The flow characteristics of cohesive and non-cohesive particles in the double barrel with differential velocity (DBDV) were studied. The viscosity of particles was represented by a contact model. The flow process of particles was simulated by the discrete element method. A representation method for mixing characteristics was proposed. The velocity distribution of particles, spatial distribution of polydisperse particles, and mixing characteristics caused by particle flow were analyzed. The flow characteristics of cohesive and non-cohesive particles were compared. The results show that the velocity of cohesive particles is higher than that of non-cohesive particles. The non-cohesive particles at the bottom of the drum are prone to form a dead zone of motion. The sensitivity of the velocity of cohesive particles to particle size is lower than that of non- cohesive particles. Compared with non-cohesive particles, the mixing uniformity of cohesive small-sized particles is improved by 13.51

Moisture damage is a common hidden distress in asphalt pavements in hot and rainy regions, where it can rapidly develop into severe surface deterioration if not detected in time. To address this issue, this study proposes an automated framework integrating ground-penetrating radar (GPR) data and the YOLOv13 model for multi-scale moisture damage detection on the Jingang Expressway in Cambodia. A total of 1672 GPR images containing moisture damage were collected through field surveys using a 2.3 GHz GPR system. Based on field statistical analysis, the detected damage was classified into three scale levels: large-scale (>2 m), medium-scale (0.8-2 m), and tiny-scale (<0.8 m). Several recent YOLO variants were compared, and YOLOv13s was identified as the optimal model, achieving the best balance between detection accuracy and inference efficiency, with an mAP@0.5 of 85.3% and an FPS of 48. The proposed method was further validated through laboratory and field tests. The results indicate that the developed framework can effectively detect and localize multi-scale moisture damage under practical engineering conditions, providing a non-destructive and efficient approach for pavement condition assessment in hot and rainy regions. By enabling early-stage detection of moisture damage deterioration, the proposed framework may contribute to more sustainable pavement maintenance and long-term transportation infrastructure management.

This paper addresses the bistable motion that occurs in hydraulic rock drills during drilling by applying a nonsmooth-system-based bistability control method to the drill's hard-impact system. This method enables the transition from undesired attractors to desired attractors via continuous pathways without affecting the original state of the system. First, a multi-degree-of-freedom dynamic model of the rock drill is established based on the dry friction rock theory, examining both viscous and non-viscous movements. An analysis of unidirectional bifurcation with respect to impact frequency reveals that viscous motion can easily induce chaos in the system. When omega = 3, the system is in the optimal motion state p1q1r1, and the rate of penetration (ROP) reaches its maximum value. In the high-frequency range, the system exhibits poor stability, showing chaotic behavior and bistability. Bidirectional bifurcation analysis at omega = 9, conducted with different initial values, reveals two bistable attractors, p0q1r1 and p1q0r2. The global attraction domain of the p1q0r2 attractor accounts for up to 90.6%. Subsequently, linear and nonlinear control methods are employed to facilitate transitions between these two attractor regions. Additionally, a pseudoarclength continuation method is adopted for path tracking to determine the parameter domains a and b suitable for the proposed control strategy. Numerical simulations show that controlling parameters such as impact force amplitude, offset, or collision spacing can effectively alter the attractor states. Finally, the validity of the model is verified experimentally. The hydraulic rock drill exhibits bistable motion, which is attributed to unstable stress waves generated by the impact of drill tools on heterogeneous rock during high-frequency drilling. Furthermore, improper matching among the impact force, propulsion force, and buffer force causes jumps in hydraulic pressure frequency at the moment of direction reversal, which consequently alters the initial working parameters of the rock drill.

Welding strength matching has significant influences on fatigue properties of welded structures. In this study, gradient strength matched (GSM) process with different filler metals ER50-6 (E5) and ER80S-G (E8) are applied for Q235/Q460 welded joints. Three filling configurations are used: E5 + E8 + E5 (GSM3), E5 + E8 + E5 + E8 + E5 (GSM5) and E5 (conventional joint). Experiments including microstructure and hardness, tensile, and fatigue crack growth (FCG) tests for base metals, welded metal (WM) and heat affected zones (HAZ) are conducted. Residual stress fields for welded joints are simulated using Abaqus software. Results show improved FCG resistance and fracture toughness of WM of GSM3 and GSM5 with corresponding crack opening force values increasing 3.5 %-19.6 % and 9.2 %-40 %, respectively. The fracture toughness in WM of GSM3 and GSM5 increases 11 % and 30.4 %, respectively. However, GSM process has limited impact on microstructures, residual stress distribution, and FCG behavior in HAZ due to fixed welding heat input. Additionally, a modified Forman model is proposed by introducing number of E8/E5 interfaces in GSM joints, which provide a new methods and theoretical foundations for fatigue-resistant design and life prediction of dissimilarsteel welded structures.

