization Algorithm (IEMOA). Subsequently, these extracted features are fed into the Multi-Scale feature fusion-based Convolution Autoencoder with a Gated Recurrent Unit (MS-CAGRU) network for hydroponic and soil compound prediction. Thus, the hydroponic and soil compound prediction data is attained in the end. Finally, the performance evaluation of the suggested work is conducted and contrasted with numerous conventional models to showcase the system’s efficacy.

A data-driven online calibration method for enhancing accuracy in biomedical tactile sensing

Soft tactile sensors have been used to acquire biomedical signals for a variety of applications. Lack of proper calibration can lead to data discrepancies and, in some cases, misdiagnoses. This study introduces a channel-specific calibration technique using an enhanced stacked Simple Recurrent Unit (SRU) network for real-time calibration of each sensor channel. The network is designed to correct deviations due to sensor characteristics or operational pressures through a dynamic weighting mechanism that adjusts to each channel’s energy level. A pneumatic-controlled platform was developed to ensure quality training data for the deep learning model, facilitating a robust public dataset. For pulse wave signals, our approach significantly outperformed existing methods, reducing the Root Mean Square Error (RMSE) by 68.21% over raw signals and 25.68% compared to LSTM-based calibration. It also improved the accuracy of radial augmentation index (radial-AIx) measurements from −3.19% to −0.03%, making biomedical sensors more reliable.

Interference Management for a Wireless Communication Network Using a Recurrent Neural Network Approach

Wireless communication technologies have profoundly impacted the interconnectivity of mobile users and terminals. Nevertheless, the exponential increase in the number of users poses significant challenges, particularly in interference management, which is a major concern in wireless communication. Machine learning (ML) approaches have emerged as powerful tools for solving various problems in this domain. However, existing studies have not fully addressed the problem of interference management for wireless communication using ML techniques. In this paper, we explore the application of recurrent neural network (RNN) approaches to address co-channel interference in wireless communication. Specifically, we investigate the effectiveness of long short-term memory (LSTM), bidirectional LSTM (Bi-LSTM), and gated recurrent unit (GRU) network architectures in two different network settings. The first network comprises 10 connected devices, while the second network involves 20 devices. Our experimental results demonstrate that Bi-LSTM outperforms LSTM and GRU in terms of mean squared error, normalized mean squared error, and sum rate. While LSTM and GRU produce similar results, LSTM exhibits a marginal advantage over GRU. In addition, a combined RNN approach is also studied, and it can provide better results in dense networks.

Energy-Efficient Edge and Cloud Image Classification with Multi-Reservoir Echo State Network and Data Processing Units.

In an era dominated by Internet of Things (IoT) devices, software-as-a-service (SaaS) platforms, and rapid advances in cloud and edge computing, the demand for efficient and lightweight models suitable for resource-constrained devices such as data processing units (DPUs) has surged. Traditional deep learning models, such as convolutional neural networks (CNNs), pose significant computational and memory challenges, limiting their use in resource-constrained environments. Echo State Networks (ESNs), based on reservoir computing principles, offer a promising alternative with reduced computational complexity and shorter training times. This study explores the applicability of ESN-based architectures in image classification and weather forecasting tasks, using benchmarks such as the MNIST, FashionMnist, and CloudCast datasets. Through comprehensive evaluations, the Multi-Reservoir ESN (MRESN) architecture emerges as a standout performer, demonstrating its potential for deployment on DPUs or home stations. In exploiting the dynamic adaptability of MRESN to changing input signals, such as weather forecasts, continuous on-device training becomes feasible, eliminating the need for static pre-trained models. Our results highlight the importance of lightweight models such as MRESN in cloud and edge computing applications where efficiency and sustainability are paramount. This study contributes to the advancement of efficient computing practices by providing novel insights into the performance and versatility of MRESN architectures. By facilitating the adoption of lightweight models in resource-constrained environments, our research provides a viable alternative for improved efficiency and scalability in modern computing paradigms.

