Search Results (15,687)

To enhance the path-tracking accuracy of unmanned articulated road roller (UARR) operating on low-adhesion, slippery surfaces, this paper proposes a hierarchical cascaded control (HCC) architecture integrated with real-time ground adhesion coefficient estimation. Addressing the complex nonlinear dynamics between the two rigid bodies of the vehicle and its interaction with the ground, an upper-layer nonlinear model predictive controller (NMPC) is designed. This layer, based on a 4-degree-of-freedom (4-DOF) dynamic model, calculates the required steering torque using position and heading errors. The lower layer employs a second-order sliding mode controller (SOSMC) to precisely track the steering torque and output the corresponding steering wheel angle. To accommodate the anisotropic and time-varying nature of slippery surfaces, a strong-tracking unscented Kalman filter (ST-UKF) observer is introduced for ground adhesion coefficient estimation. By dynamically adjusting the covariance matrix, the observer reduces reliance on historical data while increasing the weight of new data, significantly improving real-time estimation accuracy. The estimated adhesion coefficient is fed back to the upper-layer NMPC, enhancing the control system’s adaptability and robustness under slippery conditions. The HCC is validated through simulation and real-vehicle experiments and compared with LQR and PID controllers. The results demonstrate that HCC achieves the fastest response time and smallest steady-state error on both dry and slippery gravel soil surfaces. Under slippery conditions, while control performance decreases compared to dry surfaces, incorporating ground adhesion coefficient observation reduces steady-state error by 20.62%. Full article

This paper addresses the significant variations in model parameters observed in autonomous commercial vehicles in comparison to passenger cars, with a disparity noted largely due to changes in load. Additionally, it tackles the issue of path tracking inaccuracy caused by external factors such as delays in steering system execution. The proposed solution is a hierarchical control method, grounded in mass estimation and model predictive control(MPC). Initially, to counter the variation in model parameters, a mass estimator is developed. This estimator utilizes the recursive least squares method with a forgetting factor, coupled with M-estimation, thereby enhancing the robustness of the estimation and achieving model correction. Subsequently, an upper-level MPC controller is constructed based on the error model, thereby augmenting the precision of tracking control. To address the delay in the steering system execution common in autonomous commercial vehicles, a lower-level steering angle compensator is designed to expedite the response speed of the execution. The feasibility of the vehicle’s front wheel angle is constrained via the rollover index, thereby enhancing vehicle stability during operation. The efficacy of the proposed control strategy is demonstrated with joint simulations using TruckSim/Simulink and real vehicle tests. The results indicate that this strategy can effectively manage the model mismatch caused by load changes in commercial vehicles and the delay in steering system execution, thereby exhibiting commendable tracking accuracy, adaptability, and driving stability. Full article

According to the physical characteristics of cotton and the work characteristics of cotton pickers in the field, during the picking process, there is a risk of cotton combustion. The cotton picker working environment is complex, cotton ignition can be hidden, and fire is difficult to detect. Therefore, in this study, we designed an improved algorithm for multi-sensor data fusion; built a cotton picker fire detection system by using infrared temperature sensors, CO sensors, and the upper computer; and proposed a BP neural network model based on improved mutation operator hybrid gray wolf optimizer and particle swarm optimization (MGWO-PSO) algorithm based on the BP neural network model. This algorithm includes the introduction of a mutation operator in the gray wolf algorithm to improve the search ability of the algorithm, and, at the same time, we introduce the PSO algorithm idea. The improved fusion algorithm is used as a learning algorithm to optimize the BP neural network, and the optimized network is used to process and predict the data collected from temperature and gas sensors, which effectively improves the accuracy of fire prediction. The sensor measurements were compared with the actual values to verify the effectiveness of the GWO-PSO-optimized BP neural network model. Once experimentally verified, the improved GWO-PSO algorithm achieves a correlation coefficient R of 0.96929, a prediction accuracy rate of 96.10%, and a prediction error rate of only 3.9%, while the system monitors an accurate early warning rate of 96.07%, and the false alarm and omission rates are both less than 5%. This study can detect cotton picker fires in real time and provide timely warnings, which provides a new method for the accurate detection of fires during the field operation of cotton pickers. Full article

