Reduction In Mean Square Error Research Articles

Generating context-specific sports training plans by combining generative adversarial networks.

Personalized sports training plans are essential for addressing individual athlete needs, but traditional methods often need to integrate diverse data types, limiting adaptability and effectiveness. Existing machine learning (ML) and rule-based approaches cannot dynamically generate context-specific training programs, reducing their applicability in real-world scenarios. This study aims to develop a Generative Adversarial Network (GAN)- based framework to create context-specific training plans by integrating numeric attributes (e.g., age, heart rate) and motion features from video data. The research focuses on improving context-specific efficiency and real-time adaptability while addressing the limitations of traditional methods. The proposed GAN framework combines numeric and motion features using a generator-discriminator architecture to produce tailored training plans. The model is evaluated quantitatively through metrics like mean square error (MSE) and generation time and qualitatively through subjective ratings from athletes and coaches using a five-point Likert scale for context-specific, scientificity, applicability, and feasibility. Statistical significance is analyzed using ANOVA testing. The proposed GAN model outperforms traditional ML and rule-based methods, achieving a 22% reduction in MSE and a 45% improvement in generation time. Subjective evaluations reveal significant improvements in context-specific and applicability, with ratings averaging 4.8/5 compared to 3.9/5 for baseline models. The GAN framework effectively integrates multimodal data, demonstrating dynamic adaptability and high efficiency suitable for real-world applications. The proposed GAN-based framework advances the generation of personalized sports training plans by integrating numeric and motion data, achieving superior adaptability and efficiency. These results highlight the model's potential for practical deployment in athletic coaching systems, addressing critical gaps in existing methodologies and offering scalable solutions for individualized training.

Short-term Load Forecasting Method Based on VNCMD-TiDE Model

Background: Extreme weather conditions exert a considerable influence on power load, leading to increasingly erratic fluctuations. Consequently, the dependable and precise forecasting of power load assumes paramount importance in power system planning. Objective: Given the inadequacy of traditional forecasting approaches in handling long-term series load forecasting, this paper introduces a short-term power load forecasting model rooted in VNCMD-TiDE, aiming to enhance forecasting precision. Method: Initially, the XGBoost algorithm is employed to perform nonlinear coupling analysis between load and meteorological data, identifying crucial features. Following this, the VNCMD method is utilized to handle the nonlinear and non-stationary load data, decomposing it into multiple components with distinct frequencies. Building upon this decomposition, a TiDE-Bayesian model is constructed, wherein the decomposed components serve as inputs for prediction. Simultaneously, Bayesian optimization is leveraged to fine-tune hyperparameters. Ultimately, the prediction outcomes of each component are amalgamated to derive the final prediction. Results: The proposed model's performance is assessed through comparison with traditional machine learning models, such as LSTM. Achieving a noteworthy reduction in Root Mean Square Error (RMSE) by 4.14 underscores its exceptional predictive prowess. Conclusion: Through the analysis of actual power load data in a specific location, the model proposed in this article demonstrates superior prediction accuracy, particularly evident during extreme weather conditions like snow and rain.

The artificial intelligence-based agricultural field irrigation warning system using GA-BP neural network under smart agriculture.

This work explores an intelligent field irrigation warning system based on the Enhanced Genetic Algorithm-Backpropagation Neural Network (EGA-BPNN) model in the context of smart agriculture. To achieve this, irrigation flow prediction in agricultural fields is chosen as the research topic. Firstly, the BPNN principles are studied, revealing issues such as sensitivity to initial values, susceptibility to local optima, and sample dependency. To address these problems, a genetic algorithm (GA) is adopted for optimizing the BPNN, and the EGA-BPNN model is used to predict irrigation flow in agricultural fields. Secondly, the EGA-BPNN model can overcome the local optimization and overfitting problems of traditional BPNN through the global search ability of GA. Moreover, it is suitable for the irrigation flow prediction task with complex environmental factors in smart agriculture. Finally, comparative experiments compare the prediction accuracy of BPNN and EGA-BPNN using single and dual water level flow prediction models respectively. The results reveal that as the number of nodes in the hidden layer increases, the model's Mean Squared Error (MSE) and Relative Error (RE) show a decreasing trend, indicating an improvement in model prediction accuracy. When the number of nodes in the hidden layer increases from 6 to 16, the MSE of the single and dual water level flow prediction models decreases from 4.53×10-4 to 3.68×10-4 and 2.38×10-4 to 1.66×10-4, respectively. Under a standalone BPNN, the absolute relative error in flow prediction is 1.09%. In contrast, the EGA-BPNN model achieves a significantly lower mean absolute relative error of 0.41% for single-flow prediction, demonstrating superior prediction performance. Furthermore, compared to the BPNN, the EGA-BPNN model exhibits a 2.11 reduction in MSE, further emphasizing the positive impact of introducing the GA on model performance. The research outcomes contribute to more accurate water resource planning and management, providing a more reliable basis for decision-making.