Benefits of flexibility in current and future IM-DD based TDM-PON [Invited

Due to continuously emerging high bandwidth applications, research and standardization of time division multiplexed passive optical networks (TDM-PONs) have focused on increasing the peak bitrate. However, increasing the bitrate while supporting the stringent optical power budget of a PON becomes increasingly challenging because of the larger chromatic dispersion penalties as well as reduced receiver sensitivity when the bitrate of the intensity modulation with direct-detection (IM-DD) based PON is increased. Also, increasing bitrate generally causes higher power consumption, which leads to more challenging thermal designs and misalignment with environmental targets. In this paper we give an overview of flexible concepts that can help achieve the required optical power budget and support reduced power consumption of a future IM-DD based TDM-PON. We demonstrate that a flexible PON can provide an increased overall throughput or an extended reach and power budget with the use of flexible modulation formats, probabilistic and geometric shaping, and flexible rate forward error correction (FEC). Another dimension of flexibility in the form of a configurable optical distribution network (ODN) is described and its merits and challenges are discussed. Flexible concepts based on interleaving of FEC codewords can align signal processing like FEC decoding and the protocol processing closer to the user-rate of an optical network unit (ONU), which leads to reduced power consumption. Flexibility based on multiple channels based on wavelength multiplexing or spatial multiplexing enables optimization of the power consumption to the amount of traffic on the PON. Several flexible concepts have already been adopted in the PON standards. We highlight flexible gain FEC for upstream 50G PON, the transmitter dispersion eye closure (TDEC) metric, and the flexible split-ratio ODN for power saving. Flexible modulation has not been adopted yet in PON standards, but it is expected that flexibility is more and more needed to support the performance, cost effectiveness, and power conservation of future optical access systems.

Well Logging Reconstruction Based on a Temporal Convolutional Network and Bidirectional Gated Recurrent Unit Network with Attention Mechanism Optimized by Improved Sand Cat Swarm Optimization

Geophysical logging plays a very important role in reservoir evaluation. In the actual production process, some logging data are often missing due to well wall collapse and instrument failure. Therefore, this paper proposes a logging reconstruction method based on improved sand cat swarm optimization (ISCSO) and a temporal convolutional network (TCN) and bidirectional gated recurrent unit network with attention mechanism (BiGRU-AM). The ISCSO-TCN-BiGRU-AM can process both past and future states efficiently, thereby extracting valuable deterioration information from logging data. Firstly, the sand cat swarm optimization (SCSO) improved by the variable spiral strategy and sparrow warning mechanism is introduced. Secondly, the ISCSO’s performance is evaluated using the CEC–2022 functions and the Wilcoxon test, and the findings demonstrate that the ISCSO outperforms the rival algorithms. Finally, the logging reconstruction method based on the ISCSO-TCN-BiGRU-AM is obtained. The results are compared with the competing models, including the back propagation neural network (BPNN), GRU, and BiGRU-AM. The results show that the ISCSO-TCN-BiGRU-AM has the best performance, which verifies its high accuracy and feasibility for the missing logging reconstruction.

Nirsevimab Effectiveness Against Cases of Respiratory Syncytial Virus Bronchiolitis Hospitalised in Paediatric Intensive Care Units in France, September 2023-January 2024.

In September 2023, France was one of the first countries that started a national immunisation campaign with nirsevimab, a new monoclonal antibody against respiratory syncytial virus (RSV). Using data from a network of paediatric intensive care units (PICUs), we aimed to estimate nirsevimab effectiveness against severe cases of RSV bronchiolitis in France. We conducted a case-control study based on the test-negative design and included 288 infants reported by 20 PICUs. We estimated nirsevimab effectiveness at 75.9% (48.5-88.7) in the main analysis and 80.6% (61.6-90.3) and 80.4% (61.7-89.9) in two sensitivity analyses. These real-world estimates confirmed the efficacy observed in clinical studies.

