Search Results (311)

To address the challenges of characterizing subway passenger flow fluctuations and overcoming the slow convergence and significant errors of existing intelligent optimization algorithms in tuning deep learning parameters for flow prediction, this study proposes a novel subway passenger flow fluctuation interval prediction model based on a Symmetry-Enhanced FIG-ICPO-XGBoost model. The core innovation is an Improved Cheetah Optimization Algorithm (ICPO), which incorporates enhancements including Circle mapping for population initialization, a hybrid strategy of dimension-by-dimension pinhole imaging opposition-based learning and Cauchy mutation to escape local optima, and adaptive variable spiral search with inertia weight to balance exploration and exploitation. The construction of this methodology embodies the concept of symmetry in algorithm design. For instance, Circle mapping achieves uniformity and ergodicity in the initial distribution of the population within the solution space, reflecting the symmetric principle of spatial coverage. Dimension-by-dimension pinhole imaging opposition-based learning generates opposite solutions through the principle of mirror symmetry, effectively expanding the search space. The adaptive variable spiral search strategy dynamically adjusts the spiral shape, simulating the symmetric relationship of dynamic balance between exploration and exploitation. Utilizing fuzzy-granulated passenger flow data (LOW, R, UP) from Harbin, the ICPO was employed to optimize XGBoost hyperparameters. Experimental results demonstrate the superior performance of the ICPO on 12 benchmark functions. The ICPO-XGBoost model achieves mean MAE, RMSE, and MAPE values of 10,291, 10,612, and 5.8%, respectively, for the predictions of the LOW, R, and UP datasets. Compared to existing models such as CPO-XGBoost, PSO-BiLSTM, GA-BP, and CNN-LSTM, these values represent improvements ranging from 4541 to 13,161 for MAE, 5258 to 14,613 for RMSE, and 2.6% to 7.2% for MAPE. The proposed model provides a reliable theoretical and data-driven foundation for optimizing subway train schedules and station passenger flow management. Full article

In this study, a double-lattice photonic crystal structure was designed to achieve deep ultraviolet lasing without the use of any Distributed Bragg Reflector (DBR), which is called a photonic-crystal surface-emitting laser (PCSEL). The plane wave expansion (PWE) method was used to study the influence of various structural parameters on the resonant wavelength. Utilizing the random forest algorithm, we determined that the importance of the lattice constant to the resonant wavelength is 95.24%. Furthermore, we realized the reverse design of double-lattice photonic crystals from the target wavelength to optimal structural parameters through a radial basis function (RBF) network algorithm. Comparative analysis of the extreme learning machine (ELM) and back propagation (BP) algorithms demonstrated that RBF-based performance was notably superior to the training outcomes of other algorithms. The mean absolute error (MAE) of the lattice constant of the test set in the training results was 0.7610 nm, root mean square error (RMSE) was $1.143\times {10}^{-3}$ nm, and mean absolute relative error (MARE) was $5.489\times {10}^{-3}$. We verified the reliability of the algorithm and designed 13 groups of photonic crystals with different epitaxial structures. The mean square error (MSE) was 0.6188 ${\mathrm{nm}}^2$ compared with that of the plane wave expansion method. This work demonstrates applicability across various wavebands and epitaxial structures in GaN-based devices, providing a novel approach for the rapid iteration of deep ultraviolet PCSELs. Full article

In the domain of marine remote sensing, the real-time monitoring of ocean waves is a research hotspot, which employs acquired X-band radar images to retrieve wave information. To enhance the accuracy of the classical spectrum method using the extracted signal-to-noise ratio (SNR) from an image sequence, data from the preferred analysis area around the upwind is required. Additionally, the accuracy requires further improvement in cases of low wind speed and swell. For shore-based radar, access to the preferred analysis area cannot be guaranteed in practice, which limits the measurement accuracy of the spectrum method. In this paper, a method using extracted SNRs and an optimized genetic algorithm back-propagation (GABP) neural network model is proposed to enhance the inversion accuracy of significant wave height. The extracted SNRs from multiple selected analysis regions, included angles, and wind speed are employed to construct a feature vector as the input parameter of the GABP neural network. Considering the not-completely linear relationship of wave height to the SNR derived from radar images, the GABP network model is used to fit the relationship. Compared with the classical SNR-based method, the correlation coefficient using the GABP neural network is improved by 0.14, and the root mean square error is reduced by 0.20 m. Full article