Causal inference of whole-grain foods' risk based on a generative adversarial network and Bayesian network.

Whole-grain foods (WGFs) constitute a large part of humans' daily diet, making risk identification of WGFs important for health and safety. However, existing research on WGFs has paid more attention to revealing the effects of a single hazardous substance or various hazardous substances on food safety, neglecting the mutual influence between individual hazardous substances and between hazardous substances and basic information. Therefore, this paper proposes a causal inference of WGFs' risk based on a generative adversarial network (GAN) and Bayesian network (BN) to explore the mutual influence between hazardous substances and basic information. The experiment results show that the proposed GAN outperformed several traditional data-imputation methods, producing at least a 13.65% reduction of the root mean square error (RMSE). The classification accuracy of the BN model reached 91%. In conclusion, we distinguish the provinces, periods, food categories, and hazardous substances cause the absolute risk of WGFs and the high risk of mycotoxins and compounds (MaCs) and cadmium. PRACTICAL APPLICATION: This research can be applied to impute missing values for whole-grain foods (WGFs) sampling data, and explore the causality among hazardous substances themselves, that between hazardous substances and basic information in WGFs. Additionally, it can be applied to infer root cause of existing or potential WGFs risk (e.g., provinces, periods, food categories, and hazardous substances).

Metamodeling of turbofan engine compressor characteristics using SVM-based improved Kriging method

PurposeThe purpose of this paper is to overcome the inherent lack of precision in commonly used interpolation procedures when solving the mathematical model of turbofan engines, as well as to address the issue that the theoretical variogram model in traditional Kriging models is prone to subjective selection bias, which makes it impossible to accurately capture the inherent fluctuation patterns in compressor data.Design/methodology/approachTo mitigate this challenge, based on the spatial distribution characteristics of the compressor characteristic data of a certain type of turbofan engine, the input and output dimensions of the model are defined. By determining the stable operating region from the original component data, the authors use the proposed Kriging method improved with a support vector machine model to reconstruct the characteristics at unknown speeds within this region. The effectiveness of the proposed method is evaluated using the established assessment metrics.FindingsExperimental results demonstrate that the proposed method exhibits significant advantages over the conventional Kriging approach. Specifically, it leads to a substantial reduction in root mean square error and mean absolute error by 0.0153/0.0118 (low speed), 0.1306/0.0362 (medium speed) and −0.0066/0.2366 (high speed).Originality/valueThis refined approach not only offers notable engineering applicability but also contributes significantly to the enhancement of aerospace engine model solutions’ precision.