Blade tip timing (BTT) is a non-contact monitoring technology for rotor blade vibrations. Due to its severe undersampling, anti-aliasing spectral analysis has become a key focus. We identified certain specific frequencies that cannot be recognized during spectral analysis, a phenomenon we have termed frequency ambiguity. To investigate its mathematical basis, we assume constant rotational speed and neglect blade vibration effects on sampling time, hereby modeling BTT sampling as periodic non-uniform sampling. The time-domain multiplication of the original signal and sampling pulses corresponds to a convolution in the frequency domain. Frequency ambiguity arises from the uncertainty introduced by the complex weighted sum of two spectral peaks of the original signal. Then, building on the ambiguous frequency defined by this phenomenon, we propose the synchronous filtering using block spatial smoothing for subspace-based methods, which can be explained through the collapse of the signal subspace dimension. Simulation and experimental results demonstrate that subspace-based methods, such as multiple signal classification (MUSIC) and estimating signal parameter variational invariance techniques, exhibit synchronous filtering property that effectively suppresses ambiguous frequencies, thereby improving the accuracy of blade natural frequency identification. Notably, MUSIC exhibits superior noise immunity.

To address the system instability caused by heterogeneous rock stiffness during the percussive drilling process of rock drills, this study applies bifurcation theory and nonlinear time series methods to investigate system stability and subsequently estimate rock stiffness. First, a dynamic model of the rock drill is established based on rock contact theory, examining the drilling characteristics of the system in the optimal stable state of “1-impact-per-cycle”. Using impact frequency as the control parameter and employing the variable-step pseudo-arclength continuation method, the study precisely identifies period-doubling bifurcation points and saddle-node bifurcation points through path tracking, providing a reference interval for maintaining the “1-impact-per-cycle” state. Subsequently, simulated percussive drilling experiments are conducted to identify loading and unloading stiffness parameters, investigating the influence of simulated rock device stiffness on drilling process stability. Finally, mutual information and false nearest neighbor methods are applied to perform nonlinear time series analysis on simulated and experimental drill bit acceleration data during stable drilling. Through 3D phase space reconstruction, the penetration time is calculated using peak values of the tangent vector direction gradient. Combined with analytical prediction methods, the study concludes that higher rock stiffness corresponds to shorter penetration time. This correlation mechanism can be utilized to estimate rock stiffness, thereby optimizing rock drill parameters during stable operation.

The effect of particle shape and size on the mixing performance of a double barrel with differential velocity (DBDV) was investigated on the basis of a discrete element method (DEM). The mixing performance was determined on the basis of the velocity, mixing efficiency and mixing uniformity. The DBDV model, motion model and particle models of different shapes were constructed. The motion process of the particles was visualized by a DEM. The effect of particle shape, particle size and their coupling effect on the mixing performance of DBDV was analyzed. The simulation results were validated. The results indicate that the differential zone is less sensitive to particle shape than the nondifferential zone. In terms of mixing uniformity, choosing a lower linear velocity at low and medium inclinations is more appropriate; similarly, choosing a high linear velocity at a large inclination is more appropriate. In the differential region, the particle velocity decreases as the particle size decreases. The mixing uniformity increases but then decreases with increasing particle curvature and decreasing particle size. The particle size significantly affects the mixing time and mixing uniformity of particles. Choosing medium-sized ellipsoidal and square particles is better for achieving high mixing uniformity.

Accurate reducer noise prediction is essential for condition monitoring and predictive maintenance of mechanical transmission systems. However, the strong nonlinear coupling between operating conditions and noise responses, together with measurement uncertainties, remains a major challenge for data-driven prediction methods. This study proposes a hybrid prediction framework integrating feature selection, adaptive neural modeling, and residual compensation to improve the accuracy of planetary winch reducer noise prediction. A random forest (RF)-based feature selection strategy is first employed to identify the most informative vibration characteristics associated with reducer noise. Subsequently, a generalized regression neural network (GRNN) optimized by a hybrid whale optimization and bat algorithm (WOA-BAT) is developed to adaptively determine the smoothing factor and enhance nonlinear prediction capability. Furthermore, a residual Kalman compensation (RKC) mechanism is introduced to suppress prediction fluctuations caused by stochastic disturbances and modeling uncertainties. Experimental results demonstrate that the proposed WOA-BAT-GRNN-RKC framework achieves highly accurate noise prediction, with an RMSE of 0.05631 dB and an MAE of 0.027972 dB. The corresponding MAPE is 0.037897%. The proposed approach provides an effective pathway toward intelligent reducer condition monitoring and predictive maintenance.