Mortality and extrauterine growth restriction of necrotizing enterocolitis in very preterm infants with heart disease: a multi-center cohort study.

Congenital heart disease (CHD) and patent ductus arteriosus (PDA) are risk factors of necrotizing enterocolitis (NEC) in infants. However, it is unclear whether the prognosis of NEC is different between very preterm infants (VPIs) with and without heart diseases. This was an observational cohort study that enrolled VPIs (born between 24+0 and 31+6weeks) admitted to 79 tertiary neonatal intensive care units (NICU) in the Chinese Neonatal Network (CHNN) between 2019 and 2021. The exposure was CHD or isolated PDA, and VPIs with NEC were divided into three groups: complicated with CHD, with isolated PDA, and without heart diseases. The primary outcomes were NEC-related adverse outcomes (death or extrauterine growth restriction (EUGR)). Logistic regression models were used to adjust potential confounders and calculate the odds ratios (ORs) and 95% confidential intervals (CIs) for each outcome. A total of 1335 VPIs with NEC were enrolled in this study, including 65 VPIs with CHD and 406 VPIs with isolated PDA. The VPIs with heart diseases had smaller gestational ages and lower body weights at birth, more antenatal steroids use, and requiring inotrope prior to the onset of NEC. While suffering from NEC, there was no significant increased risks in NEC-related death in VPIs with either CHD (adjusted OR [aOR]: 1.10; 95% CI: 0.41-2.50) or isolated PDA (aOR: 1.25; 95% CI 0.82-1.87), and increased risks in EUGR were identified in either survival VPIs with CHD (aOR: 2.35; 95% CI: 1.31-4.20) or isolated PDA (aOR: 1.53; 95% CI: 1.16-2.01) in survivors. The composite outcome (death or EUGR) was also more often observed in VPIs with either CHD (aOR: 2.07; 95% confidence interval [CI]: 1.20-3.60) or isolated PDA (aOR: 1.51; 95% CI: 1.17-1.94) than that without heart diseases. VPIs with either CHD or isolated PDA were associated with significantly prolonged duration of fasting, extended time to achieve full enteral feeding, and longer ventilation duration and hospitalization duration. Similar characteristics were also seen in VPIs with isolated PDA, with the exception that VPIs with CHD are more likely to undergo surgical intervention and maintain a prolonged fast after NEC. Conclusion: In VPIs with NEC, CHD and isolated PDA are associated with an increased risk in worse outcomes. We recommend that VPIs with cardiac NEC be managed with aggressive treatment and nutrition strategies to prevent EUGR. What is Known: • CHD and PDA are risk factors for NEC in infants, which can lead to adverse outcomes such as death and EUGR. • NEC in infants with heart disease differs clinically from that in infants without heart disease and should be recognized as a separate disease process. What is New: • CHD and isolated PDA are associated with increased risks of EUGR in VPIs with NEC. • Risk factors associated with VPIs with cardiac NEC suggested these patients should be managed with aggressive treatment and nutrition strategies to adverse outcomes.

Early- and late-onset candidemia in very low birth weight infants in the Korean neonatal network, 2013–2017