Based on the N2N-SAMP for sparse underwater acoustic channel estimation

IntroductionOrthogonal Frequency Division Multiplexing (OFDM) is widely recognized for its high efficiency in modulation techniques and has been extensively applied in underwater acoustic communication. However, the unique sparsity and noise interference characteristics of the underwater channel pose significant challenges to the performance of traditional channel estimation methods.MethodsTo address these challenges, we propose a sparse underwater channel estimation method that combines the Noise2Noise (N2N) algorithm with the Sparsity Adaptive Matching Pursuit (SAMP) algorithm. This novel approach integrates the N2N technique from image denoising theory with the SAMP algorithm, utilizing a constant iteration termination threshold that does not require prior information. The method leverages the U-net neural network structure to denoise noisy pilot signals, thereby restoring channel sparsity and enhancing the accuracy of channel estimation.ResultsSimulation results indicate that our proposed method demonstrates commendable channel estimation performance across various signal-to-noise ratio (SNR) conditions. Notably, in low SNR environments, the N2N-SAMP algorithm significantly outperforms the traditional SAMP algorithm in terms of Mean Squared Error (MSE) and Bit Error Rate (BER). Specifically, at SNR levels of 0 dB, 10 dB, and 20 dB, the MSE of channel estimation is reduced by 58.95%, 76.08%, and 19.42%, respectively, compared to the SAMP algorithm that selects the optimal threshold based on noise power. Furthermore, the system’s BER is decreased by 12.35%, 26.41%, and 29.62%, respectively.DiscussionThe findings suggest that the integration of N2N and SAMP algorithms offers a promising solution for improving channel estimation in underwater communication channels, especially under low SNR conditions. The significant reduction in MSE and BER highlights the effectiveness of our proposed method in enhancing the reliability and accuracy of underwater communication systems.

Personalized dental crown design: A point-to-mesh completion network.

Designing dental crowns with computer-aided design software in dental laboratories is complex and time-consuming. Using real clinical datasets, we developed an end-to-end deep learning model that automatically generates personalized dental crown meshes. The input context includes the prepared tooth, its adjacent teeth, and the two closest teeth in the opposing jaw. The training set contains this context, the ground truth crown, and the extracted margin line. Our model consists of two components: First, a feature extractor converts the input point cloud into a set of local feature vectors, which are then fed into a transformer-based model to predict the geometric features of the crown. Second, a point-to-mesh module generates a dense array of points with normal vectors, and a differentiable Poisson surface reconstruction method produces an accurate crown mesh. Training is conducted with three losses: (1) a customized margin line loss; (2) a contrastive-based Chamfer distance loss; and (3) a mean square error (MSE) loss to control mesh quality. We compare our method with our previously published method, Dental Mesh Completion (DMC). Extensive testing confirms our method's superiority, achieving a 12.32% reduction in Chamfer distance and a 46.43% reduction in MSE compared to DMC. Margin line loss improves Chamfer distance by 5.59%.

Development of the CMA-ChemRA: China Regional Weakly Coupled Chemical-Weather Reanalysis System with product since 2007.

The CMA-ChemRA (China Regional Weakly Coupled Chemical-Weather Reanalysis System) was developped using China's first-generation global atmospheric reanalysis product (CRA-40) as initial fields and boundary conditions, coupled with the WRF-Chem atmospheric chemical model and the WRFDA/3DVar assimilation system. By constructing a joint background error covariance matrix, CMA-ChemRA achieves weak coupling between atmospheric chemistry and meteorological variables, enabling simultaneous assimilation of diverse data sources, including hourly observations from ground stations, wind profilers, upper-air soundings, aircraft reports, and atmospheric composition measurements. To extend the dataset to periods before 2013 when China lacked PM2.5 observations, the system incorporates a reconstructed PM2.5 dataset derived by AI from visibility inversion alongside various emission inventories. The CMA-ChemRA system produces a reanalysis product from 2007 to the present, with a spatial resolution of 15km and an hourly temporal resolution. It includes three-dimensional isobaric and near-surface layers for 6 key elements PM2.5, PM10, O3, SO2, NO2, and CO, as well as meteorological variables. This product is updated in near real-time, with a 50-min lag for forecast updates. Evaluation of the system shows substantial improvements in accuracy, with significant reductions in root mean square error (RMSE) for the six elements in the near-surface atmospheric layer post-assimilation. The model's depiction of ground-level PM2.5 concentrations aligns well with independent observational data across five urban regions, showing a narrow RMSE range of 15.5 to 32.8μg/m3. Additionally, CMA-ChemRA demonstrates strong performance in capturing the evolution of dust storms and pollution events, particularly in accurately modeling PM2.5 concentrations during severe pollution episodes. Our innovative approach in constructing a joint background error covariance matrix and the resulting high-resolution, real-time updating CMA-ChemRA product. This represents significant advancement in the field of atmospheric and chemical weather reanalysis. The product serves as an crucial tool for environmental monitoring and forecasting in China.