Coal and gas outbursts constitute a major hazard for mining safety, which is critical for the sustainable development of China’s energy industry. Rapid, accurate, and reliable pre-diction is pivotal for preventing and controlling outburst incidents. Nevertheless, the mechanisms driving coal and gas outbursts involve highly complex influencing factors. Four main geological indicators were identified by examining the attributes of these factors and their association to outburst intensity. This study developed a machine learning-based prediction model for outburst risk. Five algorithms were evaluated: K Nearest Neighbors (KNN), Back Propagation (BP), Random Forest (RF), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost). Model optimization was performed via Bayesian hyperparameter (BO) tuning. Model performance was assessed by the Receiver Operating Characteristic (ROC) curve; the optimized XGBoost model demonstrated strong predictive performance. To enhance model transparency and interpretability, the SHapley Additive exPlanations (SHAP) method was implemented. The SHAP analysis identified geological structure was the most important predictive feature, providing a practical decision support tool for mine executives to prevent and control outburst incidents. Full article

The intrinsically low stiffness of titanium alloy thin-walled components causes residual stresses to readily accumulate during high-speed micro-milling, leading to deformation and hindering machining precision. To clarify the residual-stress formation mechanism and enable deformation control, this study first proposes a surface residual stress characterization model based on an exponentially decaying sinusoidal function, with model parameters efficiently identified via an improved particle swarm optimization algorithm, allowing rapid characterization of stress distributions under different process conditions. A response surface model constructed using a central composite design is then employed to reveal the coupled effects of machining parameters on residual stress and top-surface deformation. On this basis, a GA-BP neural network–based prediction framework is developed to improve the accuracy of residual stress and deformation prediction, while the AGE-MOEA2 multi-objective evolutionary algorithm is used to optimize micro-milling parameters for the simultaneous minimization of residual stress and deformation via Pareto-optimal solutions. Validation experiments on thin-wall micro-milling confirm that the optimized parameters significantly reduce peak residual stress and suppress top-surface deformation. The proposed modeling and optimization strategy provides an effective reference for high-precision machining of titanium alloy thin-walled components. Full article

Thorough analysis of regional carbon emissions is essential, given their substantial contribution to overall carbon emissions, as it underpins the formulation of targeted low-carbon development strategies. This study investigates whether Hubei Province can achieve carbon peaking by 2030 and explores feasible emission reduction pathways. To address the limitations of existing regional carbon emission studies—particularly the direct use of decomposition factors in prediction models and the lack of logical separation between mechanism analysis and forecasting—a hybrid analytical-predictive framework is proposed. Specifically, the logarithmic mean Divisia index (LMDI) method is first employed to decompose historical carbon emissions and identify the driving forces, while the STIRPAT model combined with the Lasso regression is subsequently used to screen key influencing factors for emission prediction, thereby avoiding the direct use of decomposition factors in forecasting. Based on the selected factors, a genetic algorithm–optimized backpropagation neural network (GA-BP) is developed to predict carbon emissions in Hubei Province from 2024 to 2035. The predictive performance of the GA-BP model is validated using three statistical indicators (R², MAPE, and RMSE) and compared with Extreme Learning Machine (ELM), Support Vector Regression (SVR), and conventional BP models. Furthermore, six development scenarios are designed in accordance with provincial policy objectives to assess the feasibility of carbon peaking. The results indicate the following: (1) Based on the results of the LMDI decomposition, Lasso–STIRPAT analysis, and model sensitivity analysis, per capita GDP is identified as the primary driving factor of carbon emissions in Hubei Province. (2) The GA-BP model demonstrates superior predictive accuracy compared with benchmark models and (3) carbon peaking by 2030 can only be achieved under Scenario 6, highlighting the necessity of coordinated structural and technological interventions. Based on these findings, targeted policy recommendations for carbon emission reduction are proposed. Full article

With the improvement scale of grid connection renewable power, accurately forecasting and effectively coordinating systems with various energy sources has become much more important for power system scheduling and operation. Considering the uncertain characteristics of renewable energy and CHP units, this paper proposes an optimal dispatch method with multi-prediction models and an improved solving method by series correction and parallel coupling analysis. Firstly, multiple-model stationary time series are obtained by EMD (empirical mode decomposition) of the prediction results from multiple models. Then, series decomposition is updated by the UKF (unscented Kalman filter). Using the least-squares method, the parallel coupling of the correction results is solved. A complex optimal scheduling model with multiple renewable energy sources and CHP units is proposed and solved with the help of the improved GA and PSO combined algorithm to avoid the algorithm falling into local optimal conditions. Simulations show that the proposed optimal dispatch model and algorithm are able to consider the uncertain characteristics of renewable energy and CHP units with better performance than some typical methods, such as the baseline method that combines single-model BP forecasting with conventional PSO-based dispatch. These results demonstrate that the proposed EMD–UKF-based multi-model forecasting combined with the improved GA–PSO-based dispatch framework provides an effective and practically applicable tool for enhancing the economic and low-carbon operation of multi-energy systems with high renewable penetration. Full article