Extracting high-quality health indicators (HIs) is major for accurately predicting state of health (SOH) and remaining useful life (RUL). To address issues of existing algorithms heavily relying on training data, an optimized ensemble framework based on hybrid neural networks for SOH and RUL prediction is proposed. Firstly, the aging characteristics are characterized only by extracting the current, voltage, and capacity increment curves within the first 10 % of the battery data, and the Person method is introduced to analyze the correlation between HIs and capacity. Secondly, a multi-scale feature fusion method is proposed, Specifically, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is used for extracting the multi-scale components of the feature sequence, and principal component analysis (PCA) performs feature fusion on the high-dimensional components to improve the feature quality. Thirdly, an ensemble framework that combines least squares support vector machine (LSSVM) and convolutional neural network (CNN) suitable for small samples and high-dimensional data is proposed, and the Kepler optimization algorithm (KOA) is applied to optimize the hyperparameters of CNN to enhance local degradation feature mining. Finally, in the 10 % small sample scenario, the SOH and RUL errors predicted by the proposed model are controlled within 1 %.

To address the issues of high energy consumption and component collisions caused by intense nonlinear vibrations during the drilling process of hydraulic rock drills, this paper proposes a modeling and optimization method based on wake oscillator theory. By establishing a fluid–structure interaction drilling dynamics model that incorporates axial and longitudinal vibration impacts, combined with rock contact theory, the study systematically reveals the influence patterns of impact frequency and pressure on the system’s dynamic characteristics and energy utilization efficiency. As the impact frequency increases, the system undergoes period-doubling bifurcation and saddle-node bifurcation, accompanied by a decline in energy utilization efficiency. Within the chaotic regime induced by bifurcation, energy utilization shows a slight recovery; however, the increased longitudinal amplitude tends to exceed the design clearance, leading to collisions. Reducing the impact pressure compromises system stability, making it more sensitive to fluctuations in the backflow backpressure, while the impact pressure exhibits a positive correlation with the longitudinal amplitude. Experiments validated the bifurcation behavior predicted by the theoretical model, confirming “impact number 1, cycle number 1” as the optimal drilling mode. Furthermore, a collaborative optimization of key structural parameters using Six Sigma closed-loop optimization and backpropagation (BP) neural networks demonstrated that increasing the inner diameter of the impact piston’s rear end effectively reduces longitudinal vibrations. This study provides a theoretical basis and design guidance for preventing collision phenomena under high-frequency and high-pressure conditions and enhancing energy utilization efficiency.

In response to the complex multistable behavior observed in hydraulic rock drills during the drilling process, this study first establishes a four-degree-of-freedom physical model based on dry friction rock mechanics theory. The motion trajectory is classified into three states: non-viscous, impact viscous, and buffer viscous. Using the impact frequency ω as the bifurcation parameter, multistable attractors p0q1 and p1q2 are identified in the system when ω = 9. To control the multistability, a delayed feedback control method is applied, in which the infinite-dimensional delay differential equations are approximated by finite-dimensional ordinary differential equations. The reliability of this approximation is validated through a distance function. When the control gain K = 9 and the delay time τd = 0.35, both attractors p0q1 and p1q2 are successfully converted into a single p0q1 attractor. Next, the pseudo-arclength continuation method and Floquet theory are employed to investigate parameter continuation and parameter domains. The period-doubling bifurcation points PD1 and PD2 divide the parameter space of K and τd into three distinct regions. Crossing these regions induces a supercritical period-doubling bifurcation. For constant K, a smaller τd leads to an increased number of collisions and periodic motions in the system. Simulation results demonstrate that by tuning the delay parameters, the multistability during the drilling process can be effectively controlled, thereby enhancing drilling efficiency and stability. Finally, rock drilling experiments confirm the validity of the model and the presence of multistability. When drilling into rocks with high hardness and brittleness, multistable motions are more likely to occur.

Fatigue performance of high-strength steel welded joints is widely concerned in engineering fields. In this research, the fatigue crack growth (FCG) behavior of HG785D steel with and without a "soft + hard + soft" composite welded metal is investigated by conducting FCG tests, hardness tests and fracture observation. Further, FCG curves (da/dN-Delta K) quantified by Seven-point incremental polynomial method (SP) and Smith method (SM) from tested crack length and number of cycles are compared. Results show the fatigue performance of composite welded metal is improved due to the hard layer ensures strength and soft layers provide toughness. Additionally, the fatigue life assessment is safer using da/dN-Delta K curves by SP, whereas more accurate using that by SM. The da/dN-Delta K curves by SP can capture the local fluctuations in FCG rate, while that by SM provides a smooth and global description for FCG behavior with steadily increased da/dN. Considering heterogeneous properties of composite welded metal, a piecewise function combining SP method is recommend to describe the unconventional da/dN-Delta K curves. Moreover, the correlation between Paris parameters for steels is analyzed, offering a reference for fatigue reliability assessments.