Deep learning for predicting porosity in ultra-deep fractured vuggy reservoirs from the Shunbei oilfield in Tarim Basin, China

Deep and ultra-deep carbonate reservoirs in China, which account for 34% of the country’s oil and gas reserves, pose significant challenges for porosity prediction due to their complex geological features, including extensive burial depth, weak seismic signals, and high heterogeneity. To address these challenges, this study develops an advanced deep learning approach specifically designed for ultra-deep, fault-controlled, fractured-vuggy reservoirs in the Tarim Basin. The study utilizes a three-dimensional seismic dataset and applies Principal Component Analysis (PCA) to select five key features from eight seismic attributes. Additionally, seismic phase-controlled constraints are incorporated into the model. Using deep learning technology, a porosity prediction model for ultra-deep carbonate reservoirs has been constructed. Validation using blind wells from the Shunbei oilfield shows that this approach achieves a 76% reduction in Mean Square Error (MSE) compared to traditional impedance inversion techniques, highlighting its high predictive accuracy. Through SHapley Additive exPlanations (SHAP) analysis, the attributes LAMBDA_AAGFIL and PHASE_ANT are identified as the most influential, highlighting their importance in representing karst cave and fracture structures within the reservoir. These findings underscore the innovation and substantial improvement of the proposed method over conventional techniques, offering a robust and high-precision approach for porosity prediction in ultra-deep carbonate reservoirs.

A new modeling approach for microplastic drag and settling velocity

Understanding microplastics' (MPs') transport and settling behaviors in aquatic environments is crucial for devising effective management strategies. This study contributes a novel modeling framework to develop accurate and interpretable drag and velocity models for MPs using machine learning techniques. It achieves faster model creation and improved accuracy than traditional methods like theoretical analysis and data fitting. The framework demonstrates high predictive accuracy across different MP types (1D, 2D, 3D, and mixed), with a coefficient of determination CD = 0.86–0.95 for the drag models and CD = 0.92–0.95 for the velocity models. Compared with best-performing empirical approaches, the new drag models exhibit an average reduction in root mean square error (RMSE) by 59% and mean absolute error (MAE) by 62%. Similarly, the velocity models show a mean decrease in RMSE and MAE by 27% and 25%, respectively. Moreover, the framework outperforms commonly used symbolic regression methods, reducing errors by 18%–27%. The sensitivity analysis reveals that the relative density difference and the dimensionless diameter are essential for predicting the settling of all MP types, while the effective shape parameters vary across different MP categories. By providing accurate predictions of MPs' settling dynamics, this study offers insights for developing targeted mitigation strategies to reduce MPs' environmental impacts.

A novel approach for snow depth retrieval in forested areas by integrating horizontal and vertical canopy structures information

Snow cover significantly influences the Earth’s climate system and global hydrological cycle through its thermal insulation properties and high albedo, and is an important component of the cryosphere. Currently, the most efficient means of quantifying snow depth at both global and regional scales is through passive microwave remote sensing. However, the accuracy of passive microwave remote sensing inversion of snow depth in forested areas is affected by the forest canopy. In this study, a normalized maximum stem volume (NMSV) index was constructed by combining canopy height and tree cover data obtained through remote sensing techniques. The NMSV index was then incorporated into development of snow depth retrieval algorithm to improve accuracy of passive microwave snow depth estimation in forested areas. Compared to the Chang algorithm and the AMSR-E snow depth product, this study demonstrated higher accuracy in the mid- to high-latitude forested areas of Eurasia, with an R value approximately twice as high and a reduction in the overall root mean square error (RMSE) by 2.3 cm and 7.2 cm, respectively. The relative mean bias of this study in the Western Russia, the Eastern Siberian Mountains and the Northeast China is significantly reduced than that of the existing remote sensing algorithms. Against the ERA5 and GlobSnow datasets, with the exception of the Western Russia, the performance of this study in the mid- to high-latitude forested areas of Eurasia is comparable to the ERA5 dataset and superior to the other datasets. Based on the performance of the algorithms in different NMSV values, we observe a decline in the accuracy of this algorithm when the value exceeds 0.8, which was caused by small size of high NMSV values among the ground observation sites involved in the development of snow depth retrieval algorithm. Overall, the NMSV index proposed in this study, which integrates information from both the horizontal and vertical structures of forest, can better characterize the microwave radiation properties of sparse and moderately dense forests, facilitating improvements in the accuracy of passive microwave snow depth retrieval in global forested areas.