BackgroundCandidiasis is a critical infection that is associated with very low birth weight (VLBW; <1500 g). This study investigated the characteristics and clinical presentation of candidiasis in Korean VLBW infants according to the onset of candidemia. MethodsAll VLBW infants with candidemia, defined as blood culture-positive candidiasis and registered in a multicenter database with data from 70 neonatal units of the Korean Neonatal Network between 2013 and 2017, were included in this study. Early-onset candidemia (EOC; ≤10 days) and late-onset candidemia (LOC; >10 days) were analyzed. The demographic characteristics, clinical presentations, and outcomes of candidemia were also determined. ResultsThe overall incidence of candidemia was 2% (209/10,397) and 4% (173/3934) in VLBW and extremely very low birth weight (ELBW; <1000 g) infants, respectively. In ELBW infants, gestational age was significantly younger at EOC than at LOC (P = 0.015). Cesarean section, respiratory distress syndrome, severe bronchopulmonary disease, pulmonary hemorrhage, prior-bacteremia, neonatal seizures, and periventricular leukomalacia were significantly more common in the LOC group than in the EOC group (P < 0.05). The duration of invasive ventilation, total parenteral nutrition, and hospital stay were significantly longer in the LOC group than in the EOC group (P < 0.05). Most infections were caused by Candida spp. (91.8%). The mortality rate of ELBW infants with candidemia was 41%, which was higher than that of those without candidemia (29%) (P < 0.001). Mortality due to infection was also higher in infants with candidemia (55%) than in those without candidemia (15%) (P < 0.001); however, there were no significant differences between the EOC and LOC groups. ConclusionsLOC was more common than EOC in VLBW infants. Considering the risk factors of LOC, active weaning from invasive ventilators and aggressive enteral feeding are required to decrease LOC. Furthermore, preventing candidemia is necessary to reduce mortality in VLBW infants.

Differential protection of distribution lines based on Hausdorff algorithm

Abstract Given the weak time synchronization capability of the feeder terminal unit (FTU) in the distribution network, a new differential protection method for the distribution network is proposed. The method uses the abrupt change of the phase current as the starting criterion, then sends a phase current sampling sequence of power frequency period to the associated FTU, and determines whether the fault occurs in the section by calculating the Hausdorff distance between the sample sequences of the associated FTU current. This algorithm needs no voltage information and the calculation amount is small, making it suitable for multi-branch lines and accurately identifying faults inside and outside the area. A large number of simulation experiments show that the protection scheme has a 4 ms anti-synchronization error capability and is basically unaffected by transition resistance, which is suitable for active distribution networks with high penetration of distributed power supply.

Collision avoidance decision-making strategy for multiple USVs based on Deep Reinforcement Learning algorithm

—To address the issue of flexible decision-making for Unmanned Surface Vehicles (USVs) in emergency collision avoidance scenarios, this paper proposes a Collision Avoidance Decision-making Strategy (CADMS) for multiple USVs based on Deep Reinforcement Learning (DRL). The novelty of this method is: (1) The Preferential Experience Replay Mechanism (PERM) and the Gate Recurrent Unit (GRU) network are employed to enhance the Deep Deterministic Policy Gradient (DDPG) algorithm. The PERM facilitates faster convergence by increasing the sampling probability, thereby aiding in data assimilation. Meanwhile, the GRU enables self-assessment of sensor input significance, leading to more accurate action prediction. (2) In the design of the reward function, the departure from COLREGs has been fully considered, including the two-ship encounter situations and the multi-ship encounter situations. This study also provides a comprehensive analysis of neural network training and generalization verification. The experimental results demonstrate the effectiveness of the proposed approach in effectively avoiding collisions with multiple USVs in complex water environments. It is noteworthy that in comparison to conventional collision avoidance methods, the proposed CADMS model exhibits a superior collision avoidance success rate and enables safer and more efficient decision-making behaviors when confronted with intricate encounter situations.

Translating knowledge into policy: Organizational model and minimum requirements for the implementation of a regional pancreas unit network

Pancreatic and periampullary cancers pose significant challenges in oncological care due to their complexity and diagnostic difficulties. Global experiences underscore the crucial role of multidisciplinary collaboration and centralized care in improving patient outcomes in this context. Recognizing these challenges, Lombardy, Italy's most populous region, embarked on establishing pancreas units across its territory to enhance clinical outcomes and organizational efficiency. This initiative, driven by a multistakeholder approach involving the Lombardy Welfare Directorate, clinicians, and a patient association, emphasizes the centralization of complex care in high-volume hospitals, adopting a hub-and-spoke model and a multidisciplinary approach. This article outlines the process and criteria set forth for pancreas unit implementation, aiming to provide a structured framework for enhancing pancreatic cancer care. Central to this initiative is the establishment of structured criteria and minimal requirements, not only for surgery but also for other essential components of care, ensuring a comprehensive approach to pancreatic cancer management. The Lombardy model offers a structured framework for enhancing pancreatic cancer care, with potential applicability to other regions and countries seeking to improve their cancer care infrastructure