Eel (Anguilla) is an aquatic animal with high nutritional value and multiple health benefits for the human body. To fully utilize its processing by-products fish bone, this study optimized the enzymatic preparation process of using BP neural network and GA genetic algorithm, with collagen extraction yield as the key evaluation metric, and characterized the properties of the obtained collagen. The results demonstrated that the optimal extraction conditions for eel bone collagen were as follows: enzyme dosage of 2%, hydrolysis time of 2.65 h, solid-to-liquid ratio of 1:22, and ultrasonic pretreatment for 21 min at 250 W power, achieving an extraction yield of 57.6%. The main amino acids identified were glycine, glutamic acid, proline, and arginine. SDS-PAGE electrophoresis revealed that eel bone collagen exhibited structural characteristics of type I collagen. Raman spectroscopy and X-ray diffraction indicated an intact triple-helix structure with partial ordered features. The DSC and TGA results demonstrated good thermal stability, with a denaturation temperature of 106.73 °C. SEM imaging displayed a loose, porous fibrous network structure, while rheological analysis suggested potential biomedical material properties. The findings of this study provide fundamental data for the high-value utilization and development of eel bone resources. Full article

Highway under-bridge areas represent a valuable land resource while simultaneously constituting a sensitive ecological zone. Achieving a balance between its redevelopment and ecological preservation constitutes a critical challenge within the field of ecological engineering. Although prior research has addressed urban elevated underbridge space, investigations specifically focusing on highway underpasses remain limited. The absence of standardized criteria for assessing the suitability of these spaces has resulted in uncoordinated and fragmented utilization. In response, this study proposes a comprehensive evaluation framework that integrates multi-criteria decision-making (MCDM) methodologies with optimized backpropagation neural networks, specifically genetic-algorithm-optimized BP (GA-BP) and particle-swarm-optimization-optimized BP (PSO-BP). The model incorporates indicators spanning physical characteristics, environmental factors, safety considerations, and accessibility metrics, and is applied to an empirical dataset comprising 134 highway bridge underpasses in Fuzhou City. The results indicate that (1) both the GA-BP and PSO-BP models enhance convergence speed and classification accuracy, with the GA-BP model demonstrating superior stability and suitability for classifying underpass suitability; (2) the principal determinants of suitability include traffic accessibility, safety parameters, and spatial relationships with adjacent water bodies and agricultural lands; and (3) underpasses characterized as hub-type, single-sided road-adjacent, and cross-connection configurations exhibit greater potential for redevelopment. This investigation represents the first integration of MCDM and optimized neural network techniques in this context, offering a robust tool to support the scientific planning and ecological conservation of underbridge space environments. Full article

The extensive area of goaf makes high-temperature points highly concealed, and prolonged heating can easily trigger spontaneous coal combustion. Traditional temperature monitoring methods are limited in spatial coverage and thus fail to detect high-temperature points in a timely manner. To address this issue, this study proposes an integrated analytical method combining numerical simulation and intelligent inversion, with Taihe Coal Mine as the research object. First, A coupled flow–temperature–gas field model of the goaf was established in COMSOL Multiphysics 6.3 to simulate working-face CO concentration distributions corresponding to high-temperature points at different locations, thereby constructing a comprehensive dataset. Then, a BP neural network prediction model improved by the dung beetle optimization algorithm (DBO-BP) was trained to infer the spatial location of high-temperature points based on CO concentration distributions. Finally, a geometric prediction method was introduced to guide precise drilling within the predicted high-risk areas for field verification. The results demonstrate that the proposed DBO-BP model can effectively trace the locations of high-temperature points from CO concentration data. When combined with the geometric prediction method, it provides an efficient and reliable technical solution for the early prevention of spontaneous coal combustion in goaf. Full article

Perforation completion is one of the primary methods for putting oil and gas wells into production, and the influencing factors and prediction of perforation skin factors have long been key research topics in the petroleum industry. This study systematically investigates the effects of multiple factors (including perforation depth, phase angle, shot density, hole diameter, damage zone depth, and damage severity) on the perforation skin factor in both low-permeability and medium-to-high-permeability reservoirs. The research first established a coupled flow model for conventional perforation completion using ANSYS Fluent 2024R1 numerical simulation software, which integrates perforation geometric parameters and formation damage-related parameters; then, conducting single-factor sensitivity analysis to obtain qualitative relationships between individual perforation parameters and well skin factor; it subsequently uses orthogonal experimental design to explore multi-factor combined effects and applied gray correlation theory to calculate factor-skin factor correlation coefficients; finally, it adopts the least squares method for linear and nonlinear fitting based on orthogonal experimental data (with linear fitting average error of 45.3% and nonlinear fitting error of 7.4%, thus selecting the nonlinear formula as the prediction model). The findings provide valuable insights for optimizing perforation parameters and predicting well skin factors under different reservoir conditions. Full article