Exploring the integration of a global AI model with traditional data assimilation in weather forecasting

Abstract Recent advancements in artificial intelligence (AI) have profoundly transformed weather forecasting, challenging traditional reliance on numerical weather prediction (NWP) models. Despite notable progress, AI models still depend heavily on traditional NWP systems and data assimilation methods to generate analysis fields, a dependency that increases computational demands and might limit forecast accuracy. This study explored the integration of Gridpoint Statistical Interpolation (GSI) with the Pangu-Weather AI forecasting model (GSI-Pangu), and assessed the potential for AI models to autonomously generate forecasts by leveraging mature data assimilation systems. Our experiments commenced by adopting ERA5 reanalysis data for the initial cycle, and then involved assimilation of simulated observations in subsequent cycles, spanning a month-long period. Results demonstrated notable enhancements in forecast accuracy, with reductions in the root mean square error across various atmospheric variables compared with the results of a control experiment without data assimilation. Additionally, the results highlighted GSI-Pangu’s ability to predict large-scale circulation patterns of extreme precipitation events, together with its effectiveness in driving regional models to accurately forecast precipitation intensity and distribution. Successful implementation of GSI within the Pangu-Weather framework underscores the transformative potential of hybrid forecasting systems, which merge conventional meteorological techniques with AI innovations, thereby facilitating accelerated adoption of AI in weather forecasting.

The Direct Assimilation of Radar Reflectivity Data with a Two-Moment Microphysics Scheme for a Landfalling Typhoon in an OSSE Framework

Despite the well-known importance of radar data assimilation, there are limited studies on landfalling typhoons in terms of directly assimilating radar reflectivity data, especially using a reflectivity operator based on double-moment microphysics. In this study, radar reflectivity data assimilation experiments are conducted with an ensemble Kalman filter (EnKF), using simulated observations in an Observing System Simulation Experiment (OSSE) framework for the landfalling typhoon In-Fa. With an OSSE, it is convenient to analyze the impact of assimilation of radar reflectivity on analysis and forecast for various state variables, especially for hydrometeors. Our results show that the direct assimilation of radar reflectivity with EnKF does not introduce non-physical hydrometeors and is able to adjust well, not only to hydrometers, but also to some large-scale variables which are not directly related to reflectivity, especially in terms of temperature and vertical velocity. Though the most notable reduction in the Root Mean Square Errors (RMSEs) is observed through mixing the ratio of rainwater and snow, the analysis of other variables is also significantly improved with the accumulation of assimilation cycles. The correlation analysis reveals the strongest correlation between radar reflectivity data and hydrometeor-related variables as well as the correlation with certain large-scale variables, indicating that these cross-variables are updated well through the reliable multivariate ensemble covariance in the EnKF. As a result, an obvious improvement in typhoon intensity and precipitation forecast is obtained in the data assimilation experiment. The impact of assimilation on radar reflectivity can last for up to 15–16 h.

The Impact of Firework Ban Relaxation on Variations in SO2 Emissions in China During the 2023 Chinese New Year