URL and Malicious Link Prediction

Abstract: In response to the pressing cybersecurity challenges posed by the proliferation of phishing URLs and malicious links, this research introduces a ground breaking approach centered on transfer learning within deep neural networks. By leveraging transfer learning, intricate patterns within URLs and their content are unveiled, culminating in the development of a model seamlessly integrating Bidirectional Long Short-Term Memory (BiLSTM) and Bidirectional Gated Recurrent Unit (BiGRU) networks. These architectures effectively capture sequential dependencies, enhanced by their bidirectional variants accessing both past and future states to comprehend temporal dynamics and improve performance. Through meticulous evaluation and fine-tuning processes, the proposed cybersecurity solution demonstrates robustness and efficacy in defending against evolving threats. This research contributes significantly to advancing the cybersecurity domain, introducing an adaptive strategy that harnesses the strengths of BiLSTM and BiGRU networks within the framework of transfer learning, thus paving the way for more resilient and effective cybersecurity solutions.

Automatic Modulation Recognition Method Based on Phase Transformation and Deep Residual Shrinkage Network

Automatic Modulation Recognition (AMR) is currently a research hotspot, and research under low Signal-to-Noise Ratio (SNR) conditions still poses certain challenges. This paper proposes an AMR method based on phase transformation and deep residual shrinkage network to improve recognition accuracy. Firstly, the raw I/Q data from the benchmark dataset RML2016.10a are used as the input. Then, an end-to-end modulation recognition is performed using the model. Phase transformation is used to correct the raw I/Q data and reduce the interference of phase shift on modulation recognition. Convolutional neural network (CNN) and Gate Recurrent Unit (GRU) extract the spatial and temporal features of the modulation signal, respectively. The improved deep residual shrinkage network is added after CNN to eliminate unimportant features through soft thresholding. Finally, the proposed model is trained and tested. The experimental results show that the proposed model notably reduces the number of parameters compared to other models, effectively improving the recognition accuracy under low SNR conditions. The average recognition accuracy reaches 62.46%, and the highest recognition accuracy reaches 92.41%.

A cascaded GRU-based stereoscopic matching network for precise plank measurement

Wooden plank images in industrial measurements often contain numerous textureless areas. Furthermore, due to the thin plate structure, the three-dimensional (3D) disparity of these planks is predominantly confined to a narrow range. Consequently, achieving accurate 3D matching of wooden plank images has consistently presented a challenging task within the industry. In recent years, deep learning has progressively supplanted traditional stereo matching methods due to its inherent advantages, including rapid inference and end-to-end processing. Nonetheless, the acquisition of datasets for stereo matching networks poses an additional challenge, primarily attributable to the difficulty in obtaining accurate disparity data. Thus, this paper presents a novel stereo matching method incorporating three key innovations. Firstly, an enhanced gated recurrent unit network is introduced, accompanied by a redesigned structure to achieve higher matching accuracy. Secondly, an efficient preprocessing module is proposed, aimed at improving the algorithm’s efficiency. Lastly, in response to the challenges posed by datasets acquisition, we innovatively employed image simulation software to obtain a high-quality simulated dataset of wooden planks. To assess the feasibility of our approach, we conducted both simulated and real experiments. The experiments results clearly exhibit the superiority of our method when compared to existing approaches in terms of both stability and accuracy. In the simulation experiment, our method attained a bad1.0 score of 2.1% (compared to the baseline method’s 9.76%); In the real experiment, our method achieved an average error of 0.104 mm (compared to the baseline method’s 0.268 mm). It is worth noting that our study aims to address the challenge of acquiring datasets for deep learning and bridging the gap between simulated and real data, resulting in increased applicability of deep learning in more industrial measurement domains.