Thermal infrared (TIR) images capture temperature in a non-invasive manner, making them valuable for generating 3D models that reflect the spatial distribution of thermal properties within a scene. Current TIR image-based 3D reconstruction methods primarily focus on static conditions, which only capture the spatial distribution of thermal radiation but lack the ability to represent its temporal dynamics. The absence of dedicated datasets and effective methods for dynamic 3D representation are two key challenges that hinder progress in this field. To address these challenges, we propose a novel dynamic thermal 3D reconstruction method, named ThermalGS, based on 3D Gaussian Splatting (3DGS). ThermalGS employs a data-driven approach to directly learn both scene structure and dynamic thermal representation, using RGB and TIR images as input. The position, orientation, and scale of Gaussian primitives are guided by the RGB mesh. We introduce feature encoding and embedding networks to integrate semantic and temporal information into the Gaussian primitives, allowing them to capture dynamic thermal radiation characteristics. Moreover, we construct the Thermal Scene Day-and-Night (TSDN) dataset, which includes multi-view, high-resolution aerial RGB reference images and TIR images captured at five different times throughout the day and night, providing a benchmark for dynamic thermal 3D reconstruction tasks. Experimental results demonstrate that the proposed method achieves state-of-the-art performance on the TSDN dataset, with an average absolute temperature error of 1 ${}^{\circ }$C and the ability to predict surface temperature variations over time. Full article

The temporal dominance of sensations (TDS) method captures assessors’ real-time sensory experiences during food tasting, while the temporal liking (TL) method evaluates dynamic changes in food preferences or perceived deliciousness. These sensory evaluation tools are essential for understanding consumer preferences but are also resource-intensive processes in the food development cycle. In this study, we used reservoir computing, a machine learning technique well-suited for time-series data, to predict temporal changes in liking based on the temporal evolution of dominant sensations. While previous studies developed reservoir models for specific food brands, achieving cross-brand prediction—predicting the temporal liking of one brand using a model trained on other brands—is a critical step toward replacing human assessors. We applied this approach to coffee products, predicting temporal liking for a given brand from its TDS data using a model trained on three other brands. The average prediction error across all brands was approximately 10% of the maximum instantaneous liking scores, and the mean correlation coefficients between the observed and predicted temporal scores ranged from 0.79 to 0.85 across the four brands, demonstrating the model’s potential for cross-brand prediction. This approach offers a promising technique for reducing the costs of sensory evaluation and enhancing product development in the food industry. Full article

Predicting the Remaining Useful Life (RUL) is vital for ensuring the reliability and safety of equipment and components. This study introduces a novel method for predicting RUL that utilizes the Convolutional Block Attention Module (CBAM) to address the problem that Convolutional Neural Networks (CNNs) do not effectively leverage data channel features and spatial features in residual life prediction. Firstly, Fast Fourier Transform (FFT) is applied to convert the data into the frequency domain. The resulting frequency domain data is then used as input to the convolutional neural network for feature extraction; Then, the weights of channel features and spatial features are assigned to the extracted features by CBAM, and the weighted features are then input into the Long Short-Term Memory (LSTM) network to learn temporal features. Finally, the effectiveness of the proposed model is verified using the PHM2012 bearing dataset. Compared to several existing RUL prediction methods, the mean squared error, mean absolute error, and root mean squared error of the proposed method in this paper are reduced by 53%, 16.87%, and 31.68%, respectively, which verifies the superiority of the method. Meanwhile, the experimental results demonstrate that the proposed method achieves good RUL prediction accuracy across various failure modes. Full article

The biological methanation process has emerged as a promising alternative to thermo-catalytic methods due to its ability to operate under milder conditions. However, challenges such as low hydrogen solubility and the need for precise trace element supplementation (Fe(II), Ni(II), Co(II)) constrain methane production yield. This study investigates the combined effects of trace element concentrations and applied pressure on biological methanation, addressing their synergistic interactions. Using a face-centered composite design, batch mode experiments were conducted to optimize methane production. Response Surface Methodology (RSM) and Artificial Neural Network (ANN)—Genetic Algorithm (GA) approaches were employed to model and optimize the process. RSM identified optimal ranges for trace elements and pressure, while ANN-GA demonstrated superior predictive accuracy, capturing nonlinear relationships with a high R² (>0.99) and minimal prediction errors. ANN-GA optimization indicated 97.9% methane production efficiency with a reduced conversion time of 15.9 h under conditions of 1.5 bar pressure and trace metal concentrations of 25.0 mg/L Fe(II), 0.20 mg/L Ni(II), and 0.02 mg/L Co(II). Validation experiments confirmed these predictions with deviations below 5%, underscoring the robustness of the models. The results highlight the synergistic effects of pressure and trace metals in enhancing gas–liquid mass transfer and enzymatic pathways, demonstrating the potential of computational modeling and experimental validation to optimize biological methanation systems, contributing to sustainable methane production. Full article