Voids beneath concrete slabs represent a critical yet often concealed defect that can precipitate rapid slab failure, posing substantial safety hazards. Conventional void detection methodologies predominantly rely on human interpretation of Ground-Penetrating Radar (GPR) data. However, accurately identifying voids at the centimeter (cm) level remains challenging due to the nonlinear evolution of voids and the non-unique nature of void features. Herein, this study proposes an innovative data-driven framework for cm-level void morphology and location detection, testing and repair in real-world cement pavement applications through the integration of GPR and deep learning techniques. Only GPR data are used as input for the automated void detection, obviating the need for destructive core sampling or excavation while providing precise grouting repair parameters. Through comprehensive numerical simulations and field experiments, we have enhanced the analysis of GPR responses to cm-scale voids, enabling more accurate detection. Furthermore, a specially designed deep learning architecture is developed to effectively capture and analyze non-unique void features from extensive field data. The methodology was validated through field tests employing an 800 MHz GPR system on core samples, demonstrating its practical feasibility. The results demonstrate a remarkable 1500 % improvement in detection speed compared to manual processing, with a mere 4.3 % volume error for a 25 cm void, and 93.8 % AP (average precision), meeting practical engineering requirements. This technological advancement offers significant improvements in void detection, reduces maintenance expenditures and extends pavement lifespan, and enhances overall traffic safety, representing a substantial contribution to intelligent infrastructure maintenance systems and measurement science.

Ensuring accurate assessment of the state of health (SOH) and state of charge (SOC) of lithium-ion batteries (LIBs) is crucial for ensuring their safe and stable operation. However, achieving accurate joint SOH and SOC estimation for LIBs still faces many challenges. Especially the long collection of characteristic data for the SOH and SOC of LIBs results in sparse samples and noise during the data collection process. This paper proposes the transfer learning-enhanced LSTM multi-head differential attention iTransformer (TL-LSTM-MHDA-iTransformer) hybrid neural network for joint SOH and SOC estimation. Firstly, the model utilizes transfer learning for small sample prediction and reduces program time. The pre-training model dataset uses three different types of LIBs: nickel cobalt aluminum (NCA), nickel cobalt manganese (NCM), and kokam. For further training, use 10 % or 20 % of the LIBs life cycle data as the training set to adjust the model parameters. Next, LSTM captures the long-term dependency relationship between input and output in time series. Then, MHDA uses the difference between two softmax functions to cancel out noise and enhance attention to key information. Finally, iTransformer combines input features for joint SOH and SOC estimation. In addition, this study uses the predicted SOH for the next usage cycle of LIBs to correct the maximum available capacity of the SOC calculation formula for that cycle, and predicts the SOH for the next usage cycle based on information such as SOC, voltage, and current during the discharge process of the LIBs in this cycle. To verify the necessity and effectiveness of this new method and its definition, we conducted in-depth research on transfer learning and noise resistance using LIBs datasets provided by Tongji University and Oxford University with different materials and temperatures. The experiment demonstrated that this method has significant effectiveness in improving the accuracy of joint SOH and SOC estimation. Specifically, during the validation phase, the SOH predicted the root mean squared error (RMSE) of NCA, NCM, and Kokam batteries to be controlled within 2.09 %, 2.32 %, and 0.62 %, respectively, while the mean absolute error (MAE) is controlled within 1.36 %, 2.26 %, and 0.25 %, respectively. The SOC estimation of NCA, NCM, and Kokam batteries with RMSE controlled within 1.52 %, 1.55 %, and 1.29 %, respectively, and MAE controlled within 0.25 %, 0.30 %, and 0.05 %, fully demonstrates the effectiveness of this method.

Escalating global energy demands and environmental concerns have positioned Hot Dry Rock as a pivotal renewable energy source. This study investigates the thermal performance of Enhanced Geothermal Systems in the Guanzhong Basin, China, integrating experimental measurements and numerical simulations to evaluate heat extraction dynamics. While the findings are contextspecific to the Guanzhong Basin's sedimentary geology-characterized by a geothermal gradient and fracture apertures. The methodology establishes transferable principles for Enhanced Geothermal Systems design in analogous basins. Key results demonstrate that optimized fracture configurations (connectivity, aperture) and controlled injection parameters (flow rates of 10-12 kg/s, well spacings of 250-400 m) significantly enhance energy extraction efficiency by mitigating thermal short-circuiting. A multi-objective genetic algorithm correlates operational variables with extraction efficiency, providing a quantitative framework for optimization. The study advances a replicable evaluation methodology, incorporating energy, economic, and environmental metrics, to guide sustainable Hot Dry Rock exploitation in fracture-dominated sedimentary systems.