Performance and energy efficiency enhancement of existing optical communication systems by incorporating resource allocation on demand technique in FiWi networks

Abstract Fiber-wireless (FiWi) networks have begun to be fully integrated into contemporary optical access network systems as a result of utilizing the combined advantages of wired and wireless techniques. Passive optical networks (PON) are heavy energy consumers and one of the main contributors to greenhouse gas emissions (GHG) that are a result of climate change. These systems also present economic challenges. In this work, three different FiWi systems are presented such as (i) even/odd active transmitters with only free space optical (FSO) based optical network units (ONUs)/with dual channel option such as distribution fiber (DF)/FSO supported ONUs, (ii) a novel resource allocation on demand (RAoD) incorporated bidirectional FiWi system supporting DF/FSO based ONUs and four wave mixing (FWM) generated upstream wavelengths and (iii) transmitter diversity enabled 5G supported FiWi system supporting bidirectional communication. It is observed that the energy efficiency of 26.04 % is obtained in even/odd active transmitter and only FSO supported ONUs and under the harsh weather conditions, by adopting DF channel instead of FSO, 30 % energy efficiency further added. However, in novel RAoD technique, energy efficiencies for optical distribution networks (ODN1), ODN2, ODN3, and ODN4 are 48.52 %, 46.4 %, 36.96 %, and 44.31 %. Further, DF option in channel selection improve 30 % more energy saving. In performance enhanced transmitter diversity employed system, at 60 km reduction in bit error rate (BER) are 25 % for ODN1, 5.88 % for ODN2, 20 % for ODN3, and 12.5 % for ODN3.

Oral cancer diagnosis based on gated recurrent unit networks optimized by an improved version of Northern Goshawk optimization algorithm

Oral cancer early diagnosis is a critical task in the field of medical science, and one of the most necessary things is to develop sound and effective strategies for early detection. The current research investigates a new strategy to diagnose an oral cancer based upon combination of effective learning and medical imaging. The current research investigates a new strategy to diagnose an oral cancer using Gated Recurrent Unit (GRU) networks optimized by an improved model of the NGO (Northern Goshawk Optimization) algorithm. The proposed approach has several advantages over existing methods, including its ability to analyze large and complex datasets, its high accuracy, as well as its capacity to detect oral cancer at the very beginning stage. The improved NGO algorithm is utilized to improve the GRU network that helps to improve the performance of the network and increase the accuracy of the diagnosis. The paper describes the proposed approach and evaluates its performance using a dataset of oral cancer patients. The findings of the study demonstrate the efficiency of the suggested approach in accurately diagnosing oral cancer.

Monthly Runoff Prediction for Xijiang River via Gated Recurrent Unit, Discrete Wavelet Transform, and Variational Modal Decomposition

Neural networks have become widely employed in streamflow forecasting due to their ability to capture complex hydrological processes and provide accurate predictions. In this study, we propose a framework for monthly runoff prediction using antecedent monthly runoff, water level, and precipitation. This framework integrates the discrete wavelet transform (DWT) for denoising, variational modal decomposition (VMD) for sub-sequence extraction, and gated recurrent unit (GRU) networks for modeling individual sub-sequences. Our findings demonstrate that the DWT–VMD–GRU model, utilizing runoff and rainfall time series as inputs, outperforms other models such as GRU, long short-term memory (LSTM), DWT–GRU, and DWT–LSTM, consistently exhibiting superior performance across various evaluation metrics. During the testing phase, the DWT–VMD–GRU model yielded RMSE, MAE, MAPE, NSE, and KGE values of 245.5 m3/s, 200.5 m3/s, 0.033, 0.997, and 0.978, respectively. Additionally, optimal sliding window durations for different input combinations typically range from 1 to 3 months, with the DWT–VMD–GRU model (using runoff and rainfall) achieving optimal performance with a one-month sliding window. The model’s superior accuracy enhances water resource management, flood control, and reservoir operation, supporting better-informed decisions and efficient resource allocation.