This paper proposes a transformer temperature early warning method based on an adaptive sliding window and stacking ensemble learning algorithm, aiming to improve the accuracy and robustness of temperature prediction. The transformer temperature early warning system is crucial for ensuring the safe operation of the power system, and temperature prediction, as the foundation of early warning, directly affects the early warning effectiveness. This paper analyzes the characteristics of transformer temperature using support vector regression, random forest, and gradient boosting regression as base learners and ridge regression as the meta-learner to construct a stacking model. At the same time, Bayesian optimization is used to automatically adjust the sliding window size, achieving adaptive sliding window processing. The experimental results indicate that the temperature prediction method based on adaptive sliding window and stacking significantly reduces prediction errors, enhances the model’s adaptability and generalization ability, and provides more reliable technical support for transformer fault warning. Full article

The accurate prediction of marine shaft centerline trajectories is essential for ensuring the operational performance and safety of ships. In this study, we propose a novel Transformer-based model to forecast the lateral and longitudinal displacements of ship main shafts. A key challenge in this prediction task is capturing both short-term fluctuations and long-term dependencies in shaft displacement data, which traditional models struggle to address. Our Transformer-based model integrates Bidirectional Splitting–Agg Attention and Sequence Progressive Split–Aggregation mechanisms to efficiently process bidirectional temporal dependencies, decompose seasonal and trend components, and handle the inherent symmetry of the shafting system. The symmetrical nature of the shafting system, with left and right shafts experiencing similar dynamic conditions, aligns with the bidirectional attention mechanism, enabling the model to better capture the symmetric relationships in displacement data. Experimental results demonstrate that the proposed model significantly outperforms traditional methods, such as Autoformer and Informer, in terms of prediction accuracy. Specifically, for 96 steps ahead, the mean absolute error (MAE) of our model is 0.232, compared to 0.235 for Autoformer and 0.264 for Informer, while the mean squared error (MSE) of our model is 0.209, compared to 0.242 for Autoformer and 0.286 for Informer. These results underscore the effectiveness of Transformer-based models in accurately predicting long-term marine shaft centerline trajectories, leveraging both temporal dependencies and structural symmetry, thus contributing to maritime monitoring and performance optimization. Full article

Marine regions are undergoing rapid evolution, primarily driven by natural and anthropogenic activities. Safeguarding these ecosystems necessitates the ability to observe their physical features and control processes with precision in both space and time. This demands the acquisition of precise and up-to-date information regarding several marine parameters. Thus, to gain a comprehensive understanding of these ecosystems, this study employs remote sensing techniques, Machine Learning algorithms and traditional in situ approaches. Together, these serve as valuable tools to help comprehend the distinctive parametric characteristics and mechanisms occurring within these regions of the Maltese archipelago. An empirical workflow was implemented to predict the spatial and temporal variations in sea surface salinity and sea surface temperature from 2022 to 2024. This was achieved by leveraging Sentinel-2 satellite platforms, the random forest Machine Learning algorithm, and in situ data collected from sea gliders and floats. Subsequently, the numerical data generated by the random forest algorithm were validated with different error metrics and converted into visual representations to illustrate the sea surface salinity and sea surface temperature variations across the Maltese Islands. The random forest algorithm demonstrated strong performance in predicting sea surface salinity and sea surface temperature, indicating its capability to handle dynamic parameters effectively. Additionally, the parametric maps generated for all three years provided a clear understanding of both the spatial and temporal changes for these two parameters. Full article