During the 2023 Chinese New Year (CNY), many city governments temporarily relaxed firework restrictions, leading to increased sulfur dioxide (SO2) emissions from the combustion of sulfur-containing fireworks. This study employed the four-dimensional variational (4DVar) assimilation system to examine variations in SO2 emissions in China by assimilating hourly ground-based observations. Two experiments were conducted during CNY in 2022 and 2023 to quantify the variations in SO2 emissions. On CNY’s Eve in 2023, following the relaxation of the firework ban, SO2 emissions surged by 8.22 Gg nationwide compared to the previous day with significant increases in the Energy Golden Triangle (2.037 Gg), the North China Plain (1.709 Gg), and northeast China (0.945 Gg). Emissions peaked on CNY’s Eve and rapidly declined in the following two days but remained elevated compared to the pre-CNY period, indicating lingering effects of firework burning. Compared to the forecasts using the prior emissions, the optimized emissions markedly improved the model forecasts of SO2 during the 2023 CNY period, with an increase in the correlation coefficient (R) from 0.13 to 0.64 and a reduction in the root mean square error (RMSE) by 49.2%, demonstrating the effectiveness of the optimized emissions. These findings will be useful for local governments in formulating strategies for firework burning during CNY.

Daily runoff forecasting using novel optimized machine learning methods

Accurate runoff forecasting is crucial for effective water resource management, yet existing models often face challenges due to the complexity of hydrological systems. This study addresses these challenges by introducing a novel bio-inspired metaheuristic algorithm, Artificial Rabbits Optimization (ARO), integrated with various machine learning (ML) models for runoff forecasting in the Carson and Chehalis River basins. The ARO-ML models are rigorously evaluated, demonstrating their superior ability to enhance forecasting accuracy compared to widely used ML techniques, including gradient boosting (GB), multi-layer perceptron (MLP), and K-nearest neighbor regression (KNN). In the Carson River, the GB model achieves the highest forecasting accuracy, which is significantly improved by ARO, resulting in a 24.8 % reduction in root mean square error (RMSE). The MLP model also benefits notably from ARO, with RMSE improvements of 4.8 % and a substantial 48.9 % reduction in mean absolute error (MAE). The ARO-optimized KNN model shows exceptional performance, surpassing the baseline KNN during testing. In the Chehalis River Basin, ARO integration leads to substantial improvements in both GB and MLP models across various performance metrics during training and testing. Crucially, the results indicate that in both river basins, rainfall in preceding days has the most significant influence on streamflow forecasts, followed by air temperature. This study advances hydrological forecasting by validating the efficacy of ARO-ML models, offering a robust framework for enhanced predictive accuracy and providing deeper insights into the key drivers of streamflow in diverse hydrological contexts.

Ultra Enhancement of the Resolution of the Divergence Beam-Based SPR Sensor Using a Wavelet Filter

Surface Plasmon Resonance (SPR) sensors are highly sensitive to refractive index variations, making them ideal for biosensing and chemical detection applications. However, their performance can be constrained by noise, resolution limits, and signal distortions, particularly in divergence beam-based configurations. This research presents an ultra-enhancement of the performance of divergence beam-based SPR sensors by employing advanced wavelet filtering techniques. Wavelets, with their multi-resolution analysis capability, are applied to denoise and refine the sensor signal, significantly improving key performance metrics, including resolution, mean square error (MSE), root mean square error (RMSE), signal-to-noise ratio (SNR), sensitivity matrix (SM), and combined sensitivity factor (CSF).The wavelet filter effectively decomposes the SPR signal into distinct frequency bands, separating noise while retaining high-frequency components required for accurate sensing. This results in a significant reduction in MSE and RMSE of 30 and 92.26%, respectively, while simultaneously improving SNR by 54.08%, maintaining signal quality. The wavelet filter significantly improved the resolution of the SPR sensor by 99.95% (1.3772×10-7 RIU), allowing for more precise detection of refractive index changes and boosting diverge beam-based SPR sensor performance. In addition, a sensitivity matrix (SM) and combined sensitivity factor (CSF) were improved by 42 and 41.94%, respectively, allowing for a more thorough evaluation of the sensor's performance across multiple operational parameters. Simulation and experimentation show that wavelet filtering surpasses standard filtering approaches in terms of noise suppression and signal clarity. This technology has considerable potential for improving the accuracy and reliability of SPR sensors in high-sensitivity applications

New biased estimators for the Conway–Maxwell-Poisson Model

Recent studies focus on modelling count data, which often shows overdispersion or underdispersion. The Conway–Maxwell–Poisson (COMP) regression model effectively handles such dispersion. However, multicollinearity can negatively impact the maximum likelihood estimator (MLE) by increasing variance. To address this, biased estimators like the ridge estimator have been suggested to mitigate multicollinearity effects. This research proposes a new COMP hybrid estimator to further address multicollinearity issues. Theoretical comparisons between the COMP hybrid estimator, MLE, and other COMP estimators reveal that the hybrid estimator reduces mean squared error. Monte Carlo simulations and practical applications confirm that these proposed estimators outperform both MLE and existing COMP estimators.