In recent years, with the deepening of mining and tunnel excavation operations, the incidence of rock burst has also increased, prompting people to attracting increasing attention to microseismic monitoring technology. The location algorithm of microseismic events is the core of microseismic monitoring. In this study, a hybrid optimization algorithm, BP-GA-GN, which combines genetic algorithm (GA), BP neural network (BP) and Gauss-Newton method (GN), is introduced. The BP-GA-GN algorithm optimizes the initial weights and thresholds of the BP neural network through GA to avoid local optimum. The BP neural network is used to learn the nonlinear mapping between the sensor arrival time difference and the source position. Combined with the physical model constraints of GN, fine convergence is performed. We prove the robustness of the BP-GA-GN algorithm through a large number of numerical simulations. Compared with the traditional single algorithm, the algorithm shows excellent performance. Subsequently, the high precision and high efficiency of the method are further highlighted in the field data test of mine environment and tunnel environment. The average errors are 0.42 m and 2.54 m, respectively, rendering it a valuable tool for real-time microseismic monitoring. This study overcomes the limitations of traditional positioning methods. The algorithm can achieve high-speed training and high precision, thus significantly improving the early warning effect of rockburst risk. Full article

This study focuses on an integrated three-level multi-port liner ship vessel routing and scheduling optimization problem. Specifically, the three-level multi-port network consists of hub ports, feeder ports, and cargo source points, which provide the demands’ loading/unloading at each port. Considering vessel-specific constraints such as speed, capacity, and cost, we formulate the multi-port liner ship routing and scheduling optimization problem as a mixed integer linear programming model with the objective of minimizing total voyage cost and operating time. First, we employ machine learning models to forecast the short-term demand at different ports as the input. There are multiple feasible routes generated and allowed to be elected. Second, to ensure both computational efficiency and solution quality, we devise and compare genetic algorithm (GA), simulated annealing (SA), Gurobi and the branch-and-price (B&P) algorithm to optimize scheduling plans. Experimental results demonstrate that the proposed predict-then-optimization framework effectively addresses the complexity of multi-port scheduling and routing problems, achieving a reduction in total transportation cost by 0.81% to 8.08% and a decrease in computation time by 16.86% to 24.7% compared to baseline methods, particularly with the SA + B&P hybrid approach. This leads to overall efficiency and cost-saving ocean vessel operations. Full article

TBM construction projects require substantial investment, making the accurate and rational prediction of TBM advance speed essential for cost control and timely project completion. To address this, the study develops a precise predictive model for TBM advance speed by integrating the BP neural network model with the genetic algorithm. Initially, raw TBM data were processed to remove non-operational records and anomalous values recorded during construction, resulting in a refined database of TBM operational parameters. The surrounding rock conditions were classified based on FPI and TPI, two key indices reflecting rock mass excavatability and rock-breaking efficiency. Using the K-means clustering algorithm, the dataset was segmented into three distinct groups. Seven tunneling parameters were selected as input features for the neural network model. Subsequently, three GA-BP neural network models were developed for different rock mass categories, with key parameters optimized for enhanced performance. Prediction results demonstrate that the GA-BP neural network exhibits superior accuracy and generalization capability. Compared to a conventional BP neural network, the GA-BP model reduces prediction errors by more than 10%. Full article

Ambiguity Resolution (AR) is regarded as an effective technique for enhancing positioning accuracy and reducing convergence time in Precise Point Positioning (PPP). However, the Wide-Lane Fractional Cycle Bias (WL FCB) and Narrow-Lane Fractional Cycle Bias (NL FCB) needed for AR are generated from network solutions based on numerous globally distributed stations, leading to considerable computational load and processing time. A prediction model for FCB is proposed using the Genetic Algorithm Optimized Backpropagation Neural Network (GA-BPNN), and high-precision predictions of WL and NL FCB for Day of Year (DOY) 321 in 2023 are successfully achieved. Comparisons with iGMAS products show that predicted WL FCB deviations are within 0.01 cycles, and predicted NL FCB over 12 h deviates within 0.1 cycles (excluding satellite C20). The performance of three PPP schemes, Float, Fixed (based on FCB from iGMAS), and BP-Fixed (based on FCB predicted by GA-BPNN), is compared through experiments. For GPS + BDS-3, the accuracies of the BP-Fixed scheme are 0.0034 m, 0.0039 m, and 0.0100 m in the east, north, and up directions, respectively. The ambiguity fixed rates reach 98.62% for BP-Fixed. These outcomes confirm that the positioning performance using the predicted FCB of GA-BPNN is highly consistent with that using FCB products. Full article