To address the challenge that supervised learning for lithium-ion battery degradation diagnosis relies heavily on large-scale labeled datasets, this paper proposes an approach for multi-dimensional battery degradation diagnosis with limited labeled data and self-supervised learning. Our approach enables the battery degradation diagnosis of battery capacity, internal resistance, positive/negative electrode capacity, loss of lithium-ion inventory, and loss of active materials in both electrodes only using short partial charging data. The results show that our method achieves average root mean square errors below 2% at real-world drive cycle, demonstrating its capability in field applications. Comparative analysis indicates that the input with initial voltage of 3.85 V and charging duration of 400 seconds enables best diagnosis performance with physical interpretability. Our method enables real-time monitoring of battery aging states, providing theoretical support for optimizing energy management strategies in electric vehicles and energy storage systems, thus enhancing the operational reliability and safety of battery systems.

In response to the instability issues occurring during the drilling operation of rock drilling machines, this paper studies the principle of hydraulic rock drilling machines, considering the impact of the buffer system on the shock characteristics, and establishes a four-degree-of-freedom physical model. The viscous and non-viscous modes are analyzed, and the periodic trajectories of the non-smooth dynamical system's mathematical model are segmented. Using the pseudo-arclength continuation method and Floquet theory, the buffer pressure is treated as a control parameter to study the dynamic characteristics of multi-attractor coexistence in the buffer systemf. Multistable attractors cause the rock drill to switch between multiple steady states, and external disturbances allow the system to jump from one steady state to another. Simulation shows that in order for the rock drill to operate on a period-1 trajectory,d 1should be selected between 0.2 and 0.225. The correctness of the drilling dynamics model is validated through laser testing.

Fatigue crack growth (FCG) behavior related to stress ratio (R) is critical for assessing material durability. Based on fracture mechanics theory and asymptotic analysis, a progressive FCG rate model is derived by incorporating R, yield-to-tensile strength ratio, and fracture toughness by linking asymptotic limits of FCG rate curve with the R-dependent relationship between fracture toughness and fatigue crack growth threshold. Additionally, lognormal distribution theory is introduced for probabilistic analysis to consider the inevitable scatter in FCG measurements caused by material heterogeneity and testing errors. Validation results show that the proposed model can accurately describe FCG behavior for long cracks in near-threshold, Paris, and unstable stages at varying R from 0 to 0.6 for steels.

Semi-rigid base asphalt pavements, a common highway structure in China, often suffer from debonding defects which reduce road stability and shorten service life. In this study, a new method of road debonding detection based on the acoustic vibration method is proposed to address the needs of hidden debonding defects which are difficult to detect. The approach combines the Transformer model and the Transformer-based Parallel Cross-Gated Convolutional Neural Network (T-PCG-CNN) to classify and recognize semi-rigid base asphalt pavement acoustic data. Firstly, over a span of several years, an excitation device was designed and employed to collect acoustic data from different road types, creating a dedicated multi-sample dataset specifically for semi-rigid base asphalt pavements. Secondly, the improved Mel frequency cepstral coefficient (MFCC) feature and its first-order differential features (ΔMFCC) and second-order differential features (Δ2MFCC) are adopted as the input data of the network for different sample acoustic signal characteristics. Then, the proposed T-PCG-CNN model fuses the multi-frequency feature extraction advantage of a parallel cross-gate convolutional network and the long-time dependency capture ability of the Transformer model to improve the classification performance of different road acoustic features. Comprehensive experiments were conducted to analyze parameter sensitivity, feature combination strategies, and comparisons with existing classification algorithms. The results demonstrate that the proposed model achieves high accuracy and weighted F1 score. The confusion matrix indicates high per-class recall (including debonding), and the one-vs-rest ROC curves (AUC ≥ 0.95 for all classes) confirm strong class separability with low false-alarm trade-offs across operating thresholds. Moreover, the use of blockwise self-attention with global tokens and shared weight matrices significantly reduces model complexity and size. In the multi-type road data classification test, the classification accuracy reaches 0.9208 and the weighted F1 value reaches 0.9315, which is significantly better than the existing methods, demonstrating its generalizability in the identification of multiple road defect types.