Lithium battery SOC estimation based on improved sparrow search algorithm and backpropagation neural network

Accurate state-of-charge (SOC) estimation is crucial for optimal battery management. This paper proposes a novel method, the Improved Sparrow Search Algorithm-Backpropagation (ISSA-BP) neural network, to address the issue of low estimation accuracy encountered with a single BP neural network. ISSA is used to optimize the initial weights and thresholds of the BP neural network, effectively overcoming its tendency to get stuck in local minima. Compared to the single BP neural network, ISSA-BP demonstrates significantly improved accuracy under two conditions (DST and BJDST), with reductions in root mean square error by 64.0% and 50.9% and mean absolute error by 69.8% and 51.1%, respectively. These results highlight the superior robustness and accuracy of the ISSA-BP algorithm for SOC estimation in lithium batteries.

Deep learning approaches for image restoration in invasive coronary angiography

Abstract Background Despite being the gold standard for diagnosing coronary artery disease and guiding percutaneous coronary interventions (PCI), image restoration during invasive coronary angiography (ICA) has rarely been explored. Generative adversarial network (GAN), a deep learning model, has demonstrated its effectiveness in image restoration. Our hypothesis was that we could restore ICA images, damaged or deleted to varying degrees, using the inpainting GAN method, and we undertook efforts to test this hypothesis. Methods We extracted 22,659 ICA patch images (128 × 128 size) from 60 patients, focusing on sparse vessels for model training. To assess our inpainting GAN model, we degraded the extracted images to varying degrees: minimal, mild, moderate, severe, and completely blocked. The performance was quantitatively analyzed using metrics such as the peak signal-to-noise ratio (PSNR), structured similarity indexing methods (SSIM), and mean squared error (MSE). Results The inpainting GAN method significantly improved the image quality in all degraded images. For the blocked images, PSNR and SSIM were enhanced by 161.9% and 14.3%, respectively, with a 98% MSE reduction. In severely, moderately, mildly, and minimally damaged images, PSNR and SSIM increased by 14.9% and 14.8%, 15.7% and 14.7%, 19.3% and 15.6%, and 12.7% and 5.3%, respectively, with MSE reductions of 61.5%, 64.1%, 72.7%, and 33.9%, respectively. Conclusions Inpainting GAN can restore ICA images with varying damage levels, including deletions. Our deep-learning model can instantaneously improve ICA image quality, potentially aiding in emergency or chronic total occlusion PCI.Table 2.Quantitative Analysis of ImageFigure 1.Sample Cases Pre and Post-Rest

Enhancing streamflow prediction in the Wujiang River basin: a two-stage decomposition approach with deep learning integration

ABSTRACT Global climate change and human activities have profoundly impacted the geological and hydrological processes in watersheds, increasing the challenges in streamflow prediction. In this study, we propose a streamflow prediction model based on deep learning and dual-mode decomposition specifically tailored for the Wujiang River basin, a significant tributary on the southern bank of the upper Yangtze River. This model effectively addresses the prediction error issue caused by high-frequency components through dual-mode decomposition of time-series data. The results demonstrate that employing the coupled complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN)–variational mode decomposition (VMD)–temporal convolutional network (TCN)–long short-term memory network (LSTM) model reduces prediction errors by at least 35% compared to single decomposition models. Furthermore, compared to individual TCN or LSTM models, the TCN–LSTM coupled model exhibits greater stability and higher prediction accuracy during training, with reductions in mean absolute error and root mean square error by 43.13 and 24.57%, respectively. This model holds promising prospects for application and can provide crucial insights for water resource management.