In this work, we propose a Navier–Stokes-Informed Neural Network (NSINN) as a surrogate approach to predict the localized flow behavior of cementitious materials for advancing 3D additive construction technology to gain fundamental insights into multiscale mechanisms of cement paste rheology. NS equations are embedded into the NSINN to interpret the flow pattern in the 3D printing barrel. The results show that the presented NSINN has a higher accuracy compared to a traditional artificial neural network (ANN) as the Mean Square Errors (MSEs) of the u, v, and p predicted by NSINN are $1.25\times {10}^{-4}$, $1.85\times {10}^{-5}$, and $3.91\times {10}^{-3}$, respectively. Compared to the ANN, the MSE of the predictions are $5.88\times {10}^{-2}$, $4.17\times {10}^{-3}$, and $1.72\times {10}^{-2}$, respectively. Moreover, the mean prediction time used in the NSINN, the ANN, and Computational Fluid Dynamics (CFD) are 0.039 s, 0.014 s, and 3.37 s, respectively. That means the method is more computationally efficient at performing simulations compared to CFD which is mesh-based. The NSINN is also utilized in studying the relationship between geometry and extrudability. The ratio (R = 0.25, 0.5, and 0.75) between the diameter of the outlet and that of the domain is studied. It shows that a larger ratio (R = 0.75) can lead to better extrudability of the 3D concrete printing (3DCP). Full article

This study introduces a machine learning (ML)-driven approach to next-generation microwave filter design that enhances both accuracy and efficiency via repeated refinement. The approach includes generating a coupling matrix from filter specifications, followed by predicting physical parameters such as iris widths and resonator lengths using ML models, especially with the XGBoost algorithm. These predictions are validated and tuned via simulations and iterative adjustments to ensure meeting the performance criteria, such as center frequency, bandwidth, and return loss. For tuning, in this work, we used Simulated Annealing to extract a coupling matrix to reduce errors and hence allow accurate further optimization. The predicted values before optimization are more than 90 percent accurate compared to the optimized values, significantly reducing the optimization time and the number of iterations required. To demonstrate the procedure’s validity, third-, fourth-, and fifth-order filters are implemented, which shows significant improvements in design efficiency and accuracy. Full article

A short-term photovoltaic power forecasting method is proposed, integrating variational mode decomposition (VMD), an improved dung beetle algorithm (IDBO), and a deep hybrid kernel extreme learning machine (DHKELM). First, the weather factors less relevant to photovoltaic (PV) power generation are filtered using the Spearman correlation coefficient. Historical data are then clustered into three categories—sunny, cloudy, and rainy days—using the K-means algorithm. Next, the original PV power data are decomposed through VMD. A DHKELM-based combined prediction model is developed for each component of the decomposition, tailored to different weather types. The model’s hyperparameters are optimized using the IDBO. The final power forecast is determined by combining the outcomes of each individual component. Validation is performed using actual data from a PV power plant in Australia and a PV power station in Kashgar, China demonstrates. Numerical evaluation results show that the proposed method improves the Mean Absolute Error (MAE) by 3.84% and the Root-Mean-Squared Error (RMSE) by 3.38%, confirming its accuracy. Full article

The objective of this work is to develop a method for tracking human skeletal movements by integrating data from two synchronized video streams. To achieve this, two datasets were created, each consisting of four different rehabilitation exercise videos featuring various individuals in diverse environments and wearing different clothing. The prediction model is employed to create a dual-image stream system that enables the tracking of joint positions even when a joint is obscured in one of the streams. This system also mitigates depth coordinate errors by using data from both video streams. The final implementation successfully corrects the positions of the right elbow and wrist joints, though some depth error persists in the left hand. The results demonstrate that adding a second video camera, rotated 90° and aimed at the subject, can compensate for depth prediction inaccuracies, reducing errors by up to 0.4 m. By using a dual-camera setup and fusing the predicted human skeletal models, it is possible to construct a complete human model even when one camera does not capture all body parts and to refine depth coordinates through error correction using a linear regression model. Full article

Accurate localization is crucial for numerous applications. While several methods exist for outdoor localization, typically relying on GPS signals, these approaches become unreliable in environments subject to a weak GPS signal or GPS outage. Many researchers have attempted to address this limitation, primarily focusing on real-time solutions. However, for applications that do not require real-time localization, these methods remain suboptimal. This paper presents a novel Transformer-based bidirectional encoder approach to address, in postprocessing, the localization challenges during GPS weak signal phases or GPS outages. Our method predicts the velocity during periods of weak or lost GPS signals and calculates the position through bidirectional velocity integration. Additionally, it incorporates position interpolation to ensure smooth transitions between active GPS and GPS outage phases. Applied to a dataset tracking horse positions—which features velocities up to 10 times those of pedestrians and higher acceleration—our approach achieved an average trajectory error below 3 m, while maintaining stable relative distance errors regardless of the GPS outage duration